diff --git a/.buildinfo b/.buildinfo index d8fc416f6..bfb060702 100644 --- a/.buildinfo +++ b/.buildinfo @@ -1,4 +1,4 @@ # Sphinx build info version 1 # This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done. -config: 2a4bd9441600d6f87ec45cdb9c8c85e0 +config: 67a75a999b8281c3988fcac7056d30e5 tags: 645f666f9bcd5a90fca523b33c5a78b7 diff --git a/.doctrees/environment.pickle b/.doctrees/environment.pickle index e9d426296..eada649c8 100644 Binary files a/.doctrees/environment.pickle and b/.doctrees/environment.pickle differ diff --git a/.doctrees/python/ctranslate2.GenerationResult.doctree b/.doctrees/python/ctranslate2.GenerationResult.doctree index 42c0b3e6b..55c31bdc8 100644 Binary files a/.doctrees/python/ctranslate2.GenerationResult.doctree and b/.doctrees/python/ctranslate2.GenerationResult.doctree differ diff --git a/.doctrees/python/ctranslate2.TranslationResult.doctree b/.doctrees/python/ctranslate2.TranslationResult.doctree index 2b1f49038..b72689762 100644 Binary files a/.doctrees/python/ctranslate2.TranslationResult.doctree and b/.doctrees/python/ctranslate2.TranslationResult.doctree differ diff --git a/.doctrees/python/ctranslate2.models.Wav2Vec2Bert.doctree b/.doctrees/python/ctranslate2.models.Wav2Vec2Bert.doctree new file mode 100644 index 000000000..bbeb2069e Binary files /dev/null and b/.doctrees/python/ctranslate2.models.Wav2Vec2Bert.doctree differ diff --git a/.doctrees/python/ctranslate2.models.WhisperGenerationResult.doctree b/.doctrees/python/ctranslate2.models.WhisperGenerationResult.doctree index 06c0b62a1..3f1645e5a 100644 Binary files a/.doctrees/python/ctranslate2.models.WhisperGenerationResult.doctree and b/.doctrees/python/ctranslate2.models.WhisperGenerationResult.doctree differ diff --git a/.doctrees/python/ctranslate2.models.doctree b/.doctrees/python/ctranslate2.models.doctree index 1fda6cc7f..41f623e39 100644 Binary files a/.doctrees/python/ctranslate2.models.doctree and b/.doctrees/python/ctranslate2.models.doctree differ diff --git a/.doctrees/python/ctranslate2.specs.Activation.doctree b/.doctrees/python/ctranslate2.specs.Activation.doctree index 8f18fcb88..684e5767f 100644 Binary files a/.doctrees/python/ctranslate2.specs.Activation.doctree and b/.doctrees/python/ctranslate2.specs.Activation.doctree differ diff --git a/.doctrees/python/ctranslate2.specs.LanguageModelSpec.doctree b/.doctrees/python/ctranslate2.specs.LanguageModelSpec.doctree index 557b29e88..5db472253 100644 Binary files a/.doctrees/python/ctranslate2.specs.LanguageModelSpec.doctree and b/.doctrees/python/ctranslate2.specs.LanguageModelSpec.doctree differ diff --git a/.doctrees/python/ctranslate2.specs.Wav2Vec2BertSpec.doctree b/.doctrees/python/ctranslate2.specs.Wav2Vec2BertSpec.doctree new file mode 100644 index 000000000..2fcfa8e2f Binary files /dev/null and b/.doctrees/python/ctranslate2.specs.Wav2Vec2BertSpec.doctree differ diff --git a/.doctrees/python/ctranslate2.specs.doctree b/.doctrees/python/ctranslate2.specs.doctree index 243eb5426..eba78acee 100644 Binary files a/.doctrees/python/ctranslate2.specs.doctree and b/.doctrees/python/ctranslate2.specs.doctree differ diff --git a/.doctrees/quantization.doctree b/.doctrees/quantization.doctree index 9ba8ce31b..606e223d6 100644 Binary files a/.doctrees/quantization.doctree and b/.doctrees/quantization.doctree differ diff --git a/_sources/python/ctranslate2.models.Wav2Vec2Bert.rst.txt b/_sources/python/ctranslate2.models.Wav2Vec2Bert.rst.txt new file mode 100644 index 000000000..c339f4389 --- /dev/null +++ b/_sources/python/ctranslate2.models.Wav2Vec2Bert.rst.txt @@ -0,0 +1,26 @@ +Wav2Vec2Bert +============ + +.. autoclass:: ctranslate2.models.Wav2Vec2Bert + :members: + :undoc-members: + :inherited-members: + + **Inherits from:** :class:`pybind11_builtins.pybind11_object` + + **Attributes:** + + - :obj:`~ctranslate2.models.Wav2Vec2Bert.compute_type` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.device` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.device_index` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.model_is_loaded` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.num_active_batches` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.num_queued_batches` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.num_workers` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.tensor_parallel` + + **Methods:** + + - :obj:`~ctranslate2.models.Wav2Vec2Bert.encode` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.load_model` + - :obj:`~ctranslate2.models.Wav2Vec2Bert.unload_model` diff --git a/_sources/python/ctranslate2.models.WhisperGenerationResult.rst.txt b/_sources/python/ctranslate2.models.WhisperGenerationResult.rst.txt index 000d7c44e..2b18aa21b 100644 --- a/_sources/python/ctranslate2.models.WhisperGenerationResult.rst.txt +++ b/_sources/python/ctranslate2.models.WhisperGenerationResult.rst.txt @@ -10,6 +10,7 @@ WhisperGenerationResult **Attributes:** + - :obj:`~ctranslate2.models.WhisperGenerationResult.logits` - :obj:`~ctranslate2.models.WhisperGenerationResult.no_speech_prob` - :obj:`~ctranslate2.models.WhisperGenerationResult.scores` - :obj:`~ctranslate2.models.WhisperGenerationResult.sequences` diff --git a/_sources/python/ctranslate2.models.rst.txt b/_sources/python/ctranslate2.models.rst.txt index 8a9eb0b96..d876402e3 100644 --- a/_sources/python/ctranslate2.models.rst.txt +++ b/_sources/python/ctranslate2.models.rst.txt @@ -6,6 +6,7 @@ ctranslate2.models .. toctree:: ctranslate2.models.Wav2Vec2 + ctranslate2.models.Wav2Vec2Bert ctranslate2.models.Whisper ctranslate2.models.WhisperGenerationResult ctranslate2.models.WhisperGenerationResultAsync diff --git a/_sources/python/ctranslate2.specs.LanguageModelSpec.rst.txt b/_sources/python/ctranslate2.specs.LanguageModelSpec.rst.txt index d997060b9..d33c4c9cd 100644 --- a/_sources/python/ctranslate2.specs.LanguageModelSpec.rst.txt +++ b/_sources/python/ctranslate2.specs.LanguageModelSpec.rst.txt @@ -12,6 +12,7 @@ LanguageModelSpec - :class:`ctranslate2.specs.TransformerDecoderModelSpec` - :class:`ctranslate2.specs.TransformerEncoderModelSpec` + - :class:`ctranslate2.specs.Wav2Vec2BertSpec` - :class:`ctranslate2.specs.Wav2Vec2Spec` - :class:`ctranslate2.specs.WhisperSpec` diff --git a/_sources/python/ctranslate2.specs.Wav2Vec2BertSpec.rst.txt b/_sources/python/ctranslate2.specs.Wav2Vec2BertSpec.rst.txt new file mode 100644 index 000000000..3688b9c00 --- /dev/null +++ b/_sources/python/ctranslate2.specs.Wav2Vec2BertSpec.rst.txt @@ -0,0 +1,26 @@ +Wav2Vec2BertSpec +================ + +.. autoclass:: ctranslate2.specs.Wav2Vec2BertSpec + :members: + :undoc-members: + :inherited-members: + + **Inherits from:** :class:`ctranslate2.specs.LanguageModelSpec` + + **Attributes:** + + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.config` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.name` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.revision` + + **Methods:** + + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.get_default_config` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.get_vocabulary_size` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.optimize` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.register_file` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.register_vocabulary` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.save` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.validate` + - :obj:`~ctranslate2.specs.Wav2Vec2BertSpec.variables` diff --git a/_sources/python/ctranslate2.specs.rst.txt b/_sources/python/ctranslate2.specs.rst.txt index a2d177d03..4015d9b45 100644 --- a/_sources/python/ctranslate2.specs.rst.txt +++ b/_sources/python/ctranslate2.specs.rst.txt @@ -17,5 +17,6 @@ ctranslate2.specs ctranslate2.specs.TransformerEncoderModelSpec ctranslate2.specs.TransformerEncoderSpec ctranslate2.specs.TransformerSpec + ctranslate2.specs.Wav2Vec2BertSpec ctranslate2.specs.Wav2Vec2Spec ctranslate2.specs.WhisperSpec diff --git a/_sources/quantization.md.txt b/_sources/quantization.md.txt index 296c57000..ae79ee6b9 100644 --- a/_sources/quantization.md.txt +++ b/_sources/quantization.md.txt @@ -165,18 +165,26 @@ In this mode, all model weights are stored in BF16 and all layers are run with t ### 4-bit AWQ -The compute type would be `int32_float16` - **Supported on:** * NVIDIA GPU with Compute Capability >= 7.5 +CTranslate2 internally handles the compute type for AWQ quantization. In this mode, all model weights are stored in half precision and all layers are run in half precision. Other parameters like scale and zero are stored in ``int32``. -For example, +**Steps to use AWQ Quantization:** + +* Download a AWQ quantized model from Hugging Face for example (TheBloke/Llama-2-7B-AWQ){https://huggingface.co/TheBloke/Llama-2-7B-AWQ} or quantize your own model with using this (AutoAWQ example){https://casper-hansen.github.io/AutoAWQ/examples/}. +* Convert AWQ Quantized model to Ctranslate2 model: ```bash ct2-transformers-converter --model TheBloke/Llama-2-7B-AWQ --copy_files tokenizer.model --output_dir ct2_model ``` -We have to quantize the model with AWQ first, then convert it to CT2 format. \ No newline at end of file +* Run inference as usual with Ctranslate2: +```bash +model = ctranslate2.Generator('ct2_model', device='cuda') +outputs = model.generate_batch([tokens]) +``` + +Currently, CTranslate2 only supports the GEMM and GEMV kernels for AWQ quantization. \ No newline at end of file diff --git a/_static/documentation_options.js b/_static/documentation_options.js index b6ce93da8..2a1a60ed2 100644 --- a/_static/documentation_options.js +++ b/_static/documentation_options.js @@ -1,6 +1,6 @@ var DOCUMENTATION_OPTIONS = { URL_ROOT: document.getElementById("documentation_options").getAttribute('data-url_root'), - VERSION: '4.4.0', + VERSION: '4.5.0', LANGUAGE: 'en', COLLAPSE_INDEX: false, BUILDER: 'html', diff --git a/conversion.html b/conversion.html index 9fb4df1bc..61b9e1092 100644 --- a/conversion.html +++ b/conversion.html @@ -4,7 +4,7 @@ - Model conversion — CTranslate2 4.4.0 documentation + Model conversion — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/decoding.html b/decoding.html index 7f74ffa10..5519487e7 100644 --- a/decoding.html +++ b/decoding.html @@ -4,7 +4,7 @@ - Decoding features — CTranslate2 4.4.0 documentation + Decoding features — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/encoding.html b/encoding.html index 588ce4eb2..6a9b4806f 100644 --- a/encoding.html +++ b/encoding.html @@ -4,7 +4,7 @@ - Text encoding — CTranslate2 4.4.0 documentation + Text encoding — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/environment_variables.html b/environment_variables.html index 899e15e6b..e2b0ce23e 100644 --- a/environment_variables.html +++ b/environment_variables.html @@ -4,7 +4,7 @@ - Environment variables — CTranslate2 4.4.0 documentation + Environment variables — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/faq.html b/faq.html index d21090641..45e94367f 100644 --- a/faq.html +++ b/faq.html @@ -4,7 +4,7 @@ - Frequently asked questions — CTranslate2 4.4.0 documentation + Frequently asked questions — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/generation.html b/generation.html index caa3b8d7b..2f14838d1 100644 --- a/generation.html +++ b/generation.html @@ -4,7 +4,7 @@ - Text generation — CTranslate2 4.4.0 documentation + Text generation — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
- 4.4 + 4.5
diff --git a/genindex.html b/genindex.html index a2e37d7e3..9f8f54983 100644 --- a/genindex.html +++ b/genindex.html @@ -3,7 +3,7 @@ - Index — CTranslate2 4.4.0 documentation + Index — CTranslate2 4.5.0 documentation @@ -29,7 +29,7 @@ CTranslate2
- 4.4 + 4.5
@@ -158,6 +158,8 @@

_

  • (ctranslate2.Generator method)
  • (ctranslate2.models.Wav2Vec2 method) +
  • +
  • (ctranslate2.models.Wav2Vec2Bert method)
  • (ctranslate2.models.Whisper method)
  • @@ -176,6 +178,8 @@

    _

  • (ctranslate2.specs.TransformerEncoderSpec method)
  • (ctranslate2.specs.TransformerSpec method) +
  • +
  • (ctranslate2.specs.Wav2Vec2BertSpec method)
  • (ctranslate2.specs.Wav2Vec2Spec method)
  • @@ -232,6 +236,8 @@

    C

  • (ctranslate2.Generator property)
  • (ctranslate2.models.Wav2Vec2 property) +
  • +
  • (ctranslate2.models.Wav2Vec2Bert property)
  • (ctranslate2.models.Whisper property)
  • @@ -254,6 +260,8 @@

    C

  • (ctranslate2.specs.TransformerEncoderModelSpec property)
  • (ctranslate2.specs.TransformerSpec property) +
  • +
  • (ctranslate2.specs.Wav2Vec2BertSpec property)
  • (ctranslate2.specs.Wav2Vec2Spec property)
  • @@ -370,6 +378,8 @@

    D

  • (ctranslate2.Generator property)
  • (ctranslate2.models.Wav2Vec2 property) +
  • +
  • (ctranslate2.models.Wav2Vec2Bert property)
  • (ctranslate2.models.Whisper property)
  • @@ -386,6 +396,8 @@

    D

  • (ctranslate2.Generator property)
  • (ctranslate2.models.Wav2Vec2 property) +
  • +
  • (ctranslate2.models.Wav2Vec2Bert property)
  • (ctranslate2.models.Whisper property)
  • @@ -417,6 +429,8 @@

    E

  • encode() (ctranslate2.models.Wav2Vec2 method)
  • @@ -502,6 +516,8 @@

    G

  • (ctranslate2.specs.TransformerEncoderModelSpec method)
  • (ctranslate2.specs.TransformerSpec method) +
  • +
  • (ctranslate2.specs.Wav2Vec2BertSpec method)
  • (ctranslate2.specs.Wav2Vec2Spec method)
  • @@ -534,6 +550,8 @@

    G

  • (ctranslate2.specs.TransformerDecoderModelSpec method)
  • (ctranslate2.specs.TransformerEncoderModelSpec method) +
  • +
  • (ctranslate2.specs.Wav2Vec2BertSpec method)
  • (ctranslate2.specs.Wav2Vec2Spec method)
  • @@ -600,6 +618,8 @@

    L

  • (ctranslate2.Generator method)
  • (ctranslate2.models.Wav2Vec2 method) +
  • +
  • (ctranslate2.models.Wav2Vec2Bert method)
  • (ctranslate2.models.Whisper method)
  • @@ -618,6 +638,8 @@

    L

    • (ctranslate2.GenerationStepResult property) +
    • +
    • (ctranslate2.models.WhisperGenerationResult property)
    • (ctranslate2.TranslationResult property)
    • @@ -636,6 +658,8 @@

      M

    • (ctranslate2.Generator property)
    • (ctranslate2.models.Wav2Vec2 property) +
    • +
    • (ctranslate2.models.Wav2Vec2Bert property)
    • (ctranslate2.models.Whisper property)
    • @@ -685,6 +709,8 @@

      N

    • (ctranslate2.specs.TransformerEncoderModelSpec property)
    • (ctranslate2.specs.TransformerSpec property) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec property)
    • (ctranslate2.specs.Wav2Vec2Spec property)
    • @@ -699,6 +725,8 @@

      N

    • (ctranslate2.Generator property)
    • (ctranslate2.models.Wav2Vec2 property) +
    • +
    • (ctranslate2.models.Wav2Vec2Bert property)
    • (ctranslate2.models.Whisper property)
    • @@ -721,6 +749,8 @@

      N

    • (ctranslate2.Generator property)
    • (ctranslate2.models.Wav2Vec2 property) +
    • +
    • (ctranslate2.models.Wav2Vec2Bert property)
    • (ctranslate2.models.Whisper property)
    • @@ -734,6 +764,8 @@

      N

    • num_workers (ctranslate2.models.Wav2Vec2 property)
    • @@ -767,6 +799,8 @@

      O

    • (ctranslate2.specs.TransformerEncoderSpec method)
    • (ctranslate2.specs.TransformerSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -803,6 +837,8 @@

      R

    • (ctranslate2.specs.TransformerEncoderModelSpec method)
    • (ctranslate2.specs.TransformerSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -827,6 +863,8 @@

      R

    • (ctranslate2.specs.TransformerDecoderModelSpec method)
    • (ctranslate2.specs.TransformerEncoderModelSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -865,6 +903,8 @@

      R

    • (ctranslate2.specs.TransformerEncoderModelSpec property)
    • (ctranslate2.specs.TransformerSpec property) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec property)
    • (ctranslate2.specs.Wav2Vec2Spec property)
    • @@ -891,6 +931,8 @@

      S

    • (ctranslate2.specs.TransformerEncoderModelSpec method)
    • (ctranslate2.specs.TransformerSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -942,6 +984,8 @@

      S

    • set_random_seed() (in module ctranslate2)
    • shape (ctranslate2.StorageView property) +
    • +
    • Sigmoid (ctranslate2.specs.Activation attribute)
    • step (ctranslate2.GenerationStepResult property)
    • @@ -965,6 +1009,8 @@

      T

    • (ctranslate2.Generator property)
    • (ctranslate2.models.Wav2Vec2 property) +
    • +
    • (ctranslate2.models.Wav2Vec2Bert property)
    • (ctranslate2.models.Whisper property)
    • @@ -1019,6 +1065,8 @@

      U

    • (ctranslate2.Generator method)
    • (ctranslate2.models.Wav2Vec2 method) +
    • +
    • (ctranslate2.models.Wav2Vec2Bert method)
    • (ctranslate2.models.Whisper method)
    • @@ -1049,6 +1097,8 @@

      V

    • (ctranslate2.specs.TransformerEncoderSpec method)
    • (ctranslate2.specs.TransformerSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -1081,6 +1131,8 @@

      V

    • (ctranslate2.specs.TransformerEncoderSpec method)
    • (ctranslate2.specs.TransformerSpec method) +
    • +
    • (ctranslate2.specs.Wav2Vec2BertSpec method)
    • (ctranslate2.specs.Wav2Vec2Spec method)
    • @@ -1095,12 +1147,16 @@

      W

        +
      • Whisper (class in ctranslate2.models) +
      • WhisperGenerationResult (class in ctranslate2.models)
      • WhisperGenerationResultAsync (class in ctranslate2.models) diff --git a/guides/fairseq.html b/guides/fairseq.html index d2711880a..1173a0611 100644 --- a/guides/fairseq.html +++ b/guides/fairseq.html @@ -4,7 +4,7 @@ - Fairseq — CTranslate2 4.4.0 documentation + Fairseq — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/guides/marian.html b/guides/marian.html index 7cc913be8..732037f5b 100644 --- a/guides/marian.html +++ b/guides/marian.html @@ -4,7 +4,7 @@ - Marian — CTranslate2 4.4.0 documentation + Marian — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/guides/opennmt_py.html b/guides/opennmt_py.html index 04c01e207..ab6c9a87a 100644 --- a/guides/opennmt_py.html +++ b/guides/opennmt_py.html @@ -4,7 +4,7 @@ - OpenNMT-py — CTranslate2 4.4.0 documentation + OpenNMT-py — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/guides/opennmt_tf.html b/guides/opennmt_tf.html index 315d9ec2b..a6eace5c7 100644 --- a/guides/opennmt_tf.html +++ b/guides/opennmt_tf.html @@ -4,7 +4,7 @@ - OpenNMT-tf — CTranslate2 4.4.0 documentation + OpenNMT-tf — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/guides/opus_mt.html b/guides/opus_mt.html index 2ab2dd520..f58be860e 100644 --- a/guides/opus_mt.html +++ b/guides/opus_mt.html @@ -4,7 +4,7 @@ - OPUS-MT — CTranslate2 4.4.0 documentation + OPUS-MT — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/guides/transformers.html b/guides/transformers.html index 78410acfe..850966e42 100644 --- a/guides/transformers.html +++ b/guides/transformers.html @@ -4,7 +4,7 @@ - Transformers — CTranslate2 4.4.0 documentation + Transformers — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/hardware_support.html b/hardware_support.html index 8a3d432a0..6741300f7 100644 --- a/hardware_support.html +++ b/hardware_support.html @@ -4,7 +4,7 @@ - Hardware support — CTranslate2 4.4.0 documentation + Hardware support — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/index.html b/index.html index 9b5dea223..432f688b7 100644 --- a/index.html +++ b/index.html @@ -4,7 +4,7 @@ - Index — CTranslate2 4.4.0 documentation + Index — CTranslate2 4.5.0 documentation @@ -31,7 +31,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/installation.html b/installation.html index 50fa5f76e..e11ef7355 100644 --- a/installation.html +++ b/installation.html @@ -4,7 +4,7 @@ - Installation — CTranslate2 4.4.0 documentation + Installation — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/memory.html b/memory.html index 89711b2f4..f81a68e17 100644 --- a/memory.html +++ b/memory.html @@ -4,7 +4,7 @@ - Memory management — CTranslate2 4.4.0 documentation + Memory management — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/objects.inv b/objects.inv index 9d6f2db3b..7815a1bc3 100644 Binary files a/objects.inv and b/objects.inv differ diff --git a/parallel.html b/parallel.html index 780a6fb65..d181d728a 100644 --- a/parallel.html +++ b/parallel.html @@ -4,7 +4,7 @@ - Multithreading and parallelism — CTranslate2 4.4.0 documentation + Multithreading and parallelism — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/performance.html b/performance.html index 58fa95214..3817256d0 100644 --- a/performance.html +++ b/performance.html @@ -4,7 +4,7 @@ - Performance tips — CTranslate2 4.4.0 documentation + Performance tips — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/py-modindex.html b/py-modindex.html index 908d5d4be..892b463d6 100644 --- a/py-modindex.html +++ b/py-modindex.html @@ -3,7 +3,7 @@ - Python Module Index — CTranslate2 4.4.0 documentation + Python Module Index — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.AsyncGenerationResult.html b/python/ctranslate2.AsyncGenerationResult.html index 799276048..fdae21f38 100644 --- a/python/ctranslate2.AsyncGenerationResult.html +++ b/python/ctranslate2.AsyncGenerationResult.html @@ -4,7 +4,7 @@ - AsyncGenerationResult — CTranslate2 4.4.0 documentation + AsyncGenerationResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.AsyncScoringResult.html b/python/ctranslate2.AsyncScoringResult.html index 3975952c8..2398b4551 100644 --- a/python/ctranslate2.AsyncScoringResult.html +++ b/python/ctranslate2.AsyncScoringResult.html @@ -4,7 +4,7 @@ - AsyncScoringResult — CTranslate2 4.4.0 documentation + AsyncScoringResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.AsyncTranslationResult.html b/python/ctranslate2.AsyncTranslationResult.html index cef820a0d..559cfd949 100644 --- a/python/ctranslate2.AsyncTranslationResult.html +++ b/python/ctranslate2.AsyncTranslationResult.html @@ -4,7 +4,7 @@ - AsyncTranslationResult — CTranslate2 4.4.0 documentation + AsyncTranslationResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.DataType.html b/python/ctranslate2.DataType.html index 35ec70299..02610b333 100644 --- a/python/ctranslate2.DataType.html +++ b/python/ctranslate2.DataType.html @@ -4,7 +4,7 @@ - DataType — CTranslate2 4.4.0 documentation + DataType — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.Device.html b/python/ctranslate2.Device.html index 687f936e6..9d7cf6c91 100644 --- a/python/ctranslate2.Device.html +++ b/python/ctranslate2.Device.html @@ -4,7 +4,7 @@ - Device — CTranslate2 4.4.0 documentation + Device — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.Encoder.html b/python/ctranslate2.Encoder.html index 5573945b0..5be56a4ee 100644 --- a/python/ctranslate2.Encoder.html +++ b/python/ctranslate2.Encoder.html @@ -4,7 +4,7 @@ - Encoder — CTranslate2 4.4.0 documentation + Encoder — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.EncoderForwardOutput.html b/python/ctranslate2.EncoderForwardOutput.html index 78b071921..d7d6febef 100644 --- a/python/ctranslate2.EncoderForwardOutput.html +++ b/python/ctranslate2.EncoderForwardOutput.html @@ -4,7 +4,7 @@ - EncoderForwardOutput — CTranslate2 4.4.0 documentation + EncoderForwardOutput — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.ExecutionStats.html b/python/ctranslate2.ExecutionStats.html index f733e5fad..5f489ac10 100644 --- a/python/ctranslate2.ExecutionStats.html +++ b/python/ctranslate2.ExecutionStats.html @@ -4,7 +4,7 @@ - ExecutionStats — CTranslate2 4.4.0 documentation + ExecutionStats — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.GenerationResult.html b/python/ctranslate2.GenerationResult.html index 7e9d965eb..081de7e83 100644 --- a/python/ctranslate2.GenerationResult.html +++ b/python/ctranslate2.GenerationResult.html @@ -4,7 +4,7 @@ - GenerationResult — CTranslate2 4.4.0 documentation + GenerationResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        @@ -156,7 +156,7 @@

        GenerationResult
        property logits
        -

        Score of each sequence (empty if return_logits_vocab was disabled).

        +

        Logits of each sequence (empty if return_logits_vocab was disabled).

        diff --git a/python/ctranslate2.GenerationStepResult.html b/python/ctranslate2.GenerationStepResult.html index d0a2ba525..3001a9de2 100644 --- a/python/ctranslate2.GenerationStepResult.html +++ b/python/ctranslate2.GenerationStepResult.html @@ -4,7 +4,7 @@ - GenerationStepResult — CTranslate2 4.4.0 documentation + GenerationStepResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.Generator.html b/python/ctranslate2.Generator.html index df6580bad..5286e6779 100644 --- a/python/ctranslate2.Generator.html +++ b/python/ctranslate2.Generator.html @@ -4,7 +4,7 @@ - Generator — CTranslate2 4.4.0 documentation + Generator — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.MpiInfo.html b/python/ctranslate2.MpiInfo.html index 7bb29897b..44dc12f57 100644 --- a/python/ctranslate2.MpiInfo.html +++ b/python/ctranslate2.MpiInfo.html @@ -4,7 +4,7 @@ - MpiInfo — CTranslate2 4.4.0 documentation + MpiInfo — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.ScoringResult.html b/python/ctranslate2.ScoringResult.html index 8ed0dfb77..4ce1890c5 100644 --- a/python/ctranslate2.ScoringResult.html +++ b/python/ctranslate2.ScoringResult.html @@ -4,7 +4,7 @@ - ScoringResult — CTranslate2 4.4.0 documentation + ScoringResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.StorageView.html b/python/ctranslate2.StorageView.html index 2c9ef4942..7fc47c93f 100644 --- a/python/ctranslate2.StorageView.html +++ b/python/ctranslate2.StorageView.html @@ -4,7 +4,7 @@ - StorageView — CTranslate2 4.4.0 documentation + StorageView — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.TranslationResult.html b/python/ctranslate2.TranslationResult.html index 937d766f0..3b58fec0e 100644 --- a/python/ctranslate2.TranslationResult.html +++ b/python/ctranslate2.TranslationResult.html @@ -4,7 +4,7 @@ - TranslationResult — CTranslate2 4.4.0 documentation + TranslationResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        @@ -168,7 +168,7 @@

        TranslationResult
        property logits
        -

        Score of each translation hypothesis (empty if return_logits_vocab was disabled).

        +

        Logits of each translation hypothesis (empty if return_logits_vocab was disabled).

        diff --git a/python/ctranslate2.Translator.html b/python/ctranslate2.Translator.html index 5bba9fdf9..aa4de95a1 100644 --- a/python/ctranslate2.Translator.html +++ b/python/ctranslate2.Translator.html @@ -4,7 +4,7 @@ - Translator — CTranslate2 4.4.0 documentation + Translator — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.contains_model.html b/python/ctranslate2.contains_model.html index 0ae3d8d90..b51a9ca4d 100644 --- a/python/ctranslate2.contains_model.html +++ b/python/ctranslate2.contains_model.html @@ -4,7 +4,7 @@ - contains_model — CTranslate2 4.4.0 documentation + contains_model — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.Converter.html b/python/ctranslate2.converters.Converter.html index 61a0afe29..ae55a6305 100644 --- a/python/ctranslate2.converters.Converter.html +++ b/python/ctranslate2.converters.Converter.html @@ -4,7 +4,7 @@ - Converter — CTranslate2 4.4.0 documentation + Converter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.FairseqConverter.html b/python/ctranslate2.converters.FairseqConverter.html index ec9cdd43b..551cdbaea 100644 --- a/python/ctranslate2.converters.FairseqConverter.html +++ b/python/ctranslate2.converters.FairseqConverter.html @@ -4,7 +4,7 @@ - FairseqConverter — CTranslate2 4.4.0 documentation + FairseqConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.MarianConverter.html b/python/ctranslate2.converters.MarianConverter.html index 19e5b7dac..9199860b9 100644 --- a/python/ctranslate2.converters.MarianConverter.html +++ b/python/ctranslate2.converters.MarianConverter.html @@ -4,7 +4,7 @@ - MarianConverter — CTranslate2 4.4.0 documentation + MarianConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.OpenAIGPT2Converter.html b/python/ctranslate2.converters.OpenAIGPT2Converter.html index 357dd5ae5..68637e85a 100644 --- a/python/ctranslate2.converters.OpenAIGPT2Converter.html +++ b/python/ctranslate2.converters.OpenAIGPT2Converter.html @@ -4,7 +4,7 @@ - OpenAIGPT2Converter — CTranslate2 4.4.0 documentation + OpenAIGPT2Converter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.OpenNMTPyConverter.html b/python/ctranslate2.converters.OpenNMTPyConverter.html index 09ca3a2f4..dc15f3a78 100644 --- a/python/ctranslate2.converters.OpenNMTPyConverter.html +++ b/python/ctranslate2.converters.OpenNMTPyConverter.html @@ -4,7 +4,7 @@ - OpenNMTPyConverter — CTranslate2 4.4.0 documentation + OpenNMTPyConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.OpenNMTTFConverter.html b/python/ctranslate2.converters.OpenNMTTFConverter.html index b1884720e..f3dc1ae9f 100644 --- a/python/ctranslate2.converters.OpenNMTTFConverter.html +++ b/python/ctranslate2.converters.OpenNMTTFConverter.html @@ -4,7 +4,7 @@ - OpenNMTTFConverter — CTranslate2 4.4.0 documentation + OpenNMTTFConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.OpusMTConverter.html b/python/ctranslate2.converters.OpusMTConverter.html index 8002a6915..c0d3cbec3 100644 --- a/python/ctranslate2.converters.OpusMTConverter.html +++ b/python/ctranslate2.converters.OpusMTConverter.html @@ -4,7 +4,7 @@ - OpusMTConverter — CTranslate2 4.4.0 documentation + OpusMTConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.TransformersConverter.html b/python/ctranslate2.converters.TransformersConverter.html index 43becef5a..34e928163 100644 --- a/python/ctranslate2.converters.TransformersConverter.html +++ b/python/ctranslate2.converters.TransformersConverter.html @@ -4,7 +4,7 @@ - TransformersConverter — CTranslate2 4.4.0 documentation + TransformersConverter — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.converters.html b/python/ctranslate2.converters.html index 6ca1db451..d5a956696 100644 --- a/python/ctranslate2.converters.html +++ b/python/ctranslate2.converters.html @@ -4,7 +4,7 @@ - ctranslate2.converters — CTranslate2 4.4.0 documentation + ctranslate2.converters — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.get_cuda_device_count.html b/python/ctranslate2.get_cuda_device_count.html index d1dda16cb..ffb21119f 100644 --- a/python/ctranslate2.get_cuda_device_count.html +++ b/python/ctranslate2.get_cuda_device_count.html @@ -4,7 +4,7 @@ - get_cuda_device_count — CTranslate2 4.4.0 documentation + get_cuda_device_count — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.get_log_level.html b/python/ctranslate2.get_log_level.html index 596f50dd6..256615967 100644 --- a/python/ctranslate2.get_log_level.html +++ b/python/ctranslate2.get_log_level.html @@ -4,7 +4,7 @@ - get_log_level — CTranslate2 4.4.0 documentation + get_log_level — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.get_supported_compute_types.html b/python/ctranslate2.get_supported_compute_types.html index 19d1131aa..b83ec98e6 100644 --- a/python/ctranslate2.get_supported_compute_types.html +++ b/python/ctranslate2.get_supported_compute_types.html @@ -4,7 +4,7 @@ - get_supported_compute_types — CTranslate2 4.4.0 documentation + get_supported_compute_types — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.html b/python/ctranslate2.html index c5192831a..552044c4e 100644 --- a/python/ctranslate2.html +++ b/python/ctranslate2.html @@ -4,7 +4,7 @@ - ctranslate2 — CTranslate2 4.4.0 documentation + ctranslate2 — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.models.Wav2Vec2.html b/python/ctranslate2.models.Wav2Vec2.html index c18a35aba..ee0229180 100644 --- a/python/ctranslate2.models.Wav2Vec2.html +++ b/python/ctranslate2.models.Wav2Vec2.html @@ -4,7 +4,7 @@ - Wav2Vec2 — CTranslate2 4.4.0 documentation + Wav2Vec2 — CTranslate2 4.5.0 documentation @@ -20,7 +20,7 @@ - + @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5

        diff --git a/python/ctranslate2.models.Wav2Vec2Bert.html b/python/ctranslate2.models.Wav2Vec2Bert.html new file mode 100644 index 000000000..b37110035 --- /dev/null +++ b/python/ctranslate2.models.Wav2Vec2Bert.html @@ -0,0 +1,302 @@ + + + + + + + Wav2Vec2Bert — CTranslate2 4.5.0 documentation + + + + + + + + + + + + + + + + + + +
        + + +
        + +
        +
        +
        + +
        +
        +
        +
        + +
        +

        Wav2Vec2Bert

        +
        +
        +class ctranslate2.models.Wav2Vec2Bert
        +

        Implements the Wav2Vec2Bert speech recognition model published by Facebook.

        + +

        Inherits from: pybind11_builtins.pybind11_object

        +

        Attributes:

        + +

        Methods:

        + +
        +
        +__init__(model_path: str, device: str = 'cpu', *, device_index: Union[int, List[int]] = 0, compute_type: Union[str, Dict[str, str]] = 'default', inter_threads: int = 1, intra_threads: int = 0, max_queued_batches: int = 0, flash_attention: bool = False, tensor_parallel: bool = False, files: object = None) None
        +

        Initializes a Wav2Vec2Bert model from a converted model.

        +
        +
        Parameters
        +
          +
        • model_path – Path to the CTranslate2 model directory.

        • +
        • device – Device to use (possible values are: cpu, cuda, auto).

        • +
        • device_index – Device IDs where to place this model on.

        • +
        • compute_type – Model computation type or a dictionary mapping a device name +to the computation type (possible values are: default, auto, int8, int8_float32, +int8_float16, int8_bfloat16, int16, float16, bfloat16, float32).

        • +
        • inter_threads – Number of workers to allow executing multiple batches in parallel.

        • +
        • intra_threads – Number of OpenMP threads per worker (0 to use a default value).

        • +
        • max_queued_batches – Maximum numbers of batches in the worker queue (-1 for unlimited, +0 for an automatic value). When the queue is full, future requests will block +until a free slot is available.

        • +
        • flash_attention – run model with flash attention 2 for self-attention layer

        • +
        • tensor_parallel – run model with tensor parallel mode

        • +
        • files – Load model files from the memory. This argument is a dictionary mapping +file names to file contents as file-like or bytes objects. If this is set, +model_path acts as an identifier for this model.

        • +
        +
        +
        +
        + +
        +
        +encode(features: StorageView, to_cpu: bool = False) StorageView
        +

        Encodes the input features.

        +
        +
        Parameters
        +
          +
        • features – Mel spectogram of the audio, as a float array with shape +[batch_size, 80, 3000].

        • +
        • to_cpu – Copy the encoder output to the CPU before returning the value.

        • +
        +
        +
        Returns
        +

        The encoder output.

        +
        +
        +
        + +
        +
        +load_model(keep_cache: bool = False) None
        +

        Loads the model back to the initial device.

        +
        +
        Parameters
        +

        keep_cache – If True, the model cache in the CPU memory is not deleted if it exists.

        +
        +
        +
        + +
        +
        +unload_model(to_cpu: bool = False) None
        +

        Unloads the model attached to this wav2vec2bert but keep enough runtime context +to quickly resume wav2vec2bert on the initial device.

        +
        +
        Parameters
        +

        to_cpu – If True, the model is moved to the CPU memory and not fully unloaded.

        +
        +
        +
        + +
        +
        +property compute_type
        +

        Computation type used by the model.

        +
        + +
        +
        +property device
        +

        Device this model is running on.

        +
        + +
        +
        +property device_index
        +

        List of device IDs where this model is running on.

        +
        + +
        +
        +property model_is_loaded
        +

        Whether the model is loaded on the initial device and ready to be used.

        +
        + +
        +
        +property num_active_batches
        +

        Number of batches waiting to be processed or currently processed.

        +
        + +
        +
        +property num_queued_batches
        +

        Number of batches waiting to be processed.

        +
        + +
        +
        +property num_workers
        +

        Number of model workers backing this instance.

        +
        + +
        +
        +property tensor_parallel
        +

        Run model with tensor parallel mode.

        +
        + +
        + +
        + + +
        +
        + +
        +
        +
        +
        + + + + \ No newline at end of file diff --git a/python/ctranslate2.models.Whisper.html b/python/ctranslate2.models.Whisper.html index 47192f9cc..ce79ee5a5 100644 --- a/python/ctranslate2.models.Whisper.html +++ b/python/ctranslate2.models.Whisper.html @@ -4,7 +4,7 @@ - Whisper — CTranslate2 4.4.0 documentation + Whisper — CTranslate2 4.5.0 documentation @@ -21,7 +21,7 @@ - + @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        diff --git a/python/ctranslate2.models.WhisperGenerationResult.html b/python/ctranslate2.models.WhisperGenerationResult.html index 9eb518015..adc0bd1bc 100644 --- a/python/ctranslate2.models.WhisperGenerationResult.html +++ b/python/ctranslate2.models.WhisperGenerationResult.html @@ -4,7 +4,7 @@ - WhisperGenerationResult — CTranslate2 4.4.0 documentation + WhisperGenerationResult — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        @@ -80,6 +80,7 @@
      • ctranslate2.converters
      • ctranslate2.models diff --git a/python/ctranslate2.specs.LanguageModelSpec.html b/python/ctranslate2.specs.LanguageModelSpec.html index 7a8de234c..cd36ba8cf 100644 --- a/python/ctranslate2.specs.LanguageModelSpec.html +++ b/python/ctranslate2.specs.LanguageModelSpec.html @@ -4,7 +4,7 @@ - LanguageModelSpec — CTranslate2 4.4.0 documentation + LanguageModelSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
        - 4.4 + 4.5
        @@ -92,6 +92,7 @@
      • TransformerEncoderModelSpec
      • TransformerEncoderSpec
      • TransformerSpec
      • +
      • Wav2Vec2BertSpec
      • Wav2Vec2Spec
      • WhisperSpec
      @@ -142,6 +143,7 @@

      LanguageModelSpec
    • ctranslate2.specs.TransformerDecoderModelSpec

    • ctranslate2.specs.TransformerEncoderModelSpec

    • +
    • ctranslate2.specs.Wav2Vec2BertSpec

    • ctranslate2.specs.Wav2Vec2Spec

    • ctranslate2.specs.WhisperSpec

    diff --git a/python/ctranslate2.specs.LayerSpec.html b/python/ctranslate2.specs.LayerSpec.html index 3f3c9b162..8cb061644 100644 --- a/python/ctranslate2.specs.LayerSpec.html +++ b/python/ctranslate2.specs.LayerSpec.html @@ -4,7 +4,7 @@ - LayerSpec — CTranslate2 4.4.0 documentation + LayerSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.ModelSpec.html b/python/ctranslate2.specs.ModelSpec.html index 9eeeba679..f2e6a6557 100644 --- a/python/ctranslate2.specs.ModelSpec.html +++ b/python/ctranslate2.specs.ModelSpec.html @@ -4,7 +4,7 @@ - ModelSpec — CTranslate2 4.4.0 documentation + ModelSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.RotaryScalingType.html b/python/ctranslate2.specs.RotaryScalingType.html index b5c30c031..2feec3fc4 100644 --- a/python/ctranslate2.specs.RotaryScalingType.html +++ b/python/ctranslate2.specs.RotaryScalingType.html @@ -4,7 +4,7 @@ - RotaryScalingType — CTranslate2 4.4.0 documentation + RotaryScalingType — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.SequenceToSequenceModelSpec.html b/python/ctranslate2.specs.SequenceToSequenceModelSpec.html index 974c40acc..4da8c5164 100644 --- a/python/ctranslate2.specs.SequenceToSequenceModelSpec.html +++ b/python/ctranslate2.specs.SequenceToSequenceModelSpec.html @@ -4,7 +4,7 @@ - SequenceToSequenceModelSpec — CTranslate2 4.4.0 documentation + SequenceToSequenceModelSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.TransformerDecoderModelSpec.html b/python/ctranslate2.specs.TransformerDecoderModelSpec.html index 6a92a1495..443d17720 100644 --- a/python/ctranslate2.specs.TransformerDecoderModelSpec.html +++ b/python/ctranslate2.specs.TransformerDecoderModelSpec.html @@ -4,7 +4,7 @@ - TransformerDecoderModelSpec — CTranslate2 4.4.0 documentation + TransformerDecoderModelSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.TransformerDecoderSpec.html b/python/ctranslate2.specs.TransformerDecoderSpec.html index 6bb628089..2a87dbf2e 100644 --- a/python/ctranslate2.specs.TransformerDecoderSpec.html +++ b/python/ctranslate2.specs.TransformerDecoderSpec.html @@ -4,7 +4,7 @@ - TransformerDecoderSpec — CTranslate2 4.4.0 documentation + TransformerDecoderSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.TransformerEncoderModelSpec.html b/python/ctranslate2.specs.TransformerEncoderModelSpec.html index a9b00eaaf..bab9123a2 100644 --- a/python/ctranslate2.specs.TransformerEncoderModelSpec.html +++ b/python/ctranslate2.specs.TransformerEncoderModelSpec.html @@ -4,7 +4,7 @@ - TransformerEncoderModelSpec — CTranslate2 4.4.0 documentation + TransformerEncoderModelSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.TransformerEncoderSpec.html b/python/ctranslate2.specs.TransformerEncoderSpec.html index def76e3bd..e7a0bb6ba 100644 --- a/python/ctranslate2.specs.TransformerEncoderSpec.html +++ b/python/ctranslate2.specs.TransformerEncoderSpec.html @@ -4,7 +4,7 @@ - TransformerEncoderSpec — CTranslate2 4.4.0 documentation + TransformerEncoderSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.TransformerSpec.html b/python/ctranslate2.specs.TransformerSpec.html index 469105420..a9dbf6ff8 100644 --- a/python/ctranslate2.specs.TransformerSpec.html +++ b/python/ctranslate2.specs.TransformerSpec.html @@ -4,7 +4,7 @@ - TransformerSpec — CTranslate2 4.4.0 documentation + TransformerSpec — CTranslate2 4.5.0 documentation @@ -20,7 +20,7 @@ - + @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5

    diff --git a/python/ctranslate2.specs.Wav2Vec2BertSpec.html b/python/ctranslate2.specs.Wav2Vec2BertSpec.html new file mode 100644 index 000000000..b2d31d419 --- /dev/null +++ b/python/ctranslate2.specs.Wav2Vec2BertSpec.html @@ -0,0 +1,300 @@ + + + + + + + Wav2Vec2BertSpec — CTranslate2 4.5.0 documentation + + + + + + + + + + + + + + + + + + +
    + + +
    + +
    +
    +
    + +
    +
    +
    +
    + +
    +

    Wav2Vec2BertSpec

    +
    +
    +class ctranslate2.specs.Wav2Vec2BertSpec
    +

    Inherits from: ctranslate2.specs.LanguageModelSpec

    +

    Attributes:

    + +

    Methods:

    + +
    +
    +__init__(num_hidden_layers, num_adapter_layers)
    +

    Initializes a language model specification.

    +
    + +
    +
    +get_default_config()
    +

    Returns the default configuration used by this model.

    +
    + +
    +
    +get_vocabulary_size()
    +

    Returns the vocabulary size expected by the model.

    +
    + +
    +
    +optimize(quantization: Optional[str] = None) None
    +

    Recursively applies some optimizations to this layer:

    +
      +
    • Alias variables with the same shape and value.

    • +
    • Quantize weights.

    • +
    +
    +
    Parameters
    +

    quantization – Weight quantization scheme (possible values are: int8, int8_float32, +int8_float16, int8_bfloat16, int16, float16, bfloat16, float32).

    +
    +
    +
    + +
    +
    +register_file(path: str, filename: Optional[str] = None) None
    +

    Registers a file to be saved in the model directory.

    +
    + +
    +
    +register_vocabulary(tokens: List[str]) None
    +

    Registers the vocabulary of tokens.

    +
    +
    Parameters
    +

    tokens – List of tokens.

    +
    +
    +
    + +
    +
    +save(output_dir: str) None
    +

    Saves this model on disk.

    +
    +
    Parameters
    +

    output_dir – Output directory where the model is saved.

    +
    +
    +
    + +
    +
    +validate() None
    +

    Verify that the required weights are set.

    +
    +
    Raises
    +

    ValueError – If a required weight is not set in the specification.

    +
    +
    +
    + +
    +
    +variables(prefix: str = '', ordered: bool = False) Dict[str, ndarray]
    +

    Recursively returns the weights from this layer and its children.

    +
    +
    Parameters
    +
      +
    • prefix – Prefix to prepend to all variable names.

    • +
    • ordered – If set, an ordered list is returned instead.

    • +
    +
    +
    Returns
    +

    Dictionary mapping variables name to value.

    +
    +
    +
    + +
    +
    +property config
    +

    The model configuration.

    +
    + +
    +
    +property name
    +

    The name of the model specification.

    +
    + +
    +
    +property revision
    +

    The model specification revision.

    +

    This value is incremented each time the weights layout of the model is +changed (e.g. a weight is renamed).

    +
    + +
    + +
    + + +
    +
    + +
    +
    +
    +
    + + + + \ No newline at end of file diff --git a/python/ctranslate2.specs.Wav2Vec2Spec.html b/python/ctranslate2.specs.Wav2Vec2Spec.html index 3d7bfb794..4ff4db6d7 100644 --- a/python/ctranslate2.specs.Wav2Vec2Spec.html +++ b/python/ctranslate2.specs.Wav2Vec2Spec.html @@ -4,7 +4,7 @@ - Wav2Vec2Spec — CTranslate2 4.4.0 documentation + Wav2Vec2Spec — CTranslate2 4.5.0 documentation @@ -21,7 +21,7 @@ - + @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    diff --git a/python/ctranslate2.specs.WhisperSpec.html b/python/ctranslate2.specs.WhisperSpec.html index 35fcd6eae..eb88d23cc 100644 --- a/python/ctranslate2.specs.WhisperSpec.html +++ b/python/ctranslate2.specs.WhisperSpec.html @@ -4,7 +4,7 @@ - WhisperSpec — CTranslate2 4.4.0 documentation + WhisperSpec — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/ctranslate2.specs.html b/python/ctranslate2.specs.html index 1d4505ca2..49fd5e2ea 100644 --- a/python/ctranslate2.specs.html +++ b/python/ctranslate2.specs.html @@ -4,7 +4,7 @@ - ctranslate2.specs — CTranslate2 4.4.0 documentation + ctranslate2.specs — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -92,6 +92,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • @@ -146,6 +147,7 @@
  • TransformerEncoderModelSpec
  • TransformerEncoderSpec
  • TransformerSpec
  • +
  • Wav2Vec2BertSpec
  • Wav2Vec2Spec
  • WhisperSpec
  • diff --git a/python/overview.html b/python/overview.html index b8e8cca74..41faae644 100644 --- a/python/overview.html +++ b/python/overview.html @@ -4,7 +4,7 @@ - Python — CTranslate2 4.4.0 documentation + Python — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -155,6 +155,7 @@

    Pythonctranslate2.models diff --git a/quantization.html b/quantization.html index fabfc5d52..7fdb24e60 100644 --- a/quantization.html +++ b/quantization.html @@ -4,7 +4,7 @@ - Quantization — CTranslate2 4.4.0 documentation + Quantization — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    @@ -381,17 +381,28 @@

    16-bit brain floating points (

    4-bit AWQ

    -

    The compute type would be int32_float16

    Supported on:

    • NVIDIA GPU with Compute Capability >= 7.5

    -

    In this mode, all model weights are stored in half precision and all layers are run in half precision. Other parameters like scale and zero are stored in int32.

    -

    For example,

    +

    CTranslate2 internally handles the compute type for AWQ quantization. +In this mode, all model weights are stored in half precision and all layers are run in half precision. Other parameters like scale and zero are stored in int32.

    +

    Steps to use AWQ Quantization:

    +
      +
    • Download a AWQ quantized model from Hugging Face for example (TheBloke/Llama-2-7B-AWQ){https://huggingface.co/TheBloke/Llama-2-7B-AWQ} or quantize your own model with using this (AutoAWQ example){https://casper-hansen.github.io/AutoAWQ/examples/}.

    • +
    • Convert AWQ Quantized model to Ctranslate2 model:

    • +
     ct2-transformers-converter --model TheBloke/Llama-2-7B-AWQ --copy_files tokenizer.model --output_dir ct2_model
     
    -

    We have to quantize the model with AWQ first, then convert it to CT2 format.

    +
      +
    • Run inference as usual with Ctranslate2:

    • +
    +
    model = ctranslate2.Generator('ct2_model', device='cuda')
    +outputs = model.generate_batch([tokens])
    +
    +
    +

    Currently, CTranslate2 only supports the GEMM and GEMV kernels for AWQ quantization.

    diff --git a/quickstart.html b/quickstart.html index 025a0e5fd..c750e7296 100644 --- a/quickstart.html +++ b/quickstart.html @@ -4,7 +4,7 @@ - Quickstart — CTranslate2 4.4.0 documentation + Quickstart — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    diff --git a/search.html b/search.html index e3a4adf68..b3eed01b5 100644 --- a/search.html +++ b/search.html @@ -3,7 +3,7 @@ - Search — CTranslate2 4.4.0 documentation + Search — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    diff --git a/searchindex.js b/searchindex.js index b8860299e..c90c697fa 100644 --- a/searchindex.js +++ b/searchindex.js @@ -1 +1 @@ -Search.setIndex({docnames:["conversion","decoding","encoding","environment_variables","faq","generation","guides/fairseq","guides/marian","guides/opennmt_py","guides/opennmt_tf","guides/opus_mt","guides/transformers","hardware_support","index","installation","memory","parallel","performance","python/ctranslate2","python/ctranslate2.AsyncGenerationResult","python/ctranslate2.AsyncScoringResult","python/ctranslate2.AsyncTranslationResult","python/ctranslate2.DataType","python/ctranslate2.Device","python/ctranslate2.Encoder","python/ctranslate2.EncoderForwardOutput","python/ctranslate2.ExecutionStats","python/ctranslate2.GenerationResult","python/ctranslate2.GenerationStepResult","python/ctranslate2.Generator","python/ctranslate2.MpiInfo","python/ctranslate2.ScoringResult","python/ctranslate2.StorageView","python/ctranslate2.TranslationResult","python/ctranslate2.Translator","python/ctranslate2.contains_model","python/ctranslate2.converters","python/ctranslate2.converters.Converter","python/ctranslate2.converters.FairseqConverter","python/ctranslate2.converters.MarianConverter","python/ctranslate2.converters.OpenAIGPT2Converter","python/ctranslate2.converters.OpenNMTPyConverter","python/ctranslate2.converters.OpenNMTTFConverter","python/ctranslate2.converters.OpusMTConverter","python/ctranslate2.converters.TransformersConverter","python/ctranslate2.get_cuda_device_count","python/ctranslate2.get_log_level","python/ctranslate2.get_supported_compute_types","python/ctranslate2.models","python/ctranslate2.models.Wav2Vec2","python/ctranslate2.models.Whisper","python/ctranslate2.models.WhisperGenerationResult","python/ctranslate2.models.WhisperGenerationResultAsync","python/ctranslate2.set_log_level","python/ctranslate2.set_random_seed","python/ctranslate2.specs","python/ctranslate2.specs.Activation","python/ctranslate2.specs.EmbeddingsMerge","python/ctranslate2.specs.LanguageModelSpec","python/ctranslate2.specs.LayerSpec","python/ctranslate2.specs.ModelSpec","python/ctranslate2.specs.RotaryScalingType","python/ctranslate2.specs.SequenceToSequenceModelSpec","python/ctranslate2.specs.TransformerDecoderModelSpec","python/ctranslate2.specs.TransformerDecoderSpec","python/ctranslate2.specs.TransformerEncoderModelSpec","python/ctranslate2.specs.TransformerEncoderSpec","python/ctranslate2.specs.TransformerSpec","python/ctranslate2.specs.Wav2Vec2Spec","python/ctranslate2.specs.WhisperSpec","python/overview","quantization","quickstart","speech_recognition","translation","versioning"],envversion:{"sphinx.domains.c":2,"sphinx.domains.changeset":1,"sphinx.domains.citation":1,"sphinx.domains.cpp":5,"sphinx.domains.index":1,"sphinx.domains.javascript":2,"sphinx.domains.math":2,"sphinx.domains.python":3,"sphinx.domains.rst":2,"sphinx.domains.std":2,sphinx:56},filenames:["conversion.md","decoding.md","encoding.md","environment_variables.md","faq.md","generation.md","guides/fairseq.md","guides/marian.md","guides/opennmt_py.md","guides/opennmt_tf.md","guides/opus_mt.md","guides/transformers.md","hardware_support.md","index.rst","installation.md","memory.md","parallel.md","performance.md","python/ctranslate2.rst","python/ctranslate2.AsyncGenerationResult.rst","python/ctranslate2.AsyncScoringResult.rst","python/ctranslate2.AsyncTranslationResult.rst","python/ctranslate2.DataType.rst","python/ctranslate2.Device.rst","python/ctranslate2.Encoder.rst","python/ctranslate2.EncoderForwardOutput.rst","python/ctranslate2.ExecutionStats.rst","python/ctranslate2.GenerationResult.rst","python/ctranslate2.GenerationStepResult.rst","python/ctranslate2.Generator.rst","python/ctranslate2.MpiInfo.rst","python/ctranslate2.ScoringResult.rst","python/ctranslate2.StorageView.rst","python/ctranslate2.TranslationResult.rst","python/ctranslate2.Translator.rst","python/ctranslate2.contains_model.rst","python/ctranslate2.converters.rst","python/ctranslate2.converters.Converter.rst","python/ctranslate2.converters.FairseqConverter.rst","python/ctranslate2.converters.MarianConverter.rst","python/ctranslate2.converters.OpenAIGPT2Converter.rst","python/ctranslate2.converters.OpenNMTPyConverter.rst","python/ctranslate2.converters.OpenNMTTFConverter.rst","python/ctranslate2.converters.OpusMTConverter.rst","python/ctranslate2.converters.TransformersConverter.rst","python/ctranslate2.get_cuda_device_count.rst","python/ctranslate2.get_log_level.rst","python/ctranslate2.get_supported_compute_types.rst","python/ctranslate2.models.rst","python/ctranslate2.models.Wav2Vec2.rst","python/ctranslate2.models.Whisper.rst","python/ctranslate2.models.WhisperGenerationResult.rst","python/ctranslate2.models.WhisperGenerationResultAsync.rst","python/ctranslate2.set_log_level.rst","python/ctranslate2.set_random_seed.rst","python/ctranslate2.specs.rst","python/ctranslate2.specs.Activation.rst","python/ctranslate2.specs.EmbeddingsMerge.rst","python/ctranslate2.specs.LanguageModelSpec.rst","python/ctranslate2.specs.LayerSpec.rst","python/ctranslate2.specs.ModelSpec.rst","python/ctranslate2.specs.RotaryScalingType.rst","python/ctranslate2.specs.SequenceToSequenceModelSpec.rst","python/ctranslate2.specs.TransformerDecoderModelSpec.rst","python/ctranslate2.specs.TransformerDecoderSpec.rst","python/ctranslate2.specs.TransformerEncoderModelSpec.rst","python/ctranslate2.specs.TransformerEncoderSpec.rst","python/ctranslate2.specs.TransformerSpec.rst","python/ctranslate2.specs.Wav2Vec2Spec.rst","python/ctranslate2.specs.WhisperSpec.rst","python/overview.rst","quantization.md","quickstart.md","speech_recognition.md","translation.md","versioning.md"],objects:{"":[[18,0,0,"-","ctranslate2"]],"ctranslate2.AsyncGenerationResult":[[19,2,1,"","done"],[19,2,1,"","result"]],"ctranslate2.AsyncScoringResult":[[20,2,1,"","done"],[20,2,1,"","result"]],"ctranslate2.AsyncTranslationResult":[[21,2,1,"","done"],[21,2,1,"","result"]],"ctranslate2.DataType":[[22,2,1,"","__init__"],[22,3,1,"","bfloat16"],[22,3,1,"","float16"],[22,3,1,"","float32"],[22,3,1,"","int16"],[22,3,1,"","int32"],[22,3,1,"","int8"],[22,4,1,"","name"],[22,4,1,"","value"]],"ctranslate2.Device":[[23,2,1,"","__init__"],[23,3,1,"","cpu"],[23,3,1,"","cuda"],[23,4,1,"","name"],[23,4,1,"","value"]],"ctranslate2.Encoder":[[24,2,1,"","__init__"],[24,4,1,"","compute_type"],[24,4,1,"","device"],[24,4,1,"","device_index"],[24,2,1,"","forward_batch"],[24,2,1,"","load_model"],[24,4,1,"","model_is_loaded"],[24,4,1,"","num_active_batches"],[24,4,1,"","num_encoders"],[24,4,1,"","num_queued_batches"],[24,4,1,"","tensor_parallel"],[24,2,1,"","unload_model"]],"ctranslate2.EncoderForwardOutput":[[25,4,1,"","last_hidden_state"],[25,4,1,"","pooler_output"]],"ctranslate2.ExecutionStats":[[26,4,1,"","num_examples"],[26,4,1,"","num_tokens"],[26,4,1,"","total_time_in_ms"]],"ctranslate2.GenerationResult":[[27,4,1,"","logits"],[27,4,1,"","scores"],[27,4,1,"","sequences"],[27,4,1,"","sequences_ids"]],"ctranslate2.GenerationStepResult":[[28,4,1,"","batch_id"],[28,4,1,"","hypothesis_id"],[28,4,1,"","is_last"],[28,4,1,"","log_prob"],[28,4,1,"","logits"],[28,4,1,"","step"],[28,4,1,"","token"],[28,4,1,"","token_id"]],"ctranslate2.Generator":[[29,2,1,"","__init__"],[29,2,1,"","async_generate_tokens"],[29,4,1,"","compute_type"],[29,4,1,"","device"],[29,4,1,"","device_index"],[29,2,1,"","forward_batch"],[29,2,1,"","generate_batch"],[29,2,1,"","generate_iterable"],[29,2,1,"","generate_tokens"],[29,2,1,"","load_model"],[29,4,1,"","model_is_loaded"],[29,4,1,"","num_active_batches"],[29,4,1,"","num_generators"],[29,4,1,"","num_queued_batches"],[29,2,1,"","score_batch"],[29,2,1,"","score_iterable"],[29,4,1,"","tensor_parallel"],[29,2,1,"","unload_model"]],"ctranslate2.MpiInfo":[[30,2,1,"","getCurRank"],[30,2,1,"","getLocalRank"],[30,2,1,"","getNRanks"]],"ctranslate2.ScoringResult":[[31,4,1,"","log_probs"],[31,4,1,"","tokens"]],"ctranslate2.StorageView":[[32,4,1,"","device"],[32,4,1,"","device_index"],[32,4,1,"","dtype"],[32,2,1,"","from_array"],[32,4,1,"","shape"],[32,2,1,"","to"],[32,2,1,"","to_device"]],"ctranslate2.TranslationResult":[[33,4,1,"","attention"],[33,4,1,"","hypotheses"],[33,4,1,"","logits"],[33,4,1,"","scores"]],"ctranslate2.Translator":[[34,2,1,"","__init__"],[34,4,1,"","compute_type"],[34,4,1,"","device"],[34,4,1,"","device_index"],[34,2,1,"","generate_tokens"],[34,2,1,"","load_model"],[34,4,1,"","model_is_loaded"],[34,4,1,"","num_active_batches"],[34,4,1,"","num_queued_batches"],[34,4,1,"","num_translators"],[34,2,1,"","score_batch"],[34,2,1,"","score_file"],[34,2,1,"","score_iterable"],[34,4,1,"","tensor_parallel"],[34,2,1,"","translate_batch"],[34,2,1,"","translate_file"],[34,2,1,"","translate_iterable"],[34,2,1,"","unload_model"]],"ctranslate2.converters":[[37,1,1,"","Converter"],[38,1,1,"","FairseqConverter"],[39,1,1,"","MarianConverter"],[40,1,1,"","OpenAIGPT2Converter"],[41,1,1,"","OpenNMTPyConverter"],[42,1,1,"","OpenNMTTFConverter"],[43,1,1,"","OpusMTConverter"],[44,1,1,"","TransformersConverter"]],"ctranslate2.converters.Converter":[[37,2,1,"","convert"],[37,2,1,"","convert_from_args"],[37,2,1,"","declare_arguments"]],"ctranslate2.converters.FairseqConverter":[[38,2,1,"","__init__"],[38,2,1,"","convert"],[38,2,1,"","convert_from_args"],[38,2,1,"","declare_arguments"]],"ctranslate2.converters.MarianConverter":[[39,2,1,"","__init__"],[39,2,1,"","convert"],[39,2,1,"","convert_from_args"],[39,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenAIGPT2Converter":[[40,2,1,"","__init__"],[40,2,1,"","convert"],[40,2,1,"","convert_from_args"],[40,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenNMTPyConverter":[[41,2,1,"","__init__"],[41,2,1,"","convert"],[41,2,1,"","convert_from_args"],[41,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenNMTTFConverter":[[42,2,1,"","__init__"],[42,2,1,"","convert"],[42,2,1,"","convert_from_args"],[42,2,1,"","declare_arguments"],[42,2,1,"","from_config"]],"ctranslate2.converters.OpusMTConverter":[[43,2,1,"","__init__"],[43,2,1,"","convert"],[43,2,1,"","convert_from_args"],[43,2,1,"","declare_arguments"]],"ctranslate2.converters.TransformersConverter":[[44,2,1,"","__init__"],[44,2,1,"","convert"],[44,2,1,"","convert_from_args"],[44,2,1,"","declare_arguments"],[44,2,1,"","get_model_file"],[44,2,1,"","load_model"],[44,2,1,"","load_tokenizer"]],"ctranslate2.models":[[49,1,1,"","Wav2Vec2"],[50,1,1,"","Whisper"],[51,1,1,"","WhisperGenerationResult"],[52,1,1,"","WhisperGenerationResultAsync"]],"ctranslate2.models.Wav2Vec2":[[49,2,1,"","__init__"],[49,4,1,"","compute_type"],[49,4,1,"","device"],[49,4,1,"","device_index"],[49,2,1,"","encode"],[49,2,1,"","load_model"],[49,4,1,"","model_is_loaded"],[49,4,1,"","num_active_batches"],[49,4,1,"","num_queued_batches"],[49,4,1,"","num_workers"],[49,4,1,"","tensor_parallel"],[49,2,1,"","unload_model"]],"ctranslate2.models.Whisper":[[50,2,1,"","__init__"],[50,2,1,"","align"],[50,4,1,"","compute_type"],[50,2,1,"","detect_language"],[50,4,1,"","device"],[50,4,1,"","device_index"],[50,2,1,"","encode"],[50,2,1,"","generate"],[50,4,1,"","is_multilingual"],[50,2,1,"","load_model"],[50,4,1,"","model_is_loaded"],[50,4,1,"","n_mels"],[50,4,1,"","num_active_batches"],[50,4,1,"","num_languages"],[50,4,1,"","num_queued_batches"],[50,4,1,"","num_workers"],[50,4,1,"","tensor_parallel"],[50,2,1,"","unload_model"]],"ctranslate2.models.WhisperGenerationResult":[[51,4,1,"","no_speech_prob"],[51,4,1,"","scores"],[51,4,1,"","sequences"],[51,4,1,"","sequences_ids"]],"ctranslate2.models.WhisperGenerationResultAsync":[[52,2,1,"","done"],[52,2,1,"","result"]],"ctranslate2.specs":[[56,1,1,"","Activation"],[57,1,1,"","EmbeddingsMerge"],[58,1,1,"","LanguageModelSpec"],[59,1,1,"","LayerSpec"],[60,1,1,"","ModelSpec"],[61,1,1,"","RotaryScalingType"],[62,1,1,"","SequenceToSequenceModelSpec"],[63,1,1,"","TransformerDecoderModelSpec"],[64,1,1,"","TransformerDecoderSpec"],[65,1,1,"","TransformerEncoderModelSpec"],[66,1,1,"","TransformerEncoderSpec"],[67,1,1,"","TransformerSpec"],[68,1,1,"","Wav2Vec2Spec"],[69,1,1,"","WhisperSpec"]],"ctranslate2.specs.Activation":[[56,3,1,"","GELU"],[56,3,1,"","GELUSigmoid"],[56,3,1,"","GELUTanh"],[56,3,1,"","RELU"],[56,3,1,"","SWISH"],[56,3,1,"","Tanh"]],"ctranslate2.specs.EmbeddingsMerge":[[57,3,1,"","ADD"],[57,3,1,"","CONCAT"]],"ctranslate2.specs.LanguageModelSpec":[[58,2,1,"","__init__"],[58,4,1,"","config"],[58,2,1,"","get_default_config"],[58,2,1,"","get_vocabulary_size"],[58,4,1,"","name"],[58,2,1,"","optimize"],[58,2,1,"","register_file"],[58,2,1,"","register_vocabulary"],[58,4,1,"","revision"],[58,2,1,"","save"],[58,2,1,"","validate"],[58,2,1,"","variables"]],"ctranslate2.specs.LayerSpec":[[59,2,1,"","optimize"],[59,2,1,"","validate"],[59,2,1,"","variables"]],"ctranslate2.specs.ModelSpec":[[60,2,1,"","__init__"],[60,4,1,"","config"],[60,2,1,"","get_default_config"],[60,4,1,"","name"],[60,2,1,"","optimize"],[60,2,1,"","register_file"],[60,4,1,"","revision"],[60,2,1,"","save"],[60,2,1,"","validate"],[60,2,1,"","variables"]],"ctranslate2.specs.RotaryScalingType":[[61,3,1,"","Linear"],[61,3,1,"","Llama3"],[61,3,1,"","Su"]],"ctranslate2.specs.SequenceToSequenceModelSpec":[[62,2,1,"","__init__"],[62,4,1,"","config"],[62,2,1,"","get_default_config"],[62,2,1,"","get_source_vocabulary_size"],[62,2,1,"","get_target_vocabulary_size"],[62,4,1,"","name"],[62,2,1,"","optimize"],[62,2,1,"","register_file"],[62,2,1,"","register_source_vocabulary"],[62,2,1,"","register_target_vocabulary"],[62,2,1,"","register_vocabulary_mapping"],[62,4,1,"","revision"],[62,2,1,"","save"],[62,2,1,"","validate"],[62,2,1,"","variables"]],"ctranslate2.specs.TransformerDecoderModelSpec":[[63,2,1,"","__init__"],[63,4,1,"","config"],[63,2,1,"","from_config"],[63,2,1,"","get_default_config"],[63,2,1,"","get_vocabulary_size"],[63,4,1,"","name"],[63,2,1,"","optimize"],[63,2,1,"","register_file"],[63,2,1,"","register_vocabulary"],[63,4,1,"","revision"],[63,2,1,"","save"],[63,2,1,"","validate"],[63,2,1,"","variables"]],"ctranslate2.specs.TransformerDecoderSpec":[[64,2,1,"","__init__"],[64,4,1,"","config"],[64,2,1,"","optimize"],[64,2,1,"","validate"],[64,2,1,"","variables"]],"ctranslate2.specs.TransformerEncoderModelSpec":[[65,2,1,"","__init__"],[65,4,1,"","config"],[65,2,1,"","get_default_config"],[65,2,1,"","get_vocabulary_size"],[65,4,1,"","name"],[65,2,1,"","optimize"],[65,2,1,"","register_file"],[65,2,1,"","register_vocabulary"],[65,4,1,"","revision"],[65,2,1,"","save"],[65,2,1,"","validate"],[65,2,1,"","variables"]],"ctranslate2.specs.TransformerEncoderSpec":[[66,2,1,"","__init__"],[66,2,1,"","optimize"],[66,2,1,"","validate"],[66,2,1,"","variables"]],"ctranslate2.specs.TransformerSpec":[[67,2,1,"","__init__"],[67,4,1,"","config"],[67,2,1,"","from_config"],[67,2,1,"","get_default_config"],[67,2,1,"","get_source_vocabulary_size"],[67,2,1,"","get_target_vocabulary_size"],[67,4,1,"","name"],[67,2,1,"","optimize"],[67,2,1,"","register_file"],[67,2,1,"","register_source_vocabulary"],[67,2,1,"","register_target_vocabulary"],[67,2,1,"","register_vocabulary_mapping"],[67,4,1,"","revision"],[67,2,1,"","save"],[67,2,1,"","validate"],[67,2,1,"","variables"]],"ctranslate2.specs.Wav2Vec2Spec":[[68,2,1,"","__init__"],[68,4,1,"","config"],[68,2,1,"","get_default_config"],[68,2,1,"","get_vocabulary_size"],[68,4,1,"","name"],[68,2,1,"","optimize"],[68,2,1,"","register_file"],[68,2,1,"","register_vocabulary"],[68,4,1,"","revision"],[68,2,1,"","save"],[68,2,1,"","validate"],[68,2,1,"","variables"]],"ctranslate2.specs.WhisperSpec":[[69,2,1,"","__init__"],[69,4,1,"","config"],[69,2,1,"","get_default_config"],[69,2,1,"","get_vocabulary_size"],[69,4,1,"","name"],[69,2,1,"","optimize"],[69,2,1,"","register_file"],[69,2,1,"","register_vocabulary"],[69,4,1,"","revision"],[69,2,1,"","save"],[69,2,1,"","validate"],[69,2,1,"","variables"]],ctranslate2:[[19,1,1,"","AsyncGenerationResult"],[20,1,1,"","AsyncScoringResult"],[21,1,1,"","AsyncTranslationResult"],[22,1,1,"","DataType"],[23,1,1,"","Device"],[24,1,1,"","Encoder"],[25,1,1,"","EncoderForwardOutput"],[26,1,1,"","ExecutionStats"],[27,1,1,"","GenerationResult"],[28,1,1,"","GenerationStepResult"],[29,1,1,"","Generator"],[30,1,1,"","MpiInfo"],[31,1,1,"","ScoringResult"],[32,1,1,"","StorageView"],[33,1,1,"","TranslationResult"],[34,1,1,"","Translator"],[35,5,1,"","contains_model"],[36,0,0,"-","converters"],[45,5,1,"","get_cuda_device_count"],[46,5,1,"","get_log_level"],[47,5,1,"","get_supported_compute_types"],[48,0,0,"-","models"],[53,5,1,"","set_log_level"],[54,5,1,"","set_random_seed"],[55,0,0,"-","specs"]]},objnames:{"0":["py","module","Python module"],"1":["py","class","Python class"],"2":["py","method","Python method"],"3":["py","attribute","Python attribute"],"4":["py","property","Python property"],"5":["py","function","Python function"]},objtypes:{"0":"py:module","1":"py:class","2":"py:method","3":"py:attribute","4":"py:property","5":"py:function"},terms:{"0":[0,1,3,5,6,10,11,14,16,17,22,23,24,29,32,34,47,49,50,51,56,57,61,63,64,71,72,74,75],"001":1,"02":10,"02155":[64,66,67],"04":[14,74],"05202":[63,64,66,67],"05424":[29,34,50],"1":[0,1,3,5,6,11,12,16,17,22,23,24,29,32,34,49,50,56,57,61,63,64,66,67,71,72],"10":[1,5,6,11,71,72],"1000":10,"10000":[63,64],"100mb":71,"1024":[1,5,29,34],"10683":[64,66,67],"11":[3,14,72],"12":[3,14,71],"127":71,"128k":6,"15":[0,14],"16000":11,"17":[0,14],"1803":[64,66,67],"182mb":71,"187mb":71,"19":14,"1910":[64,66,67],"1t":11,"2":[0,1,3,5,8,14,16,17,22,24,29,32,34,40,49,50,56,61,63,67,71,72,74,75],"20":[5,11,29,71],"200":11,"2002":[63,64,66,67],"200mb":3,"2016":71,"2017":71,"2018":1,"2019":14,"2020":[10,17],"209715200":3,"20b":11,"21":11,"2204":[29,34,50],"23":11,"24gb":11,"256":34,"26":10,"2b":6,"2b_last_checkpoint":6,"2x4":32,"3":[0,1,3,12,14,16,22,56,72],"30":11,"3000":49,"32":[29,34],"32k":71,"350m":11,"364mb":71,"39m":11,"4":[1,3,11,12,16,22,32,56,72],"418m":6,"418m_last_checkpoint":6,"448":50,"4bit":[63,64],"5":[1,6,12,14,22,50,56,71,72],"50":[29,50],"50277":5,"50278":5,"50279":5,"512":[5,29],"560m":11,"6":71,"600m":11,"6291455":3,"64":[11,12,14,29,34,71],"6b":11,"7":[3,5,11,14,17,50,71],"70":11,"7b":[5,11,71],"8":[3,5,11,14,16,72],"80":49,"800":11,"9":11,"95mb":71,"\u00fcbersetzungen":1,"\u00fcbersetzungsmodel":1,"\u00fcbersetzungsmodellen":1,"abstract":[58,62],"boolean":3,"break":5,"byte":[24,29,34,49,50],"case":[9,14,15,16,72],"char":72,"class":[2,5,11,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,48,49,50,51,52,56,57,58,59,60,61,62,63,64,65,66,67,68,69,74],"const":[72,74],"default":[1,3,6,14,16,17,24,29,34,49,50,58,60,62,63,65,67,68,69,71],"do":[1,5,63,71],"enum":[56,57,61],"erm\u00f6glicht":1,"export":[3,4,16,72],"f\u00fcr":1,"final":[0,1,17,63,64,66,67],"float":[29,34,49,50,63,64],"function":[0,1,3,29,34,35,37,38,39,40,41,42,43,44,53,71],"import":[3,5,6,10,11,53,72,74],"int":[22,23,24,29,30,34,45,46,47,49,50,53,54,63,64,66,67,72,74],"long":[0,29],"new":[6,14,32],"return":[1,5,11,16,19,20,21,24,29,32,34,37,38,39,40,41,42,43,44,45,46,47,49,50,52,58,59,60,62,63,64,65,66,67,68,69,74],"short":5,"static":[5,29,30,32,37,38,39,40,41,42,43,44],"true":[1,3,5,11,16,19,20,21,24,29,34,49,50,52,63,64,66,67],"try":[1,16],A:[0,1,24,26,27,29,31,32,33,34,42,46,48,50,51,53,59,74],As:71,At:5,By:[1,3,14,71],For:[0,3,5,6,11,13,17,29,50,71],If:[5,9,11,12,14,16,17,19,20,21,24,29,34,37,38,39,40,41,42,43,44,49,50,52,58,59,60,62,63,64,65,66,67,68,69,72,74],In:[0,1,9,11,12,15,16,71,74],Is:6,It:[0,1,5,11,14,29,30,34,72,74],No:[5,6],ON:14,On:[12,14,71],The:[0,1,2,3,4,5,6,7,8,10,11,12,13,14,15,16,17,24,28,29,31,32,34,49,50,53,58,60,62,63,64,65,67,68,69,71,72,74,75],There:[0,4,6],These:1,To:[5,11,14,15,74],_:[11,16],__de__:6,__en__:6,__init__:[22,23,24,29,34,38,39,40,41,42,43,44,49,50,58,60,62,63,64,65,66,67,68,69],__x__:6,_batch:17,_file:17,_iter:17,aarch64:[12,14,71],ab:[29,34,50,63,64,66,67,71],abc:37,aber:1,abl:[4,5],about:[3,30,71],abov:[3,11],absolut:71,accelar:4,acceler:[0,1,14,29,71],accept:[6,11,29,34,50,71],access:[0,9,11],accord:16,accumul:[3,71],accuraci:[11,44,71],act:[24,29,34,49,50],activ:[11,44,55,63,64,65,66,67,70],activation_scal:[11,44],actual:71,ad:[0,4,5,11,29,74],adapt:5,add:[11,14,37,38,39,40,41,42,43,44,57,63,64,65,72],add_execut:72,addit:[14,16,74],address:4,advanc:74,affect:11,after:[1,29,34,63,64,65,66,67],against:14,aggreg:11,agnost:[0,4],ai:5,aim:11,al:[1,71],algorithm:1,alia:[58,59,60,62,63,64,65,66,67,68,69],alibi:[63,64],alibi_use_positive_posit:[63,64],align:[50,64,67],alignment_head:[64,67],alignment_lay:[64,67],all:[1,6,11,16,28,38,58,59,60,62,63,64,65,66,67,68,69,71,74],alloc:[3,32],allow:[1,3,44,49,50],alpha:5,alreadi:[7,11,37,38,39,40,41,42,43,44],also:[1,3,5,6,8,9,10,11,14,16,24,29,50,71],altern:[8,9,29,34],alwai:[1,5,8,16,74],am:11,amazonaw:72,amd:[12,14],amid:11,an:[0,1,4,5,8,11,14,17,19,20,21,24,25,29,30,32,34,42,49,50,52,58,59,60,62,63,64,65,66,67,68,69,74],ani:[16,29,34],anim:11,anoth:[0,6,14,32,63,71],anyth:5,api:[0,1,3,72,74,75],appear:1,append:[5,11],appl:[14,71],appli:[1,6,10,17,29,34,50,58,59,60,62,63,64,65,66,67,68,69,71],applic:[4,14,16],approach:4,ar:[0,1,3,5,6,7,8,10,11,12,14,16,17,24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69,71,72,74,75],arbitrarili:[29,34],architectur:[0,3,4,11,12,14,63,64,66,67,71],arg:[37,38,39,40,41,42,43,44],argc:72,argmax:11,argument:[1,5,11,17,24,29,34,37,38,39,40,41,42,43,44,49,50,53],argumentpars:[37,38,39,40,41,42,43,44],argv:72,arm64:[12,14,16,71],around:[1,19,20,21,52],arrai:[11,24,29,32,49,50],articl:3,arxiv:[29,34,50,63,64,66,67],as_tensor:[11,32],ask:13,assert:1,assioc:74,assist:5,associ:74,assumpt:4,async:29,async_generate_token:29,async_result:16,asyncgenerationresult:[18,29,70],asynchron:[3,19,20,21,29,34,50,52,74],asynciter:29,asyncscoringresult:[18,29,34,70],asynctranslationresult:[18,34,70],attach:[24,29,34,49,50],attent:[24,29,33,34,49,50,63,64,66,67,69],attribut:[22,23,24,25,26,27,28,29,31,32,33,34,49,50,51,58,60,62,63,64,65,67,68,69],auch:1,audio:[11,49,50],auf:1,ausgerichtet:1,author:71,auto:[14,24,29,34,49,50,71,72,74],auto_config:42,automat:[3,11,12,24,29,34,42,49,50],automodelforsequenceclassif:11,autoregress:11,autotoken:[5,11],avail:[1,16,19,20,21,24,29,34,49,50,52,74],averag:72,avoid:[3,5,17],avx2:[3,12],avx512:[3,12,17],avx:[3,12],awq:[63,64],back:[1,15,24,29,34,49,50],backend:[3,12,14,71],background:5,backward:75,bart:6,base:[3,11,16,37,58,62,63,64,71],basic:1,batch:[5,16,17,24,28,29,34,49,50,72,74],batch_gener:16,batch_id:28,batch_siz:[24,29,49,50],batch_typ:[17,29,34],bdist_wheel:14,beam:[17,29,34,50],beam_siz:[1,6,17,29,34,50],becaus:16,bedienen:1,bedienung:1,been:3,befor:[3,11,34,49,50,63,64],begin:50,behavior:[6,12,16,71],believ:11,below:[1,5,11,12,16,17,71],benchmark:71,bert:[2,65],bertmodel:11,beschleunigen:1,beschleunigung:1,best:[1,3,12,50],better:[1,3,4,11],between:[5,16,30,50,71],bf16:[3,71],bfloat16:[22,24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69],bia:[1,64,66,67],bias:[34,63,64],big:16,bigcod:11,bigscienc:11,billion:11,bin:[0,6],bin_growth:3,binari:[0,12,14,71],bit:[11,63,64],blackout:11,blank:50,block:[16,19,20,21,24,29,34,49,50,52,63,64],blue:6,bold:14,bool:[19,20,21,24,29,34,35,37,38,39,40,41,42,43,44,49,50,52,58,59,60,62,63,64,65,66,67,68,69],boost:16,bos_token:11,both:[4,12,74],bpe:6,branch:11,broken:0,brows:72,bs:16,buffer:[3,32],build:[0,72],build_cli:14,build_test:14,built:[29,34,42],bunch:72,bz2:6,c:[0,4,29,34,53,75],cach:[3,17,24,29,34,49,50,63,64],cache_static_prompt:29,calcul:34,call:[4,5,11,16,29,34,37,38,39,40,41,42,43,44,71],callabl:[29,34],callback:[5,29,34],can:[0,1,3,5,6,8,9,11,14,15,16,17,24,29,32,34,42,71,72,74],candid:[1,29,34,50,74],cannot:[37,38,39,40,41,42,43,44],capabl:[12,17,71],catalog:42,cd:[14,72],chang:[0,1,58,60,62,63,65,67,68,69,71],chat:[5,11],cheap:11,check:[0,16,35],checkpoint:[9,42],checkpoint_path:42,chief:6,children:[58,59,60,62,63,64,65,66,67,68,69],chunk:[17,29,34],chunk_length:50,classif:2,classifi:11,classmethod:[42,63,67],clear:5,cli:11,client:[14,75],clone:14,close:5,cmake:[14,72],cmake_cxx_flag:14,cmake_install_prefix:14,cmake_minimum_requir:72,cmake_prefix_path:72,cmakelist:72,cnn:11,code:[0,3,4,6,11,44,72],codegen:11,collect:[10,11,48],com:[6,11,14,40,44,49,50,72],combin:1,come:0,comma:3,command:[8,11,14,37,38,39,40,41,42,43,44],common:[1,37,38,39,40,41,42,43,44,71],commun:30,comp:14,compar:[11,71],compat:[10,12,29,34,75],compil:[12,72],complet:[1,5,8,11,29,34],compress:1,comput:[3,4,5,6,11,12,16,17,19,20,21,24,29,34,44,47,49,50,52,71],compute_typ:[24,29,34,49,50,71],concat:[57,66,67],concurr:[16,29,34],condit:11,confid:1,config:[0,9,16,42,50,58,60,62,63,64,65,67,68,69],configur:[0,3,9,12,14,16,42,58,60,62,63,65,67,68,69],connect:[16,63,64],connexion:30,consciou:11,consid:[5,11,17,29,72],constant:17,constrain:29,constructor:16,contain:[0,6,10,14,26,35,37,38,39,40,41,42,43,44,72],contains_model:[0,18,70],content:[24,29,34,49,50],context:[1,24,29,34,49,50],continu:[29,34,50],control:[1,3,4,12,53],conv:7,conveni:[5,10,53],convers:[4,6,7,8,9,11,13,37,38,39,40,41,42,43,44],convert:[6,7,8,9,10,11,32,38,39,40,41,42,43,44,49,50,59,70,71,72,74,75],convert_from_arg:[37,38,39,40,41,42,43,44],convert_ids_to_token:[5,11],convert_tokens_to_id:11,convolut:14,copi:[5,32,44,49,50],copy_fil:[44,71],core:[0,3,4,17],correspond:[1,9,11,71],costli:3,could:[0,4,5,16],cout:[72,74],cover:[4,75],coverag:34,coverage_penalti:34,cpp:72,cpu:[0,3,4,5,6,11,14,15,23,24,29,32,34,47,49,50,71,72,74],creat:[16,24,29,32,42,63,67],critic:3,csc:10,ct2:[5,6,7,8,9,10,11,71,72],ct2_force_cpu_isa:12,ct2_model:[5,6,7,8,9,10,11,71],ct2_use_experimental_packed_gemm:17,ct2_use_mkl:12,ct2_verbos:71,ctranslat:72,ctranslate2:[0,1,2,3,4,5,6,7,8,9,10,11,13,14,16,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,37,38,39,40,41,42,43,44,45,46,47,49,50,51,52,53,54,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75],ctranslate2_demo:72,ctranslate2_root:14,ctranslate_install_path:72,cub:3,cub_cach:3,cubla:[4,14],cuda11:[14,74],cuda:[3,11,12,14,16,23,24,29,32,34,47,49,50,74],cuda_arch_list:14,cuda_dynamic_load:14,cuda_malloc_async:3,cuda_nvcc_flag:14,cuda_select_nvcc_arch_flag:14,cudamalloc:3,cudnn:14,cumul:[29,34],current:[4,6,11,24,29,30,34,46,49,50,71],custom:[3,4,11,14,16,38,44,72],d_model:50,da:1,daili:11,data:[6,17,29,32,34,38,74],data_dir:[6,38],dataset:11,datatyp:[18,32,70],dazu:1,dcmake_install_prefix:72,dcmake_prefix_path:72,de:[1,6,10,11,72],debug:3,decid:3,declar:[38,59],declare_argu:[37,38,39,40,41,42,43,44],decod:[3,5,6,8,10,11,13,17,28,29,34,38,50,63,64,67,69,72],def:11,defin:[0,6,14,50,71],definit:[63,64],degrad:71,del:15,delet:[15,24,29,34,49,50],demonstr:[1,11],den:1,dens:[24,29],depend:[4,12,14,16,29,72],deprec:4,der:1,describ:[1,11,29,34,50,63,64,65,66,67,69],descript:[3,5,13,14],detail:1,detect:[11,71],detect_languag:[11,50],detoken:[1,34],detokenis:72,detokenize_fn:34,develop:[5,11,14],devic:[3,5,6,11,16,18,24,29,32,34,45,47,49,50,70,71,72,74],device_index:[16,17,24,29,32,34,47,49,50],devlin:71,dict:[6,24,29,34,42,49,50,58,59,60,62,63,64,65,66,67,68,69],dictionari:[6,24,29,34,38,42,49,50,58,59,60,62,63,64,65,66,67,68,69],die:1,dienst:1,dient:1,dies:1,differ:[0,1,5,6,16,67],dimens:[50,63,64,67],direct:4,directli:[8,9,14,74],directori:[0,6,9,11,14,24,29,34,35,37,38,39,40,41,42,43,44,49,50,58,60,62,63,65,67,68,69,72],disabl:[3,14,16,17,27,28,29,33,34,50,51,64,66,67],disable_unk:[29,34],disk:[4,58,60,62,63,65,67,68,69,71],dispatch:[3,14],dist:14,distil:11,distilbertmodel:11,distribut:1,diverg:1,dl:6,dn:16,dnnl:14,doc:16,docker:74,document:[0,9,13,14,71,74],doe:[1,3,5,6,32,74],don:[9,11,48],done:[19,20,21,52],doption:14,dot:[63,64],download:[6,11,44,72],driver:12,dry:11,dtype:32,dure:[8,11,14,29,34,50,71,74],dwith_acceler:14,dwith_cuda:14,dwith_cudnn:14,dwith_dnnl:14,dwith_mkl:14,dwith_openbla:14,dwith_rui:14,dynam:[14,17],e:[3,8,11,14,15,24,29,32,38,58,60,62,63,65,67,68,69,71,72],each:[0,1,3,5,11,16,24,27,29,31,33,34,50,51,58,60,62,63,64,65,67,68,69,74],earli:5,earth:11,easi:0,echo:74,edunov:1,effect:[1,16],effekt:1,effici:[1,3,17,29,34,71,75],effizi:1,effizient:1,effizienten:1,effizienter:1,effizienz:1,eigen:4,ein:1,einen:1,einsatzort:1,either:[15,24,29,42],element:17,eleutherai:11,ello:[6,74],els:11,emb:4,embed:[0,57,63,64,66,67,71],embeddings_merg:[66,67],embeddingsmerg:[55,66,67,70],empti:[1,27,33,51,74],en:[6,10,11,72],en_xx:6,enabl:[3,14,16,17,29,34,44,63,64,66,67,71,74],enable_cpu_dispatch:14,enable_profil:14,encod:[0,5,6,10,11,13,18,25,38,49,50,64,65,66,67,69,70,72,75],encoderforwardoutput:[18,24,70,75],end:[1,5,6,17,29,34,38,72],end_token:[5,29,34],ende_ctranslate2:[6,10,72,74],endian:0,endl:[72,74],enforc:1,eng_latn:11,engin:4,english:[10,11,72],enough:[24,29,34,49,50],ensur:71,entrypoint:[2,5,14,74],environ:[5,12,13,14,16,17,71],eo:[29,34],epoch:72,equal:12,equat:71,erprobung:1,error:[3,74],es:1,especi:4,et:[1,71],etc:[4,16],eval:11,even:[0,3,4,16],evolv:75,ex:[63,64],exampl:[0,1,2,3,5,6,8,11,14,16,24,26,29,32,34,47,49,53,71,72,73],exce:[29,34],except:[19,20,21,52,71],excit:5,execut:[0,3,4,14,17,26,49,50,71,72],executionstat:[18,34,70],exist:[0,4,24,29,34,37,38,39,40,41,42,43,44,49,50,74],exp:6,expand:[29,34],expans:1,expect:[0,1,11,29,58,62,63,65,67,68,69,74,75],experi:11,experiment:[1,3,73],explicitli:[5,14,67],exponenti:[29,34,50],expos:[0,2,5,71,74],extend:[16,37,39,58,59,60,62],extens:38,extern:4,extract:1,f:11,face:[11,44],facebook:[11,49],facebookresearch:49,factor:[29,34,50,57,63,64],fair:6,fairseq:[0,11,13,38,49],fairseqconvert:[36,37,70],fallback:71,fals:[11,17,24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69],fast:[0,3,4,11],faster:[4,11],fastest:[1,71],favor:4,fbaipublicfil:6,feat_lay:68,featur:[4,11,13,49,50,74],feed:[63,64,66,67],ffn:[63,64,66,67],ffn_glu:[63,64,66,67],fi:10,field:0,file:[0,6,11,16,24,29,34,37,38,39,40,41,42,43,44,49,50,58,60,62,63,65,67,68,69,74],filenam:[44,58,60,62,63,65,67,68,69],fill:0,filter:[1,16,50],fim_middl:11,fim_prefix:11,fim_suffix:11,find:[1,11],find_packag:72,fine:11,finetun:11,finish:[29,34,50],first:[1,11,29,34,50,63,64,71,72],fit:48,fix:[6,38],fixed_dictionari:[6,38],flag:[1,3,12,14,44],flan:11,flash:[24,29,34,49,50],flash_attent:[24,29,34,49,50],float16:[11,22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69],float32:[22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69,71],flush:5,folgebeschleunigung:1,follow:[1,3,6,11,14,16,29,34,71,74,75],forc:[1,3,37,38,39,40,41,42,43,44],forecast:11,format:[0,6,8,11,34,37,38,39,40,41,42,43,44,71,72,74],formkomprimierung:1,formula:71,forward:[0,1,17,24,25,29,63,64,66,67],forward_batch:[5,11,24,29],fp16:[3,11,71],fp32:[3,71],fra_latn:11,frame:50,framework:[4,11],free:[24,29,34,49,50],frequent:[1,13],from:[0,1,3,4,5,6,8,9,11,12,16,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,49,50,51,52,53,56,57,58,59,60,61,62,63,64,65,66,67,68,69,71,72,74],from_arrai:[11,32],from_config:[42,63,67],from_pretrain:[5,11,44],frozenattr:59,ft:6,full:[0,5,16,24,29,34,49,50,71],fulli:[15,24,29,34,42,49,50],further:17,futur:[5,24,29,34,49,50],g:[3,8,14,15,24,29,32,38,58,60,62,63,65,67,68,69,71,72],gate:[63,64,66,67],gear:1,gelu:56,gelusigmoid:56,gelutanh:56,gemm:[3,17],gener:[1,3,6,11,13,16,18,24,27,28,34,41,48,50,51,54,70,74,75],generate_batch:[5,6,8,11,29],generate_iter:29,generate_token:[5,29,34],generationopt:[29,75],generationresult:[18,19,29,70,75],generationstepresult:[18,29,34,70],german:[10,11,72],get:[5,11,30,71,72],get_cuda_device_count:[18,70],get_default_config:[58,60,62,63,65,67,68,69],get_log_level:[18,70],get_model_fil:44,get_source_vocabulary_s:[62,67],get_supported_compute_typ:[18,70,71],get_target_vocabulary_s:[62,67],get_vocabulary_s:[58,63,65,68,69],getcurrank:[16,30],getlocalrank:30,getnrank:30,ghcr:[14,74],gil:16,girafatron:11,giraff:11,giraftron:11,git:14,github:[11,13,14,40,44,49,50],give:71,given:[34,50],gleichzeitig:1,global:16,gloriou:11,goal:4,good:11,gpt2:11,gpt2_ct2:11,gpt2token:11,gpt:[5,40,63,64],gpt_neox_ct2:11,gptj_ct2:11,gptneox20b:11,gpu:[3,4,5,16,30,45,71,74],gram:74,graph:4,greater:[0,12,14,29,34],greedi:[29,34,50],group:[63,64],guarante:[0,29,34,75],guid:[2,10,12,72,73],gz:[6,72],h:[6,9,72,74],ha:[1,4,16,29,34,74],half:[63,64,71],hall:72,hallo:72,han:44,handl:72,hard:4,hardwar:13,harm:5,harmless:5,have:[0,3,5,6,9,11,14,16,34,50,71,72],head:[63,64,66,67,69],head_dim:[63,64],heavi:4,hei:11,hello:[6,8,10,11,24,34,72,74],help:[5,14,71,74],helper:[35,37,38,39,40,41,42,43,44],helsinki:[11,72],here:[4,11,29,71,74],hf:11,hidden:67,high:[2,3,5,11,74],higher:1,highest:[1,34],hostfil:16,hous:11,how:[0,5,6,11,66,67,74],howev:[0,4,11,71],http:[6,10,11,14,29,34,40,44,49,50,63,64,66,67,72],hub:[11,44],hug:[11,44],huggingfac:11,human:5,hypothes:[1,6,10,11,29,33,34,50,72,74],hypothesi:[1,28,33],hypothesis_id:28,i:[11,29,71,74],id:[17,24,27,28,29,30,34,49,50,51],idea:71,identifi:[24,29,34,49,50],ignor:34,imag:[14,74],immedi:[1,16],impact:[1,3],implement:[0,3,5,11,14,16,32,49,50,71],impli:[4,71],implicitli:[5,11],improv:[3,11,16,17,44],includ:[0,1,5,6,8,11,13,14,16,29,34,37,38,39,40,41,42,43,44,50,72,74],include_prompt_in_result:[11,17,29],increas:[1,3,17,29,74,75],increment:[0,58,60,62,63,65,67,68,69],independ:1,index:[28,29,32,47,50,64,67],ineffekt:1,infer:1,inferenzbeschleunigung:1,info:[3,53,71],inform:[5,12,16,30,32,71,72],inherit:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,49,50,51,52,56,57,58,59,60,61,62,63,64,65,66,67,68,69],initi:[24,29,34,38,39,40,41,42,43,44,49,50,58,60,62,63,64,65,66,67,68,69],input:[1,5,6,11,16,17,24,29,32,34,49,50,63,64,71,72,74],input_featur:11,input_fil:34,input_id:11,input_text:[11,72],input_token:[11,72],instal:[6,8,9,11,13,16,72],install_requir:14,instanc:[1,15,16,24,29,32,34,42,49,50],instead:[1,3,14,29,58,59,60,62,63,64,65,66,67,68,69,71],instruct:[3,11,12,13,14],int16:[22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69],int32:[22,24,29,32,71],int32_float16:71,int8:[17,22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69],int8_bfloat16:[24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69,71],int8_float16:[11,24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69,71],int8_float32:[24,29,34,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69,71],integr:[3,4,14,73],intel:[3,4,12,14,17,71],intenum:[56,57,61],inter_thread:[16,17,24,29,34,49,50],interact:[5,11],interfac:32,interleav:[63,64],intermedi:[0,11,44],intern:5,intra_thread:[16,17,24,29,34,49,50],introduc:[3,4,11],invari:67,io:[14,74],iostream:[72,74],ipaddress:16,irrelev:11,is_last:28,is_multilingu:50,is_new_word:5,isa:[3,14],issu:[0,4],ist:1,iter:[29,34],its:[5,14,29,58,59,60,62,63,64,65,66,67,68,69,71],j4:[14,72],j:[63,64,71],join:[6,34],joke:5,json:[0,50],just:[1,5],k:[14,29,34,50,74],keep:[1,17,24,29,34,49,50,71],keep_cach:[24,29,34,49,50],kei:[4,63,64],kernel:[14,50],known:[3,5],kv:[29,63,64],kwarg:[29,34,44],l:74,lab:44,lambda:34,lang_code_to_token:11,languag:[5,11,38,50,58,68],languagemodelspec:[55,60,63,65,68,69,70],larg:[4,5,11,17,29,34],larger:[11,17],last:[11,25,28,63],last_hidden_st:25,late:4,latenc:16,later:[15,16],latest:[14,42,74],layer:[11,14,17,24,25,29,34,49,50,58,59,60,62,63,64,65,66,67,68,69,71],layernorm_embed:[63,64,66,67],layerspec:[0,55,60,64,66,70],layout:[58,60,62,63,65,67,68,69],ld_library_path:[14,16],ldconfig:14,leaf:0,learn:[3,4],least:[1,11,12],left:[1,71],len:1,length:[17,24,29,34,50,63,64],length_penalti:[29,34,50],let:1,level:[0,2,3,5,34,46,53,60,74],librari:[3,4,29,34,53,71,72],librosa:11,life:[5,11],light:11,lightweight:4,like:[1,5,11,16,24,29,34,49,50,63,64,71],likelihood:5,limit:[1,16],line:[8,14,34,37,38,39,40,41,42,43,44,74],linear:[61,63,64,66,67,71],link:14,linux:14,list:[1,3,6,11,14,24,29,34,39,44,49,50,58,59,60,62,63,64,65,66,67,68,69,74],littl:71,llama3:61,llama:[5,71],lm:6,load:[0,5,6,10,11,14,15,16,24,29,34,42,44,49,50,75],load_as_float16:44,load_model:[15,24,29,34,44,49,50],load_token:44,local:[29,34],log:[3,5,28,29,31,34,46,50,53,71],log_prob:[6,28,31],logic:[0,4],login:11,logit:[5,11,27,28,29,33],longer:[11,16],look:72,loop:[5,29],low:1,low_cpu_mem_usag:44,lower:[1,63,64],lua:4,luatorch:4,m2m100:11,m2m100_418:11,m2m100_418m:11,m2m_100:6,m2m_100_418m_ct2:6,machin:[0,11,16,17,72],maco:[14,16],mai:[3,16,38,44],mail:11,main:[2,5,11,49,71,72,74],majesti:11,major:0,make:[4,5,8,11,14,16,17,72],manag:[0,5,13,30],mani:[4,17,29,34],manylinux2014:14,map:[24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69,74],march:14,marian:[0,10,11,13,39],marianconvert:[36,37,43,70],mask:11,master:16,match:[6,74],materi:[29,34],matmul:16,matrix:33,max:[63,64,71],max_batch_s:[16,17,29,34],max_bin:3,max_cached_byt:3,max_decoding_length:[1,34],max_initial_timestamp_index:50,max_input_length:[1,29,34],max_length:[5,6,11,24,29,50],max_position_embed:[63,64],max_queued_batch:[16,24,29,34,49,50],maximum:[1,24,29,34,49,50,71],mbart50:6,mbart50_ct2:6,mbart:11,me:11,mean:[5,6,63,64,66,67],median:50,median_filter_width:50,mel:[49,50],memori:[1,3,4,11,13,16,24,29,34,49,50,71],merg:[57,66,67],meta:11,method:[5,11,15,16,17,19,20,21,24,29,30,32,34,37,38,39,40,41,42,43,44,49,50,52,58,59,60,62,63,64,65,66,67,68,69,74],middai:11,might:15,militari:6,millisecond:26,min:71,min_alternative_expansion_prob:[1,29,34],min_bin:3,min_decoding_length:[1,34],min_length:29,minim:[6,29,34],minimum:[1,29,34],mit:44,mixtur:11,mkdir:[14,72],mkl:[3,4,12,14,71],ml50_lang:6,mode:[1,16,24,29,34,49,50,71],model:[1,2,3,4,5,7,8,9,10,11,13,14,15,17,24,25,29,30,34,35,37,38,39,40,41,42,43,44,49,50,51,52,58,60,62,63,64,65,67,68,69,70,72,73,74,75],model_class:44,model_dict:6,model_dir:[10,40,43],model_is_load:[24,29,34,49,50],model_load:72,model_nam:11,model_name_or_path:44,model_path:[6,7,8,16,24,29,34,38,39,41,49,50,71,72],model_spec:[0,59],modellen:1,modellkompress:1,modellkomprimierung:1,modelload:72,modelspec:[55,58,59,62,70],modul:[0,3,14,63],mono:11,mood:5,more:[0,1,3,4,5,12,16,17,29,34,44,50,71,72,74],mosaicml:11,most:[1,9,11,14,16,29,34,71],move:[0,4,15,24,29,34,49,50],mpi:[16,30],mpiinfo:[16,18,70],mpirun:16,mt5:11,mt:[0,11,13,43,72],multi:[11,63,64,66,67],multi_query_attent:[63,64,66,67],multilingu:[6,11,38,50],multilingual_transform:6,multipl:[0,5,14,16,17,29,34,49,50,71],multiprocess:16,multithread:13,must:0,my:11,n:[1,11,34,63,64,74],n_mel:50,name:[0,5,6,11,17,22,23,24,29,34,42,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69],namespac:[37,38,39,40,41,42,43,44],nbgpu1:16,nbgpu2:16,nbgpuexpect:16,ndaniel:11,ndarrai:[58,59,60,62,63,64,65,66,67,68,69],nearli:11,necessari:16,need:[4,11,16,72],neg:1,neox:[63,64],nest:0,network:[63,64,66,67],next:0,ngirafatron:11,ngram:[29,34,50],nlp:[11,72],nn:[0,6],no_default_special_token:38,no_final_norm:[63,64,66,67],no_repeat_ngram_s:[29,34,50],no_speech_prob:51,non:[1,14,50,71],none:[14,22,23,24,28,29,34,37,38,39,40,41,42,43,44,49,50,54,58,59,60,62,63,64,65,66,67,68,69],nonsens:1,norm:[63,64,66,67],normal:[34,63,64,66,67],note:16,notimestamp:11,notimplementederror:[37,38,39,40,41,42,43,44],now:4,np:[6,11,16,32],npz:[7,39],num_active_batch:[24,29,34,49,50],num_decoder_head:69,num_decoder_lay:69,num_encod:24,num_encoder_head:69,num_encoder_lay:69,num_exampl:26,num_fram:50,num_gener:29,num_head:[63,64,66,67,68],num_heads_kv:[63,64],num_hypothes:[1,29,34,50],num_languag:50,num_lay:[63,64,66,67,68],num_queued_batch:[24,29,34,49,50],num_source_embed:[66,67],num_token:26,num_transl:34,num_work:[49,50],number:[0,1,16,17,24,26,29,30,34,45,49,50,63,64,66,67,69,74],numpi:[6,11,24,29,32],nvcc:14,nvidia:[12,14,17,71],o:72,obj:29,object:[0,10,11,15,16,19,20,21,24,29,30,32,34,49,50,52],obsess:11,obtain:6,occur:4,occurr:1,oder:1,off:[3,14],offset:34,older:0,omp_:16,onc:[5,11,14,17,29],one:[0,1,16,29,34,63,71],onednn:[12,14,71],ones:32,onli:[0,2,4,5,8,11,12,16,29,71],onmt_release_model:8,open:[4,5,16,34],openai:[11,40,50],openaigpt2convert:[36,37,70],openbla:14,openmp:[14,16,24,29,34,49,50],openmp_runtim:14,opennmt:[0,4,13,14,41,42,71,72,74],opennmtpyconvert:[36,37,70],opennmttfconvert:[36,37,70],oper:[0,3,16],opt:14,optim:[0,17,29,34,58,59,60,62,63,64,65,66,67,68,69,71],option:[0,1,6,9,11,24,29,34,37,38,39,40,41,42,43,44,50,58,59,60,62,63,64,65,66,67,68,69,71,74,75],opu:[0,11,13,43,72],opus_model:10,opusmtconvert:[36,39,70],order:[3,50,58,59,60,62,63,64,65,66,67,68,69],org:[29,34,50,63,64,66,67],origin:[63,64,71],original_max_position_embed:[63,64],ort:1,os:[3,14],other:[4,9,11,12,16,71,72,75],otherwis:[16,63,64,72],out:[0,1,5,16],out_typ:[5,6,10,72],output:[1,5,6,8,10,11,24,25,26,29,34,37,38,39,40,41,42,43,44,49,50,58,60,62,63,65,67,68,69,72,74],output_dir:[6,7,8,9,10,11,37,38,39,40,41,42,43,44,58,60,62,63,65,67,68,69,71,72],output_id:5,output_path:34,output_text:[11,72],output_token:[11,72],outsid:[9,72],over:[17,29,34],overcom:4,overrid:[3,37,38,39,40,41,42,43,44],overview:4,own:[0,3,5,34,72],pack:[3,17],packag:[14,72],pad:[29,34,50],page:[1,11],pair:[16,50],paper:[64,66,67],parallel:[13,24,29,34,49,50,63,64],parallel_residu:[63,64],paramet:[1,3,6,11,16,24,29,32,34,37,38,39,40,41,42,43,44,47,49,50,53,58,59,60,62,63,64,65,66,67,68,69,71],pars:[37,38,39,40,41,42,43,44],parser:[37,38,39,40,41,42,43,44],part:6,particip:5,particular:12,pass:[9,11,17,32,74],path:[6,7,8,11,14,16,24,29,34,35,37,38,39,40,41,42,43,44,49,50,58,60,62,63,65,67,68,69,72],patienc:[29,34,50],pegasu:11,penal:[29,34,50],penalti:[29,34,50],per:[24,29,34,49,50,71],perform:[3,12,13,16,29,71],period:[63,64],perplex:[5,6],pg:11,physic:17,pile:11,pip:[6,8,9,11,72],place:[1,24,29,34,49,50],plan:[11,14],platform:[12,71],poetri:5,polar:11,pool:[3,11,25,65],pooler_output:[11,25],pooling_activ:65,pooling_lay:65,popul:0,posit:[29,34,63,64,66,67],possibl:[0,3,4,16,17,24,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69,71],post:[63,64],pouta:10,practic:1,pre:[11,44,63,64,66,67],pre_norm:[63,64,66,67],pre_post_layer_norm:[63,64],prebuilt:[12,71],precis:[3,44,71],predict:[17,29,34,50],predicted_class_id:11,prefer:17,prefetch:[29,34],prefix:[1,6,34,58,59,60,62,63,64,65,66,67,68,69],prefix_bias_beta:[1,34],prepend:[58,59,60,62,63,64,65,66,67,68,69],pretrain:[6,10,11,44,72],prevent:[29,34,50],previous:[0,3,29,34,50],print:[1,5,6,10,11,16,32,34,72,74],print_hello_world:11,prob:[5,29,34,50],probabilti:29,probabl:[1,4,11,28,29,31,34,50,51],process:[16,17,24,26,29,30,34,49,50],processor:[11,12],produc:0,product:[0,4,14,63,64],profil:14,programm:1,progress:5,project:[1,3,11,12,13,63,72,75],project_in_out:[63,64],projekt:1,prompt:[11,17,29,50],prompt_token:5,properli:72,properti:[22,23,24,25,26,27,28,29,31,32,33,34,49,50,51,58,60,62,63,64,65,67,68,69],propos:11,provid:[5,10,14,15,30,38,74],pt:[6,8,38,41,71,72],pubkei:16,publish:[6,11,14,49,50],pull:14,purpos:15,pwd:[72,74],py:[0,4,13,14,41,71,72],pybind11:14,pybind11_builtin:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,49,50,51,52],pybind11_object:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,49,50,51,52],pyonmt:72,pypi:14,python3:16,python:[0,3,13,16,46,53,71,75],pytorch:[0,6,8,11,24,29,32,38,41],quant_bit:[63,64],quant_group_s:[63,64],quant_typ:[63,64],quantiz:[11,13,14,17,37,38,39,40,41,42,43,44,58,59,60,62,63,64,65,66,67,68,69],queri:[63,64,66,67,71],question:13,queue:[16,24,29,34,49,50],quickli:[24,29,34,49,50],quickstart:[8,13,74],r:14,rais:[19,20,21,32,37,38,39,40,41,42,43,44,50,52,58,59,60,62,63,64,65,66,67,68,69,74],random:[29,34,50,54],randomli:[1,29,34,50],rang:[11,71],rank:30,rare:0,re:16,reach:11,read:[0,16,34],read_batch_s:34,readi:[24,29,34,49,50,72],readm:71,reason:3,recognit:[11,13,14,49,50],recommend:[11,17],recurs:[14,58,59,60,62,63,64,65,66,67,68,69],redefin:4,reduc:[4,11,71,74],reduct:17,refer:[13,71],refus:5,regist:[58,60,62,63,65,67,68,69],register_fil:[58,60,62,63,65,67,68,69],register_source_vocabulari:[62,67],register_target_vocabulari:[62,67],register_vocabulari:[58,63,65,68,69],register_vocabulary_map:[62,67],registri:14,rel:[63,64,66,67],relative_attention_bia:[64,66,67],relative_posit:[64,66],releas:[8,15,16],reli:4,relianc:4,relu:[56,63,64,66,67],remain:29,remov:11,renam:[58,60,62,63,65,67,68,69],repetit:[29,34,50],repetition_penalti:[29,34,50],replac:[1,34],replace_unknown:34,replica:5,repositori:[11,13,14],repres:0,represent:[0,63,64,66,67],request:[11,24,29,34,49,50],requir:[4,5,6,7,8,11,12,14,24,29,34,38,58,59,60,62,63,64,65,66,67,68,69],resampl:11,rescal:[11,44],residu:[63,64],resolv:42,resourc:15,respect:14,respons:11,rest:[1,4],restrict:[0,1],result:[1,6,10,11,16,19,20,21,27,28,29,31,33,34,50,51,52,71,72,74],resum:[24,29,34,49,50],retain:[63,64],retriev:16,return_altern:[1,29,34],return_attent:[33,34],return_end_token:[29,34],return_log_prob:[28,29,34],return_logits_vocab:[27,29,33,34,50],return_no_speech_prob:[50,51],return_scor:[17,27,29,33,34,50,51],return_tensor:11,reus:[3,5,29],revis:[0,11,44,58,60,62,63,65,67,68,69],richtet:1,risk:11,rm:[14,74],rms_norm:[63,64,66,67],ro_ro:6,roberta:11,root:[0,63,64,66,67],rope:[61,63,64],rotari:[63,64],rotary_bas:[63,64],rotary_dim:[63,64],rotary_interleav:[63,64],rotary_scaling_factor:[63,64],rotary_scaling_typ:[63,64],rotaryscalingtyp:[55,63,64,70],round:71,rui:[14,71],run:[2,5,6,11,14,16,24,29,30,34,49,50,71,72,74],rund:1,runtim:[3,14,24,29,34,49,50,71],runtimeerror:[37,38,39,40,41,42,43,44,50],s3:72,s:[1,3,5,6,8,11,29,34,72],sacremos:6,sai:[1,6],same:[0,3,5,8,16,24,29,32,58,59,60,62,63,64,65,66,67,68,69,71],sampl:[6,29,34,50],sampling_r:11,sampling_temperatur:[1,5,11,29,34,50],sampling_topk:[1,5,6,11,29,34,50],sampling_topp:[29,34],santacod:11,save:[0,11,16,34,37,38,39,40,41,42,43,44,58,60,62,63,65,67,68,69,71],scale:[11,44,61,63,64,71],scale_alibi:[63,64],scenario:4,schedul:11,scheme:[37,38,39,40,41,42,43,44,58,59,60,62,63,64,65,66,67,68,69],schlussfolgerung:1,scientist:11,scope:0,score:[6,17,27,29,31,33,34,50,51,74],score_batch:[5,6,29,34],score_fil:34,score_iter:[29,34],score_token_0:34,score_token_1:34,scoringopt:75,scoringresult:[18,20,29,34,70,75],scratch:11,script:[0,8,16],search:[29,34,50],second:11,section:[0,1,72],see:[0,1,2,3,4,6,8,9,11,12,13,14,16,44,71,72,73,74],seed:54,seem:35,select:[3,9,11,12,14,64,67,71],self:[24,29,34,49,50,64,66,67],semant:75,sens:0,sentenc:[1,6,17,34,72],sentencepiec:[5,6,10,72],sentencepieceprocessor:[5,6,10,72],separ:[3,5,10,14,16,71,74],sequenc:[1,2,5,6,24,27,29,34,50,51,62,63,64],sequences_id:[11,27,51],sequencetosequencemodelspec:[55,60,67,70],serv:[1,4],server:0,servic:1,servieren:1,session:[5,11],set:[1,3,6,12,14,16,17,24,29,34,37,38,39,40,41,42,43,44,47,49,50,53,54,58,59,60,62,63,64,65,66,67,68,69,71,74],set_log_level:[3,18,70,71],set_random_se:[18,70],setup:[14,16],sever:[4,5],shape:[24,29,32,49,50,58,59,60,62,63,64,65,66,67,68,69],share:[4,16,32,63,64,71],shared_layer_norm:[63,64],shock:11,should:[0,3,5,6,9,11,14,16,29,42,59,72,74],show:[5,6],shown:11,shutoff:11,sich:1,similar:[0,4,6,11,71],simpl:0,simpli:[1,8,9,15],simplic:6,simplifi:29,sinc:[10,11],singl:[3,6,14,17,28],size:[4,5,16,17,29,34,50,58,62,63,64,65,67,68,69,71],skip:17,skip_special_token:11,sky:6,slave:16,slice:[63,64],sliding_window:[63,64],slot:[16,24,29,34,49,50],slower:16,small:[1,11],smallest:11,smooth:[11,44],smoothquant:[11,44],so:[11,12,16,17,29,32,34,71,72],softmax:[16,17],solut:6,some:[3,4,6,11,14,15,16,17,26,29,34,44,58,59,60,62,63,64,65,66,67,68,69,74],sort:[17,29,34],sourc:[5,6,10,11,12,34,38,62,66,67],source_lang:38,source_path:34,source_tokenize_fn:34,source_vocabulari:0,sp:[5,6,10,72],space:[0,5],spec:[56,57,58,59,60,61,62,63,64,65,66,67,68,69,70],special:[6,29,38,74],specif:[3,4,11,16,32,58,59,60,62,63,64,65,66,67,68,69,71],specifi:[66,67],spectogram:[49,50],speech:[11,13,14,49,50,51],speed:[1,3,74],split:[16,17,29,34],spm:[5,6,10,72],squar:[63,64,66,67],sr:11,src:7,src_1:74,src_2:74,src_lang:11,src_n:74,sse:12,ssh:16,stabilityai:5,stablelm:5,standard:1,standardisierten:1,star:11,starcod:11,starcoder_ct2:11,start:[1,5,6,8,11,29,38,50,72],start_sequ:50,start_token:[11,29],startoftranscript:11,startswith:5,state:[5,11,29],static_prompt:[5,17,29],statist:[26,34],std:[72,74],step:[0,4,14,28],step_result:5,still:[0,5,11],stop:[5,29,34],storag:32,storageview:[11,18,24,29,49,50,70],store:71,stori:5,str:[5,6,10,24,29,34,35,37,38,39,40,41,42,43,44,47,49,50,58,59,60,62,63,64,65,66,67,68,69,72],strategi:[1,57],stream:[16,17,29,34],string:[1,24,28,29,50,72,74],strip:34,strong:4,stronger:1,structur:[26,29,34],style:11,su:[61,63,64],sub:16,sublay:64,submiss:17,submit:16,submodul:14,successfulli:72,sudo:14,suffix:11,suggest:71,summar:11,supervis:11,support:[3,4,6,7,8,9,11,13,16,17,47,50,74,75],suppress:50,suppress_blank:50,suppress_sequ:[29,34],suppress_token:50,sure:[8,11,16,72],swish:56,symbol:[1,50,75],symmetr:71,syria:6,system:[0,5,14,16,17,29,71],system_prompt:5,system_prompt_token:5,t5:[64,66,67],t5v1:11,t:[9,11,48],tab:74,tabl:71,take:[1,11],talk:11,tanh:[56,65],tar:[6,72],target:[1,6,11,14,34,38,62,67,74],target_detokenize_fn:34,target_lang:38,target_link_librari:72,target_path:34,target_prefix:[1,6,11,34],target_tokenize_fn:34,target_vocabulari:0,task:[6,11,17],team:6,techniqu:[1,71],temperatur:[29,34,50],templat:[0,4],temporarili:15,tensor:[17,24,29,32,34,49,50],tensor_parallel:[16,24,29,34,49,50],tensorflow:0,test:14,text:[1,6,11,13,24,29,34,50,72,73],text_token:50,textattack:11,tf:[0,13,42],tgt:7,tgt_1:74,tgt_2:74,tgt_k:74,tgt_lang:11,than:[0,5,6,17,29,34],theblok:71,thei:[0,3,5,6,10,11,12,29,34],them:[11,14,44],thi:[0,1,3,5,6,10,11,12,14,15,16,19,20,21,24,28,29,34,37,38,39,40,41,42,43,44,49,50,52,53,58,59,60,62,63,64,65,66,67,68,69,71,72,74],thousand:11,thread:[4,14,16,17,24,29,34,49,50],thread_pool:16,through:11,throughput:16,tii:11,tiiuae:11,time:[1,5,16,26,58,60,62,63,65,67,68,69],timestamp:[11,50],timestep:1,tini:11,tip:13,to_cpu:[24,29,34,49,50],to_devic:32,todai:5,token:[1,6,8,17,24,26,27,28,29,31,34,38,50,51,58,62,63,65,67,68,69,71,72,74],token_id:[5,28],token_type_id:24,tokenis:72,tokenize_fn:34,tokenizer_class:44,tomorrow:11,too:16,toolkit:[4,14],top:[0,29,34,50,60],torch:[11,32,44],torch_dtyp:44,total:[17,26],total_time_in_m:26,toward:[1,34],trace:3,track:0,tradeoff:3,train:[0,4,6,7,8,9,10,11,38,39,43,72],transcrib:11,transcript:11,transform:[0,2,4,5,6,7,8,9,10,13,29,44,63,64,65,66,67,71,72,73],transformer_align:6,transformer_lm:6,transformerdecodermodelspec:[55,58,70],transformerdecoderspec:[55,59,63,67,70],transformerencodermodelspec:[55,58,70],transformerencoderspec:[55,59,65,67,70],transformersconvert:[36,37,70],transformerspec:[55,62,70],translat:[1,5,6,10,13,14,15,16,18,33,48,70,71,72,75],translate_batch:[1,6,10,11,16,34,72,74],translate_fil:34,translate_iter:34,translationopt:[34,75],translationresult:[18,21,34,70,74,75],tree:49,tri:71,truncat:[1,29,34],trust_remote_cod:[11,44],tune:[5,11],tupl:[50,67],txt:[6,14,34,72],type:[8,24,29,32,34,47,49,50,56,61,63,64,72],typic:71,ubuntu20:[14,74],um:1,un:6,uncas:11,uncondit:11,unconstrain:[29,34],und:1,unifi:0,union:[24,29,34,42,49,50,67,74],unit:[63,64,66,67],unknown:[29,34],unlik:1,unlimit:[5,24,29,34,49,50],unload:[5,15,24,29,34,49,50],unload_model:[15,24,29,34,49,50],unscal:29,unspecifi:1,unsupervis:11,unsupport:32,until:[16,19,20,21,24,29,34,49,50,52],unzip:10,up:[11,29],updat:[0,14],us:[0,1,2,3,5,6,8,10,11,12,14,15,16,17,24,29,30,32,34,42,44,49,50,58,60,62,63,64,65,66,67,68,69,71,72,74],usag:[1,3,4,11,13,71,74],use_vmap:[34,74],user:[5,38],user_dir:38,usual:[3,5,14],v:74,valid:[12,58,59,60,62,63,64,65,66,67,68,69],valu:[0,1,3,4,14,16,22,23,24,28,29,34,37,38,39,40,41,42,43,44,49,50,58,59,60,62,63,64,65,66,67,68,69,71],valueerror:[32,58,59,60,62,63,64,65,66,67,68,69],variabl:[0,12,13,14,16,17,29,58,59,60,62,63,64,65,66,67,68,69,71],variant:11,variou:72,vector:[34,72,74],vendor:3,verbos:3,veri:[1,16],verifi:[58,59,60,62,63,64,65,66,67,68,69],version:[0,5,11,12,13,14,72],via:[4,74],view:32,virtual_env:16,virtual_env_prompt:16,visibl:45,visual:14,vmap:[37,38,39,40,41,42,43,44,74],vocab:[7,28],vocab_map:74,vocab_path:[7,39],vocabulari:[0,6,7,17,34,37,38,39,40,41,42,43,44,58,62,63,65,67,68,69,71],volum:17,von:1,w:71,wa:[11,19,20,21,27,28,33,51,52,71],wai:[4,29],wait:[24,29,34,49,50],want:[1,4,9,15,16],warn:3,wav2vec2:[48,70],wav2vec2spec:[55,58,70],wav2vec:49,wav:11,we:[1,16,71],weight:[0,4,11,16,34,37,38,39,40,41,42,43,44,58,59,60,62,63,64,65,66,67,68,69,71],welcom:13,well:74,welt:72,were:[0,4,11],wget:[6,10,72],what:[5,72,74],whatev:71,wheel:[14,16],when:[0,1,3,5,11,14,16,17,19,20,21,24,29,34,44,49,50,52,63,64,66,67,71],whenev:17,where:[0,4,6,12,24,29,32,34,37,38,39,40,41,42,43,44,49,50,58,60,62,63,65,67,68,69,74],wherea:12,whether:[3,24,28,29,34,42,49,50],which:[0,2,3,4,5,6,11,14,16,17,29,34,48,50,71,72,74],whisper:[48,51,69,70,73],whisperalignmentresult:50,whispergenerationresult:[48,50,52,70],whispergenerationresultasync:[48,50,70],whisperprocessor:11,whisperspec:[55,58,70],whl:14,whose:[17,29,34],why:1,width:50,wildfir:11,wind:11,window:[11,14],with_acceler:14,with_cuda:14,with_cudnn:14,with_dnnl:14,with_encoder_attent:64,with_mkl:14,with_openbla:14,with_relative_posit:[63,67],with_rui:14,with_scor:34,with_tokens_scor:34,without:[0,4,32,71],wmt19:5,wmt19_en_ct2:6,wmt:72,wngt:17,wonder:11,word:5,work:[3,71],worker:[16,29,34,49,50],workload:16,workshop:11,world:[6,10,11,24,34,72,74],worst:[11,50],would:[0,71],wq:71,wrapper:[19,20,21,52],write:[0,5,72],wu:71,x86:[12,14,71],x:[6,14,16,32],xf:[6,72],xlm:11,y:32,yaml:[9,42],yelp:11,yield:[29,34],yml:[7,9,39],you:[0,1,3,5,8,9,11,12,14,15,16,17,71,72],your:[0,4,5,14,34,71,72],z:32,zero:71,zip:10,zu:1,zur:1,zwar:1},titles:["Model conversion","Decoding features","Text encoding","Environment variables","Frequently asked questions","Text generation","Fairseq","Marian","OpenNMT-py","OpenNMT-tf","OPUS-MT","Transformers","Hardware support","Index","Installation","Memory management","Multithreading and parallelism","Performance tips","ctranslate2","AsyncGenerationResult","AsyncScoringResult","AsyncTranslationResult","DataType","Device","Encoder","EncoderForwardOutput","ExecutionStats","GenerationResult","GenerationStepResult","Generator","MpiInfo","ScoringResult","StorageView","TranslationResult","Translator","contains_model","ctranslate2.converters","Converter","FairseqConverter","MarianConverter","OpenAIGPT2Converter","OpenNMTPyConverter","OpenNMTTFConverter","OpusMTConverter","TransformersConverter","get_cuda_device_count","get_log_level","get_supported_compute_types","ctranslate2.models","Wav2Vec2","Whisper","WhisperGenerationResult","WhisperGenerationResultAsync","set_log_level","set_random_seed","ctranslate2.specs","Activation","EmbeddingsMerge","LanguageModelSpec","LayerSpec","ModelSpec","RotaryScalingType","SequenceToSequenceModelSpec","TransformerDecoderModelSpec","TransformerDecoderSpec","TransformerEncoderModelSpec","TransformerEncoderSpec","TransformerSpec","Wav2Vec2Spec","WhisperSpec","Python","Quantization","Quickstart","Speech recognition","Text translation","Versioning"],titleterms:{"100":[6,11],"16":71,"2":11,"4":71,"50":6,"8":71,"do":4,"float":71,"new":0,activ:56,add:0,altern:1,api:13,ar:4,ask:4,asyncgenerationresult:19,asynchron:16,asyncscoringresult:20,asynctranslationresult:21,autocomplet:1,awq:71,backward:0,bart:11,beam:[1,6],bert:11,bfloat16:71,bias:1,bit:71,bloom:11,brain:71,build:14,c:[14,72,74],cach:5,client:74,code:14,compat:0,compil:14,constraint:1,contains_model:35,convers:[0,71],convert:[0,36,37],cpu:[12,16,17],ct2_cuda_alloc:3,ct2_cuda_allow_bf16:3,ct2_cuda_allow_fp16:3,ct2_cuda_caching_allocator_config:3,ct2_cuda_true_fp16_gemm:3,ct2_force_cpu_isa:3,ct2_use_experimental_packed_gemm:3,ct2_use_mkl:3,ct2_verbos:3,ctranslat:4,ctranslate2:[18,36,48,55],data:16,datatyp:22,decod:1,develop:13,devic:23,distilbert:11,docker:14,doe:4,download:14,dynam:74,embeddingsmerg:57,encod:[2,24],encoderforwardoutput:25,english:6,environ:3,equival:6,exampl:[10,74],execut:16,executionstat:26,factor:74,fairseq:6,fairseqconvert:38,falcon:11,featur:1,float16:71,framework:[0,13],frequent:4,from:[14,15],futur:4,gener:[5,8,17,29],generationresult:27,generationstepresult:28,german:6,get:13,get_cuda_device_count:45,get_log_level:46,get_supported_compute_typ:47,gpt:11,gptbigcod:11,gpu:[12,14,17],greedi:1,guid:13,hardwar:12,how:4,i:4,implement:4,implicit:71,index:13,inform:13,instal:14,instead:4,int16:71,int8:71,integ:71,intra:16,j:11,known:4,languag:6,languagemodelspec:58,layerspec:59,length:1,librari:14,limit:4,llama:11,load:71,m2m:[6,11],manag:15,marian:7,marianconvert:39,marianmt:11,mbart:6,memori:15,model:[0,6,16,48,71],modelspec:60,mpiinfo:30,mpt:11,mt:10,multithread:16,neox:11,nllb:11,op:16,openaigpt2convert:40,opennmt:[8,9],opennmtpyconvert:41,opennmttfconvert:42,opt:11,option:14,opu:10,opusmtconvert:43,origin:4,other:13,parallel:16,perform:17,pip:14,plan:4,point:71,portabl:0,posit:1,precis:0,project:4,prompt:5,provid:4,py:8,python:[14,15,70,72,74],pytorch:4,quantiz:[0,71],question:4,quickstart:72,random:1,recognit:73,reduc:0,reduct:74,relat:4,rotaryscalingtyp:61,sampl:1,scoringresult:31,search:[1,6],sequencetosequencemodelspec:62,serial:0,server:4,set_log_level:53,set_random_se:54,should:4,sourc:[14,74],spec:55,special:[5,11],specif:0,speech:73,start:13,storageview:32,stream:5,structur:0,support:[0,12,14,71],t5:11,task:13,tensor:16,tensorflow:4,text:[2,5,8,74],tf:9,thi:4,tip:17,token:[5,11],transform:11,transformer_lm:8,transformerdecodermodelspec:63,transformerdecoderspec:64,transformerencodermodelspec:65,transformerencoderspec:66,transformersconvert:44,transformerspec:67,translat:[4,11,17,34,74],translationresult:33,type:71,us:4,variabl:3,version:75,vocabulari:74,wav2vec2:49,wav2vec2spec:68,what:4,when:4,whisper:[11,50],whispergenerationresult:51,whispergenerationresultasync:52,whisperspec:69,why:4,wmt16:6,wmt19:6,wrapper:14,you:4}}) \ No newline at end of file +Search.setIndex({docnames:["conversion","decoding","encoding","environment_variables","faq","generation","guides/fairseq","guides/marian","guides/opennmt_py","guides/opennmt_tf","guides/opus_mt","guides/transformers","hardware_support","index","installation","memory","parallel","performance","python/ctranslate2","python/ctranslate2.AsyncGenerationResult","python/ctranslate2.AsyncScoringResult","python/ctranslate2.AsyncTranslationResult","python/ctranslate2.DataType","python/ctranslate2.Device","python/ctranslate2.Encoder","python/ctranslate2.EncoderForwardOutput","python/ctranslate2.ExecutionStats","python/ctranslate2.GenerationResult","python/ctranslate2.GenerationStepResult","python/ctranslate2.Generator","python/ctranslate2.MpiInfo","python/ctranslate2.ScoringResult","python/ctranslate2.StorageView","python/ctranslate2.TranslationResult","python/ctranslate2.Translator","python/ctranslate2.contains_model","python/ctranslate2.converters","python/ctranslate2.converters.Converter","python/ctranslate2.converters.FairseqConverter","python/ctranslate2.converters.MarianConverter","python/ctranslate2.converters.OpenAIGPT2Converter","python/ctranslate2.converters.OpenNMTPyConverter","python/ctranslate2.converters.OpenNMTTFConverter","python/ctranslate2.converters.OpusMTConverter","python/ctranslate2.converters.TransformersConverter","python/ctranslate2.get_cuda_device_count","python/ctranslate2.get_log_level","python/ctranslate2.get_supported_compute_types","python/ctranslate2.models","python/ctranslate2.models.Wav2Vec2","python/ctranslate2.models.Wav2Vec2Bert","python/ctranslate2.models.Whisper","python/ctranslate2.models.WhisperGenerationResult","python/ctranslate2.models.WhisperGenerationResultAsync","python/ctranslate2.set_log_level","python/ctranslate2.set_random_seed","python/ctranslate2.specs","python/ctranslate2.specs.Activation","python/ctranslate2.specs.EmbeddingsMerge","python/ctranslate2.specs.LanguageModelSpec","python/ctranslate2.specs.LayerSpec","python/ctranslate2.specs.ModelSpec","python/ctranslate2.specs.RotaryScalingType","python/ctranslate2.specs.SequenceToSequenceModelSpec","python/ctranslate2.specs.TransformerDecoderModelSpec","python/ctranslate2.specs.TransformerDecoderSpec","python/ctranslate2.specs.TransformerEncoderModelSpec","python/ctranslate2.specs.TransformerEncoderSpec","python/ctranslate2.specs.TransformerSpec","python/ctranslate2.specs.Wav2Vec2BertSpec","python/ctranslate2.specs.Wav2Vec2Spec","python/ctranslate2.specs.WhisperSpec","python/overview","quantization","quickstart","speech_recognition","translation","versioning"],envversion:{"sphinx.domains.c":2,"sphinx.domains.changeset":1,"sphinx.domains.citation":1,"sphinx.domains.cpp":5,"sphinx.domains.index":1,"sphinx.domains.javascript":2,"sphinx.domains.math":2,"sphinx.domains.python":3,"sphinx.domains.rst":2,"sphinx.domains.std":2,sphinx:56},filenames:["conversion.md","decoding.md","encoding.md","environment_variables.md","faq.md","generation.md","guides/fairseq.md","guides/marian.md","guides/opennmt_py.md","guides/opennmt_tf.md","guides/opus_mt.md","guides/transformers.md","hardware_support.md","index.rst","installation.md","memory.md","parallel.md","performance.md","python/ctranslate2.rst","python/ctranslate2.AsyncGenerationResult.rst","python/ctranslate2.AsyncScoringResult.rst","python/ctranslate2.AsyncTranslationResult.rst","python/ctranslate2.DataType.rst","python/ctranslate2.Device.rst","python/ctranslate2.Encoder.rst","python/ctranslate2.EncoderForwardOutput.rst","python/ctranslate2.ExecutionStats.rst","python/ctranslate2.GenerationResult.rst","python/ctranslate2.GenerationStepResult.rst","python/ctranslate2.Generator.rst","python/ctranslate2.MpiInfo.rst","python/ctranslate2.ScoringResult.rst","python/ctranslate2.StorageView.rst","python/ctranslate2.TranslationResult.rst","python/ctranslate2.Translator.rst","python/ctranslate2.contains_model.rst","python/ctranslate2.converters.rst","python/ctranslate2.converters.Converter.rst","python/ctranslate2.converters.FairseqConverter.rst","python/ctranslate2.converters.MarianConverter.rst","python/ctranslate2.converters.OpenAIGPT2Converter.rst","python/ctranslate2.converters.OpenNMTPyConverter.rst","python/ctranslate2.converters.OpenNMTTFConverter.rst","python/ctranslate2.converters.OpusMTConverter.rst","python/ctranslate2.converters.TransformersConverter.rst","python/ctranslate2.get_cuda_device_count.rst","python/ctranslate2.get_log_level.rst","python/ctranslate2.get_supported_compute_types.rst","python/ctranslate2.models.rst","python/ctranslate2.models.Wav2Vec2.rst","python/ctranslate2.models.Wav2Vec2Bert.rst","python/ctranslate2.models.Whisper.rst","python/ctranslate2.models.WhisperGenerationResult.rst","python/ctranslate2.models.WhisperGenerationResultAsync.rst","python/ctranslate2.set_log_level.rst","python/ctranslate2.set_random_seed.rst","python/ctranslate2.specs.rst","python/ctranslate2.specs.Activation.rst","python/ctranslate2.specs.EmbeddingsMerge.rst","python/ctranslate2.specs.LanguageModelSpec.rst","python/ctranslate2.specs.LayerSpec.rst","python/ctranslate2.specs.ModelSpec.rst","python/ctranslate2.specs.RotaryScalingType.rst","python/ctranslate2.specs.SequenceToSequenceModelSpec.rst","python/ctranslate2.specs.TransformerDecoderModelSpec.rst","python/ctranslate2.specs.TransformerDecoderSpec.rst","python/ctranslate2.specs.TransformerEncoderModelSpec.rst","python/ctranslate2.specs.TransformerEncoderSpec.rst","python/ctranslate2.specs.TransformerSpec.rst","python/ctranslate2.specs.Wav2Vec2BertSpec.rst","python/ctranslate2.specs.Wav2Vec2Spec.rst","python/ctranslate2.specs.WhisperSpec.rst","python/overview.rst","quantization.md","quickstart.md","speech_recognition.md","translation.md","versioning.md"],objects:{"":[[18,0,0,"-","ctranslate2"]],"ctranslate2.AsyncGenerationResult":[[19,2,1,"","done"],[19,2,1,"","result"]],"ctranslate2.AsyncScoringResult":[[20,2,1,"","done"],[20,2,1,"","result"]],"ctranslate2.AsyncTranslationResult":[[21,2,1,"","done"],[21,2,1,"","result"]],"ctranslate2.DataType":[[22,2,1,"","__init__"],[22,3,1,"","bfloat16"],[22,3,1,"","float16"],[22,3,1,"","float32"],[22,3,1,"","int16"],[22,3,1,"","int32"],[22,3,1,"","int8"],[22,4,1,"","name"],[22,4,1,"","value"]],"ctranslate2.Device":[[23,2,1,"","__init__"],[23,3,1,"","cpu"],[23,3,1,"","cuda"],[23,4,1,"","name"],[23,4,1,"","value"]],"ctranslate2.Encoder":[[24,2,1,"","__init__"],[24,4,1,"","compute_type"],[24,4,1,"","device"],[24,4,1,"","device_index"],[24,2,1,"","forward_batch"],[24,2,1,"","load_model"],[24,4,1,"","model_is_loaded"],[24,4,1,"","num_active_batches"],[24,4,1,"","num_encoders"],[24,4,1,"","num_queued_batches"],[24,4,1,"","tensor_parallel"],[24,2,1,"","unload_model"]],"ctranslate2.EncoderForwardOutput":[[25,4,1,"","last_hidden_state"],[25,4,1,"","pooler_output"]],"ctranslate2.ExecutionStats":[[26,4,1,"","num_examples"],[26,4,1,"","num_tokens"],[26,4,1,"","total_time_in_ms"]],"ctranslate2.GenerationResult":[[27,4,1,"","logits"],[27,4,1,"","scores"],[27,4,1,"","sequences"],[27,4,1,"","sequences_ids"]],"ctranslate2.GenerationStepResult":[[28,4,1,"","batch_id"],[28,4,1,"","hypothesis_id"],[28,4,1,"","is_last"],[28,4,1,"","log_prob"],[28,4,1,"","logits"],[28,4,1,"","step"],[28,4,1,"","token"],[28,4,1,"","token_id"]],"ctranslate2.Generator":[[29,2,1,"","__init__"],[29,2,1,"","async_generate_tokens"],[29,4,1,"","compute_type"],[29,4,1,"","device"],[29,4,1,"","device_index"],[29,2,1,"","forward_batch"],[29,2,1,"","generate_batch"],[29,2,1,"","generate_iterable"],[29,2,1,"","generate_tokens"],[29,2,1,"","load_model"],[29,4,1,"","model_is_loaded"],[29,4,1,"","num_active_batches"],[29,4,1,"","num_generators"],[29,4,1,"","num_queued_batches"],[29,2,1,"","score_batch"],[29,2,1,"","score_iterable"],[29,4,1,"","tensor_parallel"],[29,2,1,"","unload_model"]],"ctranslate2.MpiInfo":[[30,2,1,"","getCurRank"],[30,2,1,"","getLocalRank"],[30,2,1,"","getNRanks"]],"ctranslate2.ScoringResult":[[31,4,1,"","log_probs"],[31,4,1,"","tokens"]],"ctranslate2.StorageView":[[32,4,1,"","device"],[32,4,1,"","device_index"],[32,4,1,"","dtype"],[32,2,1,"","from_array"],[32,4,1,"","shape"],[32,2,1,"","to"],[32,2,1,"","to_device"]],"ctranslate2.TranslationResult":[[33,4,1,"","attention"],[33,4,1,"","hypotheses"],[33,4,1,"","logits"],[33,4,1,"","scores"]],"ctranslate2.Translator":[[34,2,1,"","__init__"],[34,4,1,"","compute_type"],[34,4,1,"","device"],[34,4,1,"","device_index"],[34,2,1,"","generate_tokens"],[34,2,1,"","load_model"],[34,4,1,"","model_is_loaded"],[34,4,1,"","num_active_batches"],[34,4,1,"","num_queued_batches"],[34,4,1,"","num_translators"],[34,2,1,"","score_batch"],[34,2,1,"","score_file"],[34,2,1,"","score_iterable"],[34,4,1,"","tensor_parallel"],[34,2,1,"","translate_batch"],[34,2,1,"","translate_file"],[34,2,1,"","translate_iterable"],[34,2,1,"","unload_model"]],"ctranslate2.converters":[[37,1,1,"","Converter"],[38,1,1,"","FairseqConverter"],[39,1,1,"","MarianConverter"],[40,1,1,"","OpenAIGPT2Converter"],[41,1,1,"","OpenNMTPyConverter"],[42,1,1,"","OpenNMTTFConverter"],[43,1,1,"","OpusMTConverter"],[44,1,1,"","TransformersConverter"]],"ctranslate2.converters.Converter":[[37,2,1,"","convert"],[37,2,1,"","convert_from_args"],[37,2,1,"","declare_arguments"]],"ctranslate2.converters.FairseqConverter":[[38,2,1,"","__init__"],[38,2,1,"","convert"],[38,2,1,"","convert_from_args"],[38,2,1,"","declare_arguments"]],"ctranslate2.converters.MarianConverter":[[39,2,1,"","__init__"],[39,2,1,"","convert"],[39,2,1,"","convert_from_args"],[39,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenAIGPT2Converter":[[40,2,1,"","__init__"],[40,2,1,"","convert"],[40,2,1,"","convert_from_args"],[40,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenNMTPyConverter":[[41,2,1,"","__init__"],[41,2,1,"","convert"],[41,2,1,"","convert_from_args"],[41,2,1,"","declare_arguments"]],"ctranslate2.converters.OpenNMTTFConverter":[[42,2,1,"","__init__"],[42,2,1,"","convert"],[42,2,1,"","convert_from_args"],[42,2,1,"","declare_arguments"],[42,2,1,"","from_config"]],"ctranslate2.converters.OpusMTConverter":[[43,2,1,"","__init__"],[43,2,1,"","convert"],[43,2,1,"","convert_from_args"],[43,2,1,"","declare_arguments"]],"ctranslate2.converters.TransformersConverter":[[44,2,1,"","__init__"],[44,2,1,"","convert"],[44,2,1,"","convert_from_args"],[44,2,1,"","declare_arguments"],[44,2,1,"","get_model_file"],[44,2,1,"","load_model"],[44,2,1,"","load_tokenizer"]],"ctranslate2.models":[[49,1,1,"","Wav2Vec2"],[50,1,1,"","Wav2Vec2Bert"],[51,1,1,"","Whisper"],[52,1,1,"","WhisperGenerationResult"],[53,1,1,"","WhisperGenerationResultAsync"]],"ctranslate2.models.Wav2Vec2":[[49,2,1,"","__init__"],[49,4,1,"","compute_type"],[49,4,1,"","device"],[49,4,1,"","device_index"],[49,2,1,"","encode"],[49,2,1,"","load_model"],[49,4,1,"","model_is_loaded"],[49,4,1,"","num_active_batches"],[49,4,1,"","num_queued_batches"],[49,4,1,"","num_workers"],[49,4,1,"","tensor_parallel"],[49,2,1,"","unload_model"]],"ctranslate2.models.Wav2Vec2Bert":[[50,2,1,"","__init__"],[50,4,1,"","compute_type"],[50,4,1,"","device"],[50,4,1,"","device_index"],[50,2,1,"","encode"],[50,2,1,"","load_model"],[50,4,1,"","model_is_loaded"],[50,4,1,"","num_active_batches"],[50,4,1,"","num_queued_batches"],[50,4,1,"","num_workers"],[50,4,1,"","tensor_parallel"],[50,2,1,"","unload_model"]],"ctranslate2.models.Whisper":[[51,2,1,"","__init__"],[51,2,1,"","align"],[51,4,1,"","compute_type"],[51,2,1,"","detect_language"],[51,4,1,"","device"],[51,4,1,"","device_index"],[51,2,1,"","encode"],[51,2,1,"","generate"],[51,4,1,"","is_multilingual"],[51,2,1,"","load_model"],[51,4,1,"","model_is_loaded"],[51,4,1,"","n_mels"],[51,4,1,"","num_active_batches"],[51,4,1,"","num_languages"],[51,4,1,"","num_queued_batches"],[51,4,1,"","num_workers"],[51,4,1,"","tensor_parallel"],[51,2,1,"","unload_model"]],"ctranslate2.models.WhisperGenerationResult":[[52,4,1,"","logits"],[52,4,1,"","no_speech_prob"],[52,4,1,"","scores"],[52,4,1,"","sequences"],[52,4,1,"","sequences_ids"]],"ctranslate2.models.WhisperGenerationResultAsync":[[53,2,1,"","done"],[53,2,1,"","result"]],"ctranslate2.specs":[[57,1,1,"","Activation"],[58,1,1,"","EmbeddingsMerge"],[59,1,1,"","LanguageModelSpec"],[60,1,1,"","LayerSpec"],[61,1,1,"","ModelSpec"],[62,1,1,"","RotaryScalingType"],[63,1,1,"","SequenceToSequenceModelSpec"],[64,1,1,"","TransformerDecoderModelSpec"],[65,1,1,"","TransformerDecoderSpec"],[66,1,1,"","TransformerEncoderModelSpec"],[67,1,1,"","TransformerEncoderSpec"],[68,1,1,"","TransformerSpec"],[69,1,1,"","Wav2Vec2BertSpec"],[70,1,1,"","Wav2Vec2Spec"],[71,1,1,"","WhisperSpec"]],"ctranslate2.specs.Activation":[[57,3,1,"","GELU"],[57,3,1,"","GELUSigmoid"],[57,3,1,"","GELUTanh"],[57,3,1,"","RELU"],[57,3,1,"","SWISH"],[57,3,1,"","Sigmoid"],[57,3,1,"","Tanh"]],"ctranslate2.specs.EmbeddingsMerge":[[58,3,1,"","ADD"],[58,3,1,"","CONCAT"]],"ctranslate2.specs.LanguageModelSpec":[[59,2,1,"","__init__"],[59,4,1,"","config"],[59,2,1,"","get_default_config"],[59,2,1,"","get_vocabulary_size"],[59,4,1,"","name"],[59,2,1,"","optimize"],[59,2,1,"","register_file"],[59,2,1,"","register_vocabulary"],[59,4,1,"","revision"],[59,2,1,"","save"],[59,2,1,"","validate"],[59,2,1,"","variables"]],"ctranslate2.specs.LayerSpec":[[60,2,1,"","optimize"],[60,2,1,"","validate"],[60,2,1,"","variables"]],"ctranslate2.specs.ModelSpec":[[61,2,1,"","__init__"],[61,4,1,"","config"],[61,2,1,"","get_default_config"],[61,4,1,"","name"],[61,2,1,"","optimize"],[61,2,1,"","register_file"],[61,4,1,"","revision"],[61,2,1,"","save"],[61,2,1,"","validate"],[61,2,1,"","variables"]],"ctranslate2.specs.RotaryScalingType":[[62,3,1,"","Linear"],[62,3,1,"","Llama3"],[62,3,1,"","Su"]],"ctranslate2.specs.SequenceToSequenceModelSpec":[[63,2,1,"","__init__"],[63,4,1,"","config"],[63,2,1,"","get_default_config"],[63,2,1,"","get_source_vocabulary_size"],[63,2,1,"","get_target_vocabulary_size"],[63,4,1,"","name"],[63,2,1,"","optimize"],[63,2,1,"","register_file"],[63,2,1,"","register_source_vocabulary"],[63,2,1,"","register_target_vocabulary"],[63,2,1,"","register_vocabulary_mapping"],[63,4,1,"","revision"],[63,2,1,"","save"],[63,2,1,"","validate"],[63,2,1,"","variables"]],"ctranslate2.specs.TransformerDecoderModelSpec":[[64,2,1,"","__init__"],[64,4,1,"","config"],[64,2,1,"","from_config"],[64,2,1,"","get_default_config"],[64,2,1,"","get_vocabulary_size"],[64,4,1,"","name"],[64,2,1,"","optimize"],[64,2,1,"","register_file"],[64,2,1,"","register_vocabulary"],[64,4,1,"","revision"],[64,2,1,"","save"],[64,2,1,"","validate"],[64,2,1,"","variables"]],"ctranslate2.specs.TransformerDecoderSpec":[[65,2,1,"","__init__"],[65,4,1,"","config"],[65,2,1,"","optimize"],[65,2,1,"","validate"],[65,2,1,"","variables"]],"ctranslate2.specs.TransformerEncoderModelSpec":[[66,2,1,"","__init__"],[66,4,1,"","config"],[66,2,1,"","get_default_config"],[66,2,1,"","get_vocabulary_size"],[66,4,1,"","name"],[66,2,1,"","optimize"],[66,2,1,"","register_file"],[66,2,1,"","register_vocabulary"],[66,4,1,"","revision"],[66,2,1,"","save"],[66,2,1,"","validate"],[66,2,1,"","variables"]],"ctranslate2.specs.TransformerEncoderSpec":[[67,2,1,"","__init__"],[67,2,1,"","optimize"],[67,2,1,"","validate"],[67,2,1,"","variables"]],"ctranslate2.specs.TransformerSpec":[[68,2,1,"","__init__"],[68,4,1,"","config"],[68,2,1,"","from_config"],[68,2,1,"","get_default_config"],[68,2,1,"","get_source_vocabulary_size"],[68,2,1,"","get_target_vocabulary_size"],[68,4,1,"","name"],[68,2,1,"","optimize"],[68,2,1,"","register_file"],[68,2,1,"","register_source_vocabulary"],[68,2,1,"","register_target_vocabulary"],[68,2,1,"","register_vocabulary_mapping"],[68,4,1,"","revision"],[68,2,1,"","save"],[68,2,1,"","validate"],[68,2,1,"","variables"]],"ctranslate2.specs.Wav2Vec2BertSpec":[[69,2,1,"","__init__"],[69,4,1,"","config"],[69,2,1,"","get_default_config"],[69,2,1,"","get_vocabulary_size"],[69,4,1,"","name"],[69,2,1,"","optimize"],[69,2,1,"","register_file"],[69,2,1,"","register_vocabulary"],[69,4,1,"","revision"],[69,2,1,"","save"],[69,2,1,"","validate"],[69,2,1,"","variables"]],"ctranslate2.specs.Wav2Vec2Spec":[[70,2,1,"","__init__"],[70,4,1,"","config"],[70,2,1,"","get_default_config"],[70,2,1,"","get_vocabulary_size"],[70,4,1,"","name"],[70,2,1,"","optimize"],[70,2,1,"","register_file"],[70,2,1,"","register_vocabulary"],[70,4,1,"","revision"],[70,2,1,"","save"],[70,2,1,"","validate"],[70,2,1,"","variables"]],"ctranslate2.specs.WhisperSpec":[[71,2,1,"","__init__"],[71,4,1,"","config"],[71,2,1,"","get_default_config"],[71,2,1,"","get_vocabulary_size"],[71,4,1,"","name"],[71,2,1,"","optimize"],[71,2,1,"","register_file"],[71,2,1,"","register_vocabulary"],[71,4,1,"","revision"],[71,2,1,"","save"],[71,2,1,"","validate"],[71,2,1,"","variables"]],ctranslate2:[[19,1,1,"","AsyncGenerationResult"],[20,1,1,"","AsyncScoringResult"],[21,1,1,"","AsyncTranslationResult"],[22,1,1,"","DataType"],[23,1,1,"","Device"],[24,1,1,"","Encoder"],[25,1,1,"","EncoderForwardOutput"],[26,1,1,"","ExecutionStats"],[27,1,1,"","GenerationResult"],[28,1,1,"","GenerationStepResult"],[29,1,1,"","Generator"],[30,1,1,"","MpiInfo"],[31,1,1,"","ScoringResult"],[32,1,1,"","StorageView"],[33,1,1,"","TranslationResult"],[34,1,1,"","Translator"],[35,5,1,"","contains_model"],[36,0,0,"-","converters"],[45,5,1,"","get_cuda_device_count"],[46,5,1,"","get_log_level"],[47,5,1,"","get_supported_compute_types"],[48,0,0,"-","models"],[54,5,1,"","set_log_level"],[55,5,1,"","set_random_seed"],[56,0,0,"-","specs"]]},objnames:{"0":["py","module","Python module"],"1":["py","class","Python class"],"2":["py","method","Python method"],"3":["py","attribute","Python attribute"],"4":["py","property","Python property"],"5":["py","function","Python function"]},objtypes:{"0":"py:module","1":"py:class","2":"py:method","3":"py:attribute","4":"py:property","5":"py:function"},terms:{"0":[0,1,3,5,6,10,11,14,16,17,22,23,24,29,32,34,47,49,50,51,52,57,58,62,64,65,73,74,76,77],"001":1,"02":10,"02155":[65,67,68],"04":[14,76],"05202":[64,65,67,68],"05424":[29,34,51],"1":[0,1,3,5,6,11,12,16,17,22,23,24,29,32,34,49,50,51,57,58,62,64,65,67,68,73,74],"10":[1,5,6,11,73,74],"1000":10,"10000":[64,65],"100mb":73,"1024":[1,5,29,34],"10683":[65,67,68],"11":[3,14,74],"12":[3,14,73],"127":73,"128k":6,"15":[0,14],"16000":11,"17":[0,14],"1803":[65,67,68],"182mb":73,"187mb":73,"19":14,"1910":[65,67,68],"1t":11,"2":[0,1,3,5,8,14,16,17,22,24,29,32,34,40,49,50,51,57,62,64,68,73,74,76,77],"20":[5,11,29,73],"200":11,"2002":[64,65,67,68],"200mb":3,"2016":73,"2017":73,"2018":1,"2019":14,"2020":[10,17],"209715200":3,"20b":11,"21":11,"2204":[29,34,51],"23":11,"24gb":11,"256":34,"26":10,"2b":6,"2b_last_checkpoint":6,"2x4":32,"3":[0,1,3,12,14,16,22,57,74],"30":11,"3000":[49,50],"32":[29,34],"32k":73,"350m":11,"364mb":73,"39m":11,"4":[1,3,11,12,16,22,32,57,74],"418m":6,"418m_last_checkpoint":6,"448":51,"4bit":[64,65],"5":[1,6,12,14,22,51,57,73,74],"50":[29,51],"50277":5,"50278":5,"50279":5,"512":[5,29],"560m":11,"6":[57,73],"600m":11,"6291455":3,"64":[11,12,14,29,34,73],"6b":11,"7":[3,5,11,14,17,51,73],"70":11,"7b":[5,11,73],"8":[3,5,11,14,16,74],"80":[49,50],"800":11,"9":11,"95mb":73,"\u00fcbersetzungen":1,"\u00fcbersetzungsmodel":1,"\u00fcbersetzungsmodellen":1,"abstract":[59,63],"boolean":3,"break":5,"byte":[24,29,34,49,50,51],"case":[9,14,15,16,74],"char":74,"class":[2,5,11,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,48,49,50,51,52,53,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,76],"const":[74,76],"default":[1,3,6,14,16,17,24,29,34,49,50,51,59,61,63,64,66,68,69,70,71,73],"do":[1,5,64,73],"enum":[57,58,62],"erm\u00f6glicht":1,"export":[3,4,16,74],"f\u00fcr":1,"final":[0,1,17,64,65,67,68],"float":[29,34,49,50,51,64,65],"function":[0,1,3,29,34,35,37,38,39,40,41,42,43,44,54,73],"import":[3,5,6,10,11,54,74,76],"int":[22,23,24,29,30,34,45,46,47,49,50,51,54,55,64,65,67,68,74,76],"long":[0,29],"new":[6,14,32],"return":[1,5,11,16,19,20,21,24,29,32,34,37,38,39,40,41,42,43,44,45,46,47,49,50,51,53,59,60,61,63,64,65,66,67,68,69,70,71,76],"short":5,"static":[5,29,30,32,37,38,39,40,41,42,43,44],"true":[1,3,5,11,16,19,20,21,24,29,34,49,50,51,53,64,65,67,68],"try":[1,16],A:[0,1,24,26,27,29,31,32,33,34,42,46,48,51,52,54,60,76],As:73,At:5,By:[1,3,14,73],For:[0,3,5,6,11,13,17,29,51,73],If:[5,9,11,12,14,16,17,19,20,21,24,29,34,37,38,39,40,41,42,43,44,49,50,51,53,59,60,61,63,64,65,66,67,68,69,70,71,74,76],In:[0,1,9,11,12,15,16,73,76],Is:6,It:[0,1,5,11,14,29,30,34,74,76],No:[5,6],ON:14,On:[12,14,73],The:[0,1,2,3,4,5,6,7,8,10,11,12,13,14,15,16,17,24,28,29,31,32,34,49,50,51,54,59,61,63,64,65,66,68,69,70,71,73,74,76,77],There:[0,4,6],These:1,To:[5,11,14,15,76],_:[11,16],__de__:6,__en__:6,__init__:[22,23,24,29,34,38,39,40,41,42,43,44,49,50,51,59,61,63,64,65,66,67,68,69,70,71],__x__:6,_batch:17,_file:17,_iter:17,aarch64:[12,14,73],ab:[29,34,51,64,65,67,68,73],abc:37,aber:1,abl:[4,5],about:[3,30,73],abov:[3,11],absolut:73,accelar:4,acceler:[0,1,14,29,73],accept:[6,11,29,34,51,73],access:[0,9,11],accord:16,accumul:[3,73],accuraci:[11,44,73],act:[24,29,34,49,50,51],activ:[11,44,56,64,65,66,67,68,72],activation_scal:[11,44],actual:73,ad:[0,4,5,11,29,76],adapt:5,add:[11,14,37,38,39,40,41,42,43,44,58,64,65,66,74],add_execut:74,addit:[14,16,76],address:4,advanc:76,affect:11,after:[1,29,34,64,65,66,67,68],against:14,aggreg:11,agnost:[0,4],ai:5,aim:11,al:[1,73],algorithm:1,alia:[59,60,61,63,64,65,66,67,68,69,70,71],alibi:[64,65],alibi_use_positive_posit:[64,65],align:[51,65,68],alignment_head:[65,68],alignment_lay:[65,68],all:[1,6,11,16,28,38,59,60,61,63,64,65,66,67,68,69,70,71,73,76],alloc:[3,32],allow:[1,3,44,49,50,51],alpha:5,alreadi:[7,11,37,38,39,40,41,42,43,44],also:[1,3,5,6,8,9,10,11,14,16,24,29,51,73],altern:[8,9,29,34],alwai:[1,5,8,16,76],am:11,amazonaw:74,amd:[12,14],amid:11,an:[0,1,4,5,8,11,14,17,19,20,21,24,25,29,30,32,34,42,49,50,51,53,59,60,61,63,64,65,66,67,68,69,70,71,76],ani:[16,29,34],anim:11,anoth:[0,6,14,32,64,73],anyth:5,api:[0,1,3,74,76,77],appear:1,append:[5,11],appl:[14,73],appli:[1,6,10,17,29,34,51,59,60,61,63,64,65,66,67,68,69,70,71,73],applic:[4,14,16],approach:4,ar:[0,1,3,5,6,7,8,10,11,12,14,16,17,24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73,74,76,77],arbitrarili:[29,34],architectur:[0,3,4,11,12,14,64,65,67,68,73],arg:[37,38,39,40,41,42,43,44],argc:74,argmax:11,argument:[1,5,11,17,24,29,34,37,38,39,40,41,42,43,44,49,50,51,54],argumentpars:[37,38,39,40,41,42,43,44],argv:74,arm64:[12,14,16,73],around:[1,19,20,21,53],arrai:[11,24,29,32,49,50,51],articl:3,arxiv:[29,34,51,64,65,67,68],as_tensor:[11,32],ask:13,assert:1,assioc:76,assist:5,associ:76,assumpt:4,async:29,async_generate_token:29,async_result:16,asyncgenerationresult:[18,29,72],asynchron:[3,19,20,21,29,34,51,53,76],asynciter:29,asyncscoringresult:[18,29,34,72],asynctranslationresult:[18,34,72],attach:[24,29,34,49,50,51],attent:[24,29,33,34,49,50,51,64,65,67,68,71],attribut:[22,23,24,25,26,27,28,29,31,32,33,34,49,50,51,52,59,61,63,64,65,66,68,69,70,71],auch:1,audio:[11,49,50,51],auf:1,ausgerichtet:1,author:73,auto:[14,24,29,34,49,50,51,73,74,76],auto_config:42,autoawq:73,automat:[3,11,12,24,29,34,42,49,50,51],automodelforsequenceclassif:11,autoregress:11,autotoken:[5,11],avail:[1,16,19,20,21,24,29,34,49,50,51,53,76],averag:74,avoid:[3,5,17],avx2:[3,12],avx512:[3,12,17],avx:[3,12],awq:[64,65],back:[1,15,24,29,34,49,50,51],backend:[3,12,14,73],background:5,backward:77,bart:6,base:[3,11,16,37,59,63,64,65,73],basic:1,batch:[5,16,17,24,28,29,34,49,50,51,74,76],batch_gener:16,batch_id:28,batch_siz:[24,29,49,50,51],batch_typ:[17,29,34],bdist_wheel:14,beam:[17,29,34,51],beam_siz:[1,6,17,29,34,51],becaus:16,bedienen:1,bedienung:1,been:3,befor:[3,11,34,49,50,51,64,65],begin:51,behavior:[6,12,16,73],believ:11,below:[1,5,11,12,16,17,73],benchmark:73,bert:[2,66],bertmodel:11,beschleunigen:1,beschleunigung:1,best:[1,3,12,51],better:[1,3,4,11],between:[5,16,30,51,73],bf16:[3,73],bfloat16:[22,24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],bia:[1,65,67,68],bias:[34,64,65],big:16,bigcod:11,bigscienc:11,billion:11,bin:[0,6],bin_growth:3,binari:[0,12,14,73],bit:[11,64,65],blackout:11,blank:51,block:[16,19,20,21,24,29,34,49,50,51,53,64,65],blue:6,bold:14,bool:[19,20,21,24,29,34,35,37,38,39,40,41,42,43,44,49,50,51,53,59,60,61,63,64,65,66,67,68,69,70,71],boost:16,bos_token:11,both:[4,12,76],bpe:6,branch:11,broken:0,brows:74,bs:16,buffer:[3,32],build:[0,74],build_cli:14,build_test:14,built:[29,34,42],bunch:74,bz2:6,c:[0,4,29,34,54,77],cach:[3,17,24,29,34,49,50,51,64,65],cache_static_prompt:29,calcul:34,call:[4,5,11,16,29,34,37,38,39,40,41,42,43,44,73],callabl:[29,34],callback:[5,29,34],can:[0,1,3,5,6,8,9,11,14,15,16,17,24,29,32,34,42,73,74,76],candid:[1,29,34,51,76],cannot:[37,38,39,40,41,42,43,44],capabl:[12,17,73],casper:73,catalog:42,cd:[14,74],chang:[0,1,59,61,63,64,66,68,69,70,71,73],chat:[5,11],cheap:11,check:[0,16,35],checkpoint:[9,42],checkpoint_path:42,chief:6,children:[59,60,61,63,64,65,66,67,68,69,70,71],chunk:[17,29,34],chunk_length:51,classif:2,classifi:11,classmethod:[42,64,68],clear:5,cli:11,client:[14,77],clone:14,close:5,cmake:[14,74],cmake_cxx_flag:14,cmake_install_prefix:14,cmake_minimum_requir:74,cmake_prefix_path:74,cmakelist:74,cnn:11,co:73,code:[0,3,4,6,11,44,74],codegen:11,collect:[10,11,48],com:[6,11,14,40,44,49,50,51,74],combin:1,come:0,comma:3,command:[8,11,14,37,38,39,40,41,42,43,44],common:[1,37,38,39,40,41,42,43,44,73],commun:30,comp:14,compar:[11,73],compat:[10,12,29,34,77],compil:[12,74],complet:[1,5,8,11,29,34],compress:1,comput:[3,4,5,6,11,12,16,17,19,20,21,24,29,34,44,47,49,50,51,53,73],compute_typ:[24,29,34,49,50,51,73],concat:[58,67,68],concurr:[16,29,34],condit:11,confid:1,config:[0,9,16,42,51,59,61,63,64,65,66,68,69,70,71],configur:[0,3,9,12,14,16,42,59,61,63,64,66,68,69,70,71],connect:[16,64,65],connexion:30,consciou:11,consid:[5,11,17,29,74],constant:17,constrain:29,constructor:16,contain:[0,6,10,14,26,35,37,38,39,40,41,42,43,44,74],contains_model:[0,18,72],content:[24,29,34,49,50,51],context:[1,24,29,34,49,50,51],continu:[29,34,51],control:[1,3,4,12,54],conv:7,conveni:[5,10,54],convers:[4,6,7,8,9,11,13,37,38,39,40,41,42,43,44],convert:[6,7,8,9,10,11,32,38,39,40,41,42,43,44,49,50,51,60,72,73,74,76,77],convert_from_arg:[37,38,39,40,41,42,43,44],convert_ids_to_token:[5,11],convert_tokens_to_id:11,convolut:14,copi:[5,32,44,49,50,51],copy_fil:[44,73],core:[0,3,4,17],correspond:[1,9,11,73],costli:3,could:[0,4,5,16],cout:[74,76],cover:[4,77],coverag:34,coverage_penalti:34,cpp:74,cpu:[0,3,4,5,6,11,14,15,23,24,29,32,34,47,49,50,51,73,74,76],creat:[16,24,29,32,42,64,68],critic:3,csc:10,ct2:[5,6,7,8,9,10,11,73,74],ct2_force_cpu_isa:12,ct2_model:[5,6,7,8,9,10,11,73],ct2_use_experimental_packed_gemm:17,ct2_use_mkl:12,ct2_verbos:73,ctranslat:74,ctranslate2:[0,1,2,3,4,5,6,7,8,9,10,11,13,14,16,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,37,38,39,40,41,42,43,44,45,46,47,49,50,51,52,53,54,55,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77],ctranslate2_demo:74,ctranslate2_root:14,ctranslate_install_path:74,cub:3,cub_cach:3,cubla:[4,14],cuda11:[14,76],cuda:[3,11,12,14,16,23,24,29,32,34,47,49,50,51,73,76],cuda_arch_list:14,cuda_dynamic_load:14,cuda_malloc_async:3,cuda_nvcc_flag:14,cuda_select_nvcc_arch_flag:14,cudamalloc:3,cudnn:14,cumul:[29,34],current:[4,6,11,24,29,30,34,46,49,50,51,73],custom:[3,4,11,14,16,38,44,74],d_model:51,da:1,daili:11,data:[6,17,29,32,34,38,76],data_dir:[6,38],dataset:11,datatyp:[18,32,72],dazu:1,dcmake_install_prefix:74,dcmake_prefix_path:74,de:[1,6,10,11,74],debug:3,decid:3,declar:[38,60],declare_argu:[37,38,39,40,41,42,43,44],decod:[3,5,6,8,10,11,13,17,28,29,34,38,51,64,65,68,71,74],def:11,defin:[0,6,14,51,73],definit:[64,65],degrad:73,del:15,delet:[15,24,29,34,49,50,51],demonstr:[1,11],den:1,dens:[24,29],depend:[4,12,14,16,29,74],deprec:4,der:1,describ:[1,11,29,34,51,64,65,66,67,68,71],descript:[3,5,13,14],detail:1,detect:[11,73],detect_languag:[11,51],detoken:[1,34],detokenis:74,detokenize_fn:34,develop:[5,11,14],devic:[3,5,6,11,16,18,24,29,32,34,45,47,49,50,51,72,73,74,76],device_index:[16,17,24,29,32,34,47,49,50,51],devlin:73,dict:[6,24,29,34,42,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],dictionari:[6,24,29,34,38,42,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],die:1,dienst:1,dient:1,dies:1,differ:[0,1,5,6,16,68],dimens:[51,64,65,68],direct:4,directli:[8,9,14,76],directori:[0,6,9,11,14,24,29,34,35,37,38,39,40,41,42,43,44,49,50,51,59,61,63,64,66,68,69,70,71,74],disabl:[3,14,16,17,27,28,29,33,34,51,52,65,67,68],disable_unk:[29,34],disk:[4,59,61,63,64,66,68,69,70,71,73],dispatch:[3,14],dist:14,distil:11,distilbertmodel:11,distribut:1,diverg:1,dl:6,dn:16,dnnl:14,doc:16,docker:76,document:[0,9,13,14,73,76],doe:[1,3,5,6,32,76],don:[9,11,48],done:[19,20,21,53],doption:14,dot:[64,65],download:[6,11,44,73,74],driver:12,dry:11,dtype:32,dure:[8,11,14,29,34,51,73,76],dwith_acceler:14,dwith_cuda:14,dwith_cudnn:14,dwith_dnnl:14,dwith_mkl:14,dwith_openbla:14,dwith_rui:14,dynam:[14,17],e:[3,8,11,14,15,24,29,32,38,59,61,63,64,66,68,69,70,71,73,74],each:[0,1,3,5,11,16,24,27,29,31,33,34,51,52,59,61,63,64,65,66,68,69,70,71,76],earli:5,earth:11,easi:0,echo:76,edunov:1,effect:[1,16],effekt:1,effici:[1,3,17,29,34,73,77],effizi:1,effizient:1,effizienten:1,effizienter:1,effizienz:1,eigen:4,ein:1,einen:1,einsatzort:1,either:[15,24,29,42],element:17,eleutherai:11,ello:[6,76],els:11,emb:4,embed:[0,58,64,65,67,68,73],embeddings_merg:[67,68],embeddingsmerg:[56,67,68,72],empti:[1,27,33,52,76],en:[6,10,11,74],en_xx:6,enabl:[3,14,16,17,29,34,44,64,65,67,68,73,76],enable_cpu_dispatch:14,enable_profil:14,encod:[0,5,6,10,11,13,18,25,38,49,50,51,65,66,67,68,71,72,74,77],encoderforwardoutput:[18,24,72,77],end:[1,5,6,17,29,34,38,74],end_token:[5,29,34],ende_ctranslate2:[6,10,74,76],endian:0,endl:[74,76],enforc:1,eng_latn:11,engin:4,english:[10,11,74],enough:[24,29,34,49,50,51],ensur:73,entrypoint:[2,5,14,76],environ:[5,12,13,14,16,17,73],eo:[29,34],epoch:74,equal:12,equat:73,erprobung:1,error:[3,76],es:1,especi:4,et:[1,73],etc:[4,16],eval:11,even:[0,3,4,16],evolv:77,ex:[64,65],exampl:[0,1,2,3,5,6,8,11,14,16,24,26,29,32,34,47,49,50,54,73,74,75],exce:[29,34],except:[19,20,21,53,73],excit:5,execut:[0,3,4,14,17,26,49,50,51,73,74],executionstat:[18,34,72],exist:[0,4,24,29,34,37,38,39,40,41,42,43,44,49,50,51,76],exp:6,expand:[29,34],expans:1,expect:[0,1,11,29,59,63,64,66,68,69,70,71,76,77],experi:11,experiment:[1,3,75],explicitli:[5,14,68],exponenti:[29,34,51],expos:[0,2,5,73,76],extend:[16,37,39,59,60,61,63],extens:38,extern:4,extract:1,f:11,face:[11,44,73],facebook:[11,49,50],facebookresearch:[49,50],factor:[29,34,51,58,64,65],fair:6,fairseq:[0,11,13,38,49,50],fairseqconvert:[36,37,72],fallback:73,fals:[11,17,24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],fast:[0,3,4,11],faster:[4,11],fastest:[1,73],favor:4,fbaipublicfil:6,feat_lay:70,featur:[4,11,13,49,50,51,76],feed:[64,65,67,68],ffn:[64,65,67,68],ffn_glu:[64,65,67,68],fi:10,field:0,file:[0,6,11,16,24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,61,63,64,66,68,69,70,71,76],filenam:[44,59,61,63,64,66,68,69,70,71],fill:0,filter:[1,16,51],fim_middl:11,fim_prefix:11,fim_suffix:11,find:[1,11],find_packag:74,fine:11,finetun:11,finish:[29,34,51],first:[1,11,29,34,51,64,65,74],fit:48,fix:[6,38],fixed_dictionari:[6,38],flag:[1,3,12,14,44],flan:11,flash:[24,29,34,49,50,51],flash_attent:[24,29,34,49,50,51],float16:[11,22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],float32:[22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],flush:5,folgebeschleunigung:1,follow:[1,3,6,11,14,16,29,34,73,76,77],forc:[1,3,37,38,39,40,41,42,43,44],forecast:11,format:[0,6,8,11,34,37,38,39,40,41,42,43,44,74,76],formkomprimierung:1,formula:73,forward:[0,1,17,24,25,29,64,65,67,68],forward_batch:[5,11,24,29],fp16:[3,11,73],fp32:[3,73],fra_latn:11,frame:51,framework:[4,11],free:[24,29,34,49,50,51],frequent:[1,13],from:[0,1,3,4,5,6,8,9,11,12,16,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,49,50,51,52,53,54,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,73,74,76],from_arrai:[11,32],from_config:[42,64,68],from_pretrain:[5,11,44],frozenattr:60,ft:6,full:[0,5,16,24,29,34,49,50,51,73],fulli:[15,24,29,34,42,49,50,51],further:17,futur:[5,24,29,34,49,50,51],g:[3,8,14,15,24,29,32,38,59,61,63,64,66,68,69,70,71,73,74],gate:[64,65,67,68],gear:1,gelu:57,gelusigmoid:57,gelutanh:57,gemm:[3,17,73],gemv:73,gener:[1,3,6,11,13,16,18,24,27,28,34,41,48,51,52,55,72,73,76,77],generate_batch:[5,6,8,11,29,73],generate_iter:29,generate_token:[5,29,34],generationopt:[29,77],generationresult:[18,19,29,72,77],generationstepresult:[18,29,34,72],german:[10,11,74],get:[5,11,30,73,74],get_cuda_device_count:[18,72],get_default_config:[59,61,63,64,66,68,69,70,71],get_log_level:[18,72],get_model_fil:44,get_source_vocabulary_s:[63,68],get_supported_compute_typ:[18,72,73],get_target_vocabulary_s:[63,68],get_vocabulary_s:[59,64,66,69,70,71],getcurrank:[16,30],getlocalrank:30,getnrank:30,ghcr:[14,76],gil:16,girafatron:11,giraff:11,giraftron:11,git:14,github:[11,13,14,40,44,49,50,51,73],give:73,given:[34,51],gleichzeitig:1,global:16,gloriou:11,goal:4,good:11,gpt2:11,gpt2_ct2:11,gpt2token:11,gpt:[5,40,64,65],gpt_neox_ct2:11,gptj_ct2:11,gptneox20b:11,gpu:[3,4,5,16,30,45,73,76],gram:76,graph:4,greater:[0,12,14,29,34],greedi:[29,34,51],group:[64,65],guarante:[0,29,34,77],guid:[2,10,12,74,75],gz:[6,74],h:[6,9,74,76],ha:[1,4,16,29,34,76],half:[64,65,73],hall:74,hallo:74,han:44,handl:[73,74],hansen:73,hard:4,hardwar:13,harm:5,harmless:5,have:[0,3,5,6,9,11,14,16,34,51,74],head:[64,65,67,68,71],head_dim:[64,65],heavi:4,hei:11,hello:[6,8,10,11,24,34,74,76],help:[5,14,73,76],helper:[35,37,38,39,40,41,42,43,44],helsinki:[11,74],here:[4,11,29,73,76],hf:11,hidden:68,high:[2,3,5,11,76],higher:1,highest:[1,34],hostfil:16,hous:11,how:[0,5,6,11,67,68,76],howev:[0,4,11,73],http:[6,10,11,14,29,34,40,44,49,50,51,64,65,67,68,73,74],hub:[11,44],hug:[11,44,73],huggingfac:[11,73],human:5,hypothes:[1,6,10,11,29,33,34,51,74,76],hypothesi:[1,28,33],hypothesis_id:28,i:[11,29,73,76],id:[17,24,27,28,29,30,34,49,50,51,52],idea:73,identifi:[24,29,34,49,50,51],ignor:34,imag:[14,76],immedi:[1,16],impact:[1,3],implement:[0,3,5,11,14,16,32,49,50,51,73],impli:[4,73],implicitli:[5,11],improv:[3,11,16,17,44],includ:[0,1,5,6,8,11,13,14,16,29,34,37,38,39,40,41,42,43,44,51,74,76],include_prompt_in_result:[11,17,29],increas:[1,3,17,29,76,77],increment:[0,59,61,63,64,66,68,69,70,71],independ:1,index:[28,29,32,47,51,65,68],ineffekt:1,infer:[1,73],inferenzbeschleunigung:1,info:[3,54,73],inform:[5,12,16,30,32,73,74],inherit:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,37,38,39,40,41,42,43,44,49,50,51,52,53,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71],initi:[24,29,34,38,39,40,41,42,43,44,49,50,51,59,61,63,64,65,66,67,68,69,70,71],input:[1,5,6,11,16,17,24,29,32,34,49,50,51,64,65,73,74,76],input_featur:11,input_fil:34,input_id:11,input_text:[11,74],input_token:[11,74],instal:[6,8,9,11,13,16,74],install_requir:14,instanc:[1,15,16,24,29,32,34,42,49,50,51],instead:[1,3,14,29,59,60,61,63,64,65,66,67,68,69,70,71,73],instruct:[3,11,12,13,14],int16:[22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],int32:[22,24,29,32,73],int8:[17,22,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],int8_bfloat16:[24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],int8_float16:[11,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],int8_float32:[24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],integr:[3,4,14,75],intel:[3,4,12,14,17,73],intenum:[57,58,62],inter_thread:[16,17,24,29,34,49,50,51],interact:[5,11],interfac:32,interleav:[64,65],intermedi:[0,11,44],intern:[5,73],intra_thread:[16,17,24,29,34,49,50,51],introduc:[3,4,11],invari:68,io:[14,73,76],iostream:[74,76],ipaddress:16,irrelev:11,is_last:28,is_multilingu:51,is_new_word:5,isa:[3,14],issu:[0,4],ist:1,iter:[29,34],its:[5,14,29,59,60,61,63,64,65,66,67,68,69,70,71,73],j4:[14,74],j:[64,65,73],join:[6,34],joke:5,json:[0,51],just:[1,5],k:[14,29,34,51,76],keep:[1,17,24,29,34,49,50,51,73],keep_cach:[24,29,34,49,50,51],kei:[4,64,65],kernel:[14,51,73],known:[3,5],kv:[29,64,65],kwarg:[29,34,44],l:76,lab:44,lambda:34,lang_code_to_token:11,languag:[5,11,38,51,59,69,70],languagemodelspec:[56,61,64,66,69,70,71,72],larg:[4,5,11,17,29,34],larger:[11,17],last:[11,25,28,64],last_hidden_st:25,late:4,latenc:16,later:[15,16],latest:[14,42,76],layer:[11,14,17,24,25,29,34,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],layernorm_embed:[64,65,67,68],layerspec:[0,56,61,65,67,72],layout:[59,61,63,64,66,68,69,70,71],ld_library_path:[14,16],ldconfig:14,leaf:0,learn:[3,4],least:[1,11,12],left:[1,73],len:1,length:[17,24,29,34,51,64,65],length_penalti:[29,34,51],let:1,level:[0,2,3,5,34,46,54,61,76],librari:[3,4,29,34,54,73,74],librosa:11,life:[5,11],light:11,lightweight:4,like:[1,5,11,16,24,29,34,49,50,51,64,65,73],likelihood:5,limit:[1,16],line:[8,14,34,37,38,39,40,41,42,43,44,76],linear:[62,64,65,67,68,73],link:14,linux:14,list:[1,3,6,11,14,24,29,34,39,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,76],littl:73,llama3:62,llama:[5,73],lm:6,load:[0,5,6,10,11,14,15,16,24,29,34,42,44,49,50,51,77],load_as_float16:44,load_model:[15,24,29,34,44,49,50,51],load_token:44,local:[29,34],log:[3,5,28,29,31,34,46,51,54,73],log_prob:[6,28,31],logic:[0,4],login:11,logit:[5,11,27,28,29,33,52],longer:[11,16],look:74,loop:[5,29],low:1,low_cpu_mem_usag:44,lower:[1,64,65],lua:4,luatorch:4,m2m100:11,m2m100_418:11,m2m100_418m:11,m2m_100:6,m2m_100_418m_ct2:6,machin:[0,11,16,17,74],maco:[14,16],mai:[3,16,38,44],mail:11,main:[2,5,11,49,50,73,74,76],majesti:11,major:0,make:[4,5,8,11,14,16,17,74],manag:[0,5,13,30],mani:[4,17,29,34],manylinux2014:14,map:[24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,76],march:14,marian:[0,10,11,13,39],marianconvert:[36,37,43,72],mask:11,master:16,match:[6,76],materi:[29,34],matmul:16,matrix:33,max:[64,65,73],max_batch_s:[16,17,29,34],max_bin:3,max_cached_byt:3,max_decoding_length:[1,34],max_initial_timestamp_index:51,max_input_length:[1,29,34],max_length:[5,6,11,24,29,51],max_position_embed:[64,65],max_queued_batch:[16,24,29,34,49,50,51],maximum:[1,24,29,34,49,50,51,73],mbart50:6,mbart50_ct2:6,mbart:11,me:11,mean:[5,6,64,65,67,68],median:51,median_filter_width:51,mel:[49,50,51],memori:[1,3,4,11,13,16,24,29,34,49,50,51,73],merg:[58,67,68],meta:11,method:[5,11,15,16,17,19,20,21,24,29,30,32,34,37,38,39,40,41,42,43,44,49,50,51,53,59,60,61,63,64,65,66,67,68,69,70,71,76],middai:11,might:15,militari:6,millisecond:26,min:73,min_alternative_expansion_prob:[1,29,34],min_bin:3,min_decoding_length:[1,34],min_length:29,minim:[6,29,34],minimum:[1,29,34],mit:44,mixtur:11,mkdir:[14,74],mkl:[3,4,12,14,73],ml50_lang:6,mode:[1,16,24,29,34,49,50,51,73],model:[1,2,3,4,5,7,8,9,10,11,13,14,15,17,24,25,29,30,34,35,37,38,39,40,41,42,43,44,49,50,51,52,53,59,61,63,64,65,66,68,69,70,71,72,74,75,76,77],model_class:44,model_dict:6,model_dir:[10,40,43],model_is_load:[24,29,34,49,50,51],model_load:74,model_nam:11,model_name_or_path:44,model_path:[6,7,8,16,24,29,34,38,39,41,49,50,51,73,74],model_spec:[0,60],modellen:1,modellkompress:1,modellkomprimierung:1,modelload:74,modelspec:[56,59,60,63,72],modul:[0,3,14,64],mono:11,mood:5,more:[0,1,3,4,5,12,16,17,29,34,44,51,73,74,76],mosaicml:11,most:[1,9,11,14,16,29,34,73],move:[0,4,15,24,29,34,49,50,51],mpi:[16,30],mpiinfo:[16,18,72],mpirun:16,mt5:11,mt:[0,11,13,43,74],multi:[11,64,65,67,68],multi_query_attent:[64,65,67,68],multilingu:[6,11,38,51],multilingual_transform:6,multipl:[0,5,14,16,17,29,34,49,50,51,73],multiprocess:16,multithread:13,must:0,my:11,n:[1,11,34,64,65,76],n_mel:51,name:[0,5,6,11,17,22,23,24,29,34,42,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],namespac:[37,38,39,40,41,42,43,44],nbgpu1:16,nbgpu2:16,nbgpuexpect:16,ndaniel:11,ndarrai:[59,60,61,63,64,65,66,67,68,69,70,71],nearli:11,necessari:16,need:[4,11,16,74],neg:1,neox:[64,65],nest:0,network:[64,65,67,68],next:0,ngirafatron:11,ngram:[29,34,51],nlp:[11,74],nn:[0,6],no_default_special_token:38,no_final_norm:[64,65,67,68],no_repeat_ngram_s:[29,34,51],no_speech_prob:52,non:[1,14,51,73],none:[14,22,23,24,28,29,34,37,38,39,40,41,42,43,44,49,50,51,55,59,60,61,63,64,65,66,67,68,69,70,71],nonsens:1,norm:[64,65,67,68],normal:[34,64,65,67,68],note:16,notimestamp:11,notimplementederror:[37,38,39,40,41,42,43,44],now:4,np:[6,11,16,32],npz:[7,39],num_active_batch:[24,29,34,49,50,51],num_adapter_lay:69,num_decoder_head:71,num_decoder_lay:71,num_encod:24,num_encoder_head:71,num_encoder_lay:71,num_exampl:26,num_fram:51,num_gener:29,num_head:[64,65,67,68,70],num_heads_kv:[64,65],num_hidden_lay:69,num_hypothes:[1,29,34,51],num_languag:51,num_lay:[64,65,67,68,70],num_queued_batch:[24,29,34,49,50,51],num_source_embed:[67,68],num_token:26,num_transl:34,num_work:[49,50,51],number:[0,1,16,17,24,26,29,30,34,45,49,50,51,64,65,67,68,71,76],numpi:[6,11,24,29,32],nvcc:14,nvidia:[12,14,17,73],o:74,obj:29,object:[0,10,11,15,16,19,20,21,24,29,30,32,34,49,50,51,53],obsess:11,obtain:6,occur:4,occurr:1,oder:1,off:[3,14],offset:34,older:0,omp_:16,onc:[5,11,14,17,29],one:[0,1,16,29,34,64,73],onednn:[12,14,73],ones:32,onli:[0,2,4,5,8,11,12,16,29,73],onmt_release_model:8,open:[4,5,16,34],openai:[11,40,51],openaigpt2convert:[36,37,72],openbla:14,openmp:[14,16,24,29,34,49,50,51],openmp_runtim:14,opennmt:[0,4,13,14,41,42,73,74,76],opennmtpyconvert:[36,37,72],opennmttfconvert:[36,37,72],oper:[0,3,16],opt:14,optim:[0,17,29,34,59,60,61,63,64,65,66,67,68,69,70,71,73],option:[0,1,6,9,11,24,29,34,37,38,39,40,41,42,43,44,51,59,60,61,63,64,65,66,67,68,69,70,71,73,76,77],opu:[0,11,13,43,74],opus_model:10,opusmtconvert:[36,39,72],order:[3,51,59,60,61,63,64,65,66,67,68,69,70,71],org:[29,34,51,64,65,67,68],origin:[64,65,73],original_max_position_embed:[64,65],ort:1,os:[3,14],other:[4,9,11,12,16,73,74,77],otherwis:[16,64,65,74],out:[0,1,5,16],out_typ:[5,6,10,74],output:[1,5,6,8,10,11,24,25,26,29,34,37,38,39,40,41,42,43,44,49,50,51,59,61,63,64,66,68,69,70,71,73,74,76],output_dir:[6,7,8,9,10,11,37,38,39,40,41,42,43,44,59,61,63,64,66,68,69,70,71,73,74],output_id:5,output_path:34,output_text:[11,74],output_token:[11,74],outsid:[9,74],over:[17,29,34],overcom:4,overrid:[3,37,38,39,40,41,42,43,44],overview:4,own:[0,3,5,34,73,74],pack:[3,17],packag:[14,74],pad:[29,34,51],page:[1,11],pair:[16,51],paper:[65,67,68],parallel:[13,24,29,34,49,50,51,64,65],parallel_residu:[64,65],paramet:[1,3,6,11,16,24,29,32,34,37,38,39,40,41,42,43,44,47,49,50,51,54,59,60,61,63,64,65,66,67,68,69,70,71,73],pars:[37,38,39,40,41,42,43,44],parser:[37,38,39,40,41,42,43,44],part:6,particip:5,particular:12,pass:[9,11,17,32,76],path:[6,7,8,11,14,16,24,29,34,35,37,38,39,40,41,42,43,44,49,50,51,59,61,63,64,66,68,69,70,71,74],patienc:[29,34,51],pegasu:11,penal:[29,34,51],penalti:[29,34,51],per:[24,29,34,49,50,51,73],perform:[3,12,13,16,29,73],period:[64,65],perplex:[5,6],pg:11,physic:17,pile:11,pip:[6,8,9,11,74],place:[1,24,29,34,49,50,51],plan:[11,14],platform:[12,73],poetri:5,polar:11,pool:[3,11,25,66],pooler_output:[11,25],pooling_activ:66,pooling_lay:66,popul:0,posit:[29,34,64,65,67,68],possibl:[0,3,4,16,17,24,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],post:[64,65],pouta:10,practic:1,pre:[11,44,64,65,67,68],pre_norm:[64,65,67,68],pre_post_layer_norm:[64,65],prebuilt:[12,73],precis:[3,44,73],predict:[17,29,34,51],predicted_class_id:11,prefer:17,prefetch:[29,34],prefix:[1,6,34,59,60,61,63,64,65,66,67,68,69,70,71],prefix_bias_beta:[1,34],prepend:[59,60,61,63,64,65,66,67,68,69,70,71],pretrain:[6,10,11,44,74],prevent:[29,34,51],previous:[0,3,29,34,51],print:[1,5,6,10,11,16,32,34,74,76],print_hello_world:11,prob:[5,29,34,51],probabilti:29,probabl:[1,4,11,28,29,31,34,51,52],process:[16,17,24,26,29,30,34,49,50,51],processor:[11,12],produc:0,product:[0,4,14,64,65],profil:14,programm:1,progress:5,project:[1,3,11,12,13,64,74,77],project_in_out:[64,65],projekt:1,prompt:[11,17,29,51],prompt_token:5,properli:74,properti:[22,23,24,25,26,27,28,29,31,32,33,34,49,50,51,52,59,61,63,64,65,66,68,69,70,71],propos:11,provid:[5,10,14,15,30,38,76],pt:[6,8,38,41,73,74],pubkei:16,publish:[6,11,14,49,50,51],pull:14,purpos:15,pwd:[74,76],py:[0,4,13,14,41,73,74],pybind11:14,pybind11_builtin:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,49,50,51,52,53],pybind11_object:[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,49,50,51,52,53],pyonmt:74,pypi:14,python3:16,python:[0,3,13,16,46,54,73,77],pytorch:[0,6,8,11,24,29,32,38,41],quant_bit:[64,65],quant_group_s:[64,65],quant_typ:[64,65],quantiz:[11,13,14,17,37,38,39,40,41,42,43,44,59,60,61,63,64,65,66,67,68,69,70,71],queri:[64,65,67,68,73],question:13,queue:[16,24,29,34,49,50,51],quickli:[24,29,34,49,50,51],quickstart:[8,13,76],r:14,rais:[19,20,21,32,37,38,39,40,41,42,43,44,51,53,59,60,61,63,64,65,66,67,68,69,70,71,76],random:[29,34,51,55],randomli:[1,29,34,51],rang:[11,73],rank:30,rare:0,re:16,reach:11,read:[0,16,34],read_batch_s:34,readi:[24,29,34,49,50,51,74],readm:73,reason:3,recognit:[11,13,14,49,50,51],recommend:[11,17],recurs:[14,59,60,61,63,64,65,66,67,68,69,70,71],redefin:4,reduc:[4,11,73,76],reduct:17,refer:[13,73],refus:5,regist:[59,61,63,64,66,68,69,70,71],register_fil:[59,61,63,64,66,68,69,70,71],register_source_vocabulari:[63,68],register_target_vocabulari:[63,68],register_vocabulari:[59,64,66,69,70,71],register_vocabulary_map:[63,68],registri:14,rel:[64,65,67,68],relative_attention_bia:[65,67,68],relative_posit:[65,67],releas:[8,15,16],reli:4,relianc:4,relu:[57,64,65,67,68],remain:29,remov:11,renam:[59,61,63,64,66,68,69,70,71],repetit:[29,34,51],repetition_penalti:[29,34,51],replac:[1,34],replace_unknown:34,replica:5,repositori:[11,13,14],repres:0,represent:[0,64,65,67,68],request:[11,24,29,34,49,50,51],requir:[4,5,6,7,8,11,12,14,24,29,34,38,59,60,61,63,64,65,66,67,68,69,70,71],resampl:11,rescal:[11,44],residu:[64,65],resolv:42,resourc:15,respect:14,respons:11,rest:[1,4],restrict:[0,1],result:[1,6,10,11,16,19,20,21,27,28,29,31,33,34,51,52,53,73,74,76],resum:[24,29,34,49,50,51],retain:[64,65],retriev:16,return_altern:[1,29,34],return_attent:[33,34],return_end_token:[29,34],return_log_prob:[28,29,34],return_logits_vocab:[27,29,33,34,51,52],return_no_speech_prob:[51,52],return_scor:[17,27,29,33,34,51,52],return_tensor:11,reus:[3,5,29],revis:[0,11,44,59,61,63,64,66,68,69,70,71],richtet:1,risk:11,rm:[14,76],rms_norm:[64,65,67,68],ro_ro:6,roberta:11,root:[0,64,65,67,68],rope:[62,64,65],rotari:[64,65],rotary_bas:[64,65],rotary_dim:[64,65],rotary_interleav:[64,65],rotary_scaling_factor:[64,65],rotary_scaling_typ:[64,65],rotaryscalingtyp:[56,64,65,72],round:73,rui:[14,73],run:[2,5,6,11,14,16,24,29,30,34,49,50,51,73,74,76],rund:1,runtim:[3,14,24,29,34,49,50,51,73],runtimeerror:[37,38,39,40,41,42,43,44,51],s3:74,s:[1,3,5,6,8,11,29,34,74],sacremos:6,sai:[1,6],same:[0,3,5,8,16,24,29,32,59,60,61,63,64,65,66,67,68,69,70,71,73],sampl:[6,29,34,51],sampling_r:11,sampling_temperatur:[1,5,11,29,34,51],sampling_topk:[1,5,6,11,29,34,51],sampling_topp:[29,34],santacod:11,save:[0,11,16,34,37,38,39,40,41,42,43,44,59,61,63,64,66,68,69,70,71,73],scale:[11,44,62,64,65,73],scale_alibi:[64,65],scenario:4,schedul:11,scheme:[37,38,39,40,41,42,43,44,59,60,61,63,64,65,66,67,68,69,70,71],schlussfolgerung:1,scientist:11,scope:0,score:[6,17,27,29,31,33,34,51,52,76],score_batch:[5,6,29,34],score_fil:34,score_iter:[29,34],score_token_0:34,score_token_1:34,scoringopt:77,scoringresult:[18,20,29,34,72,77],scratch:11,script:[0,8,16],search:[29,34,51],second:11,section:[0,1,74],see:[0,1,2,3,4,6,8,9,11,12,13,14,16,44,73,74,75,76],seed:55,seem:35,select:[3,9,11,12,14,65,68,73],self:[24,29,34,49,50,51,65,67,68],semant:77,sens:0,sentenc:[1,6,17,34,74],sentencepiec:[5,6,10,74],sentencepieceprocessor:[5,6,10,74],separ:[3,5,10,14,16,73,76],sequenc:[1,2,5,6,24,27,29,34,51,52,63,64,65],sequences_id:[11,27,52],sequencetosequencemodelspec:[56,61,68,72],serv:[1,4],server:0,servic:1,servieren:1,session:[5,11],set:[1,3,6,12,14,16,17,24,29,34,37,38,39,40,41,42,43,44,47,49,50,51,54,55,59,60,61,63,64,65,66,67,68,69,70,71,73,76],set_log_level:[3,18,72,73],set_random_se:[18,72],setup:[14,16],sever:[4,5],shape:[24,29,32,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71],share:[4,16,32,64,65,73],shared_layer_norm:[64,65],shock:11,should:[0,3,5,6,9,11,14,16,29,42,60,74,76],show:[5,6],shown:11,shutoff:11,sich:1,sigmoid:57,similar:[0,4,6,11,73],simpl:0,simpli:[1,8,9,15],simplic:6,simplifi:29,sinc:[10,11],singl:[3,6,14,17,28],size:[4,5,16,17,29,34,51,59,63,64,65,66,68,69,70,71,73],skip:17,skip_special_token:11,sky:6,slave:16,slice:[64,65],sliding_window:[64,65],slot:[16,24,29,34,49,50,51],slower:16,small:[1,11],smallest:11,smooth:[11,44],smoothquant:[11,44],so:[11,12,16,17,29,32,34,73,74],softmax:[16,17],solut:6,some:[3,4,6,11,14,15,16,17,26,29,34,44,59,60,61,63,64,65,66,67,68,69,70,71,76],sort:[17,29,34],sourc:[5,6,10,11,12,34,38,63,67,68],source_lang:38,source_path:34,source_tokenize_fn:34,source_vocabulari:0,sp:[5,6,10,74],space:[0,5],spec:[57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72],special:[6,29,38,76],specif:[3,4,11,16,32,59,60,61,63,64,65,66,67,68,69,70,71,73],specifi:[67,68],spectogram:[49,50,51],speech:[11,13,14,49,50,51,52],speed:[1,3,76],split:[16,17,29,34],spm:[5,6,10,74],squar:[64,65,67,68],sr:11,src:7,src_1:76,src_2:76,src_lang:11,src_n:76,sse:12,ssh:16,stabilityai:5,stablelm:5,standard:1,standardisierten:1,star:11,starcod:11,starcoder_ct2:11,start:[1,5,6,8,11,29,38,51,74],start_sequ:51,start_token:[11,29],startoftranscript:11,startswith:5,state:[5,11,29],static_prompt:[5,17,29],statist:[26,34],std:[74,76],step:[0,4,14,28,73],step_result:5,still:[0,5,11],stop:[5,29,34],storag:32,storageview:[11,18,24,29,49,50,51,72],store:73,stori:5,str:[5,6,10,24,29,34,35,37,38,39,40,41,42,43,44,47,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,74],strategi:[1,58],stream:[16,17,29,34],string:[1,24,28,29,51,74,76],strip:34,strong:4,stronger:1,structur:[26,29,34],style:11,su:[62,64,65],sub:16,sublay:65,submiss:17,submit:16,submodul:14,successfulli:74,sudo:14,suffix:11,suggest:73,summar:11,supervis:11,support:[3,4,6,7,8,9,11,13,16,17,47,51,76,77],suppress:51,suppress_blank:51,suppress_sequ:[29,34],suppress_token:51,sure:[8,11,16,74],swish:57,symbol:[1,51,77],symmetr:73,syria:6,system:[0,5,14,16,17,29,73],system_prompt:5,system_prompt_token:5,t5:[65,67,68],t5v1:11,t:[9,11,48],tab:76,tabl:73,take:[1,11],talk:11,tanh:[57,66],tar:[6,74],target:[1,6,11,14,34,38,63,68,76],target_detokenize_fn:34,target_lang:38,target_link_librari:74,target_path:34,target_prefix:[1,6,11,34],target_tokenize_fn:34,target_vocabulari:0,task:[6,11,17],team:6,techniqu:[1,73],temperatur:[29,34,51],templat:[0,4],temporarili:15,tensor:[17,24,29,32,34,49,50,51],tensor_parallel:[16,24,29,34,49,50,51],tensorflow:0,test:14,text:[1,6,11,13,24,29,34,51,74,75],text_token:51,textattack:11,tf:[0,13,42],tgt:7,tgt_1:76,tgt_2:76,tgt_k:76,tgt_lang:11,than:[0,5,6,17,29,34],theblok:73,thei:[0,3,5,6,10,11,12,29,34],them:[11,14,44],thi:[0,1,3,5,6,10,11,12,14,15,16,19,20,21,24,28,29,34,37,38,39,40,41,42,43,44,49,50,51,53,54,59,60,61,63,64,65,66,67,68,69,70,71,73,74,76],thousand:11,thread:[4,14,16,17,24,29,34,49,50,51],thread_pool:16,through:11,throughput:16,tii:11,tiiuae:11,time:[1,5,16,26,59,61,63,64,66,68,69,70,71],timestamp:[11,51],timestep:1,tini:11,tip:13,to_cpu:[24,29,34,49,50,51],to_devic:32,todai:5,token:[1,6,8,17,24,26,27,28,29,31,34,38,51,52,59,63,64,66,68,69,70,71,73,74,76],token_id:[5,28],token_type_id:24,tokenis:74,tokenize_fn:34,tokenizer_class:44,tomorrow:11,too:16,toolkit:[4,14],top:[0,29,34,51,61],torch:[11,32,44],torch_dtyp:44,total:[17,26],total_time_in_m:26,toward:[1,34],trace:3,track:0,tradeoff:3,train:[0,4,6,7,8,9,10,11,38,39,43,74],transcrib:11,transcript:11,transform:[0,2,4,5,6,7,8,9,10,13,29,44,64,65,66,67,68,73,74,75],transformer_align:6,transformer_lm:6,transformerdecodermodelspec:[56,59,72],transformerdecoderspec:[56,60,64,68,72],transformerencodermodelspec:[56,59,72],transformerencoderspec:[56,60,66,68,72],transformersconvert:[36,37,72],transformerspec:[56,63,72],translat:[1,5,6,10,13,14,15,16,18,33,48,72,73,74,77],translate_batch:[1,6,10,11,16,34,74,76],translate_fil:34,translate_iter:34,translationopt:[34,77],translationresult:[18,21,34,72,76,77],tree:[49,50],tri:73,truncat:[1,29,34],trust_remote_cod:[11,44],tune:[5,11],tupl:[51,68],txt:[6,14,34,74],type:[8,24,29,32,34,47,49,50,51,57,62,64,65,74],typic:73,ubuntu20:[14,76],um:1,un:6,uncas:11,uncondit:11,unconstrain:[29,34],und:1,unifi:0,union:[24,29,34,42,49,50,51,68,76],unit:[64,65,67,68],unknown:[29,34],unlik:1,unlimit:[5,24,29,34,49,50,51],unload:[5,15,24,29,34,49,50,51],unload_model:[15,24,29,34,49,50,51],unscal:29,unspecifi:1,unsupervis:11,unsupport:32,until:[16,19,20,21,24,29,34,49,50,51,53],unzip:10,up:[11,29],updat:[0,14],us:[0,1,2,3,5,6,8,10,11,12,14,15,16,17,24,29,30,32,34,42,44,49,50,51,59,61,63,64,65,66,67,68,69,70,71,73,74,76],usag:[1,3,4,11,13,73,76],use_vmap:[34,76],user:[5,38],user_dir:38,usual:[3,5,14,73],v:76,valid:[12,59,60,61,63,64,65,66,67,68,69,70,71],valu:[0,1,3,4,14,16,22,23,24,28,29,34,37,38,39,40,41,42,43,44,49,50,51,59,60,61,63,64,65,66,67,68,69,70,71,73],valueerror:[32,59,60,61,63,64,65,66,67,68,69,70,71],variabl:[0,12,13,14,16,17,29,59,60,61,63,64,65,66,67,68,69,70,71,73],variant:11,variou:74,vector:[34,74,76],vendor:3,verbos:3,veri:[1,16],verifi:[59,60,61,63,64,65,66,67,68,69,70,71],version:[0,5,11,12,13,14,74],via:[4,76],view:32,virtual_env:16,virtual_env_prompt:16,visibl:45,visual:14,vmap:[37,38,39,40,41,42,43,44,76],vocab:[7,28],vocab_map:76,vocab_path:[7,39],vocabulari:[0,6,7,17,34,37,38,39,40,41,42,43,44,59,63,64,66,68,69,70,71,73],volum:17,von:1,w:73,wa:[11,19,20,21,27,28,33,52,53,73],wai:[4,29],wait:[24,29,34,49,50,51],want:[1,4,9,15,16],warn:3,wav2vec2:[48,72],wav2vec2bert:[48,72],wav2vec2bertspec:[56,59,72],wav2vec2spec:[56,59,72],wav2vec:[49,50],wav:11,we:[1,16,73],weight:[0,4,11,16,34,37,38,39,40,41,42,43,44,59,60,61,63,64,65,66,67,68,69,70,71,73],welcom:13,well:76,welt:74,were:[0,4,11],wget:[6,10,74],what:[5,74,76],whatev:73,wheel:[14,16],when:[0,1,3,5,11,14,16,17,19,20,21,24,29,34,44,49,50,51,53,64,65,67,68,73],whenev:17,where:[0,4,6,12,24,29,32,34,37,38,39,40,41,42,43,44,49,50,51,59,61,63,64,66,68,69,70,71,76],wherea:12,whether:[3,24,28,29,34,42,49,50,51],which:[0,2,3,4,5,6,11,14,16,17,29,34,48,51,73,74,76],whisper:[48,52,71,72,75],whisperalignmentresult:51,whispergenerationresult:[48,51,53,72],whispergenerationresultasync:[48,51,72],whisperprocessor:11,whisperspec:[56,59,72],whl:14,whose:[17,29,34],why:1,width:51,wildfir:11,wind:11,window:[11,14],with_acceler:14,with_cuda:14,with_cudnn:14,with_dnnl:14,with_encoder_attent:65,with_mkl:14,with_openbla:14,with_relative_posit:[64,68],with_rui:14,with_scor:34,with_tokens_scor:34,without:[0,4,32,73],wmt19:5,wmt19_en_ct2:6,wmt:74,wngt:17,wonder:11,word:5,work:[3,73],worker:[16,29,34,49,50,51],workload:16,workshop:11,world:[6,10,11,24,34,74,76],worst:[11,51],would:0,wq:73,wrapper:[19,20,21,53],write:[0,5,74],wu:73,x86:[12,14,73],x:[6,14,16,32],xf:[6,74],xlm:11,y:32,yaml:[9,42],yelp:11,yield:[29,34],yml:[7,9,39],you:[0,1,3,5,8,9,11,12,14,15,16,17,73,74],your:[0,4,5,14,34,73,74],z:32,zero:73,zip:10,zu:1,zur:1,zwar:1},titles:["Model conversion","Decoding features","Text encoding","Environment variables","Frequently asked questions","Text generation","Fairseq","Marian","OpenNMT-py","OpenNMT-tf","OPUS-MT","Transformers","Hardware support","Index","Installation","Memory management","Multithreading and parallelism","Performance tips","ctranslate2","AsyncGenerationResult","AsyncScoringResult","AsyncTranslationResult","DataType","Device","Encoder","EncoderForwardOutput","ExecutionStats","GenerationResult","GenerationStepResult","Generator","MpiInfo","ScoringResult","StorageView","TranslationResult","Translator","contains_model","ctranslate2.converters","Converter","FairseqConverter","MarianConverter","OpenAIGPT2Converter","OpenNMTPyConverter","OpenNMTTFConverter","OpusMTConverter","TransformersConverter","get_cuda_device_count","get_log_level","get_supported_compute_types","ctranslate2.models","Wav2Vec2","Wav2Vec2Bert","Whisper","WhisperGenerationResult","WhisperGenerationResultAsync","set_log_level","set_random_seed","ctranslate2.specs","Activation","EmbeddingsMerge","LanguageModelSpec","LayerSpec","ModelSpec","RotaryScalingType","SequenceToSequenceModelSpec","TransformerDecoderModelSpec","TransformerDecoderSpec","TransformerEncoderModelSpec","TransformerEncoderSpec","TransformerSpec","Wav2Vec2BertSpec","Wav2Vec2Spec","WhisperSpec","Python","Quantization","Quickstart","Speech recognition","Text translation","Versioning"],titleterms:{"100":[6,11],"16":73,"2":11,"4":73,"50":6,"8":73,"do":4,"float":73,"new":0,activ:57,add:0,altern:1,api:13,ar:4,ask:4,asyncgenerationresult:19,asynchron:16,asyncscoringresult:20,asynctranslationresult:21,autocomplet:1,awq:73,backward:0,bart:11,beam:[1,6],bert:11,bfloat16:73,bias:1,bit:73,bloom:11,brain:73,build:14,c:[14,74,76],cach:5,client:76,code:14,compat:0,compil:14,constraint:1,contains_model:35,convers:[0,73],convert:[0,36,37],cpu:[12,16,17],ct2_cuda_alloc:3,ct2_cuda_allow_bf16:3,ct2_cuda_allow_fp16:3,ct2_cuda_caching_allocator_config:3,ct2_cuda_true_fp16_gemm:3,ct2_force_cpu_isa:3,ct2_use_experimental_packed_gemm:3,ct2_use_mkl:3,ct2_verbos:3,ctranslat:4,ctranslate2:[18,36,48,56],data:16,datatyp:22,decod:1,develop:13,devic:23,distilbert:11,docker:14,doe:4,download:14,dynam:76,embeddingsmerg:58,encod:[2,24],encoderforwardoutput:25,english:6,environ:3,equival:6,exampl:[10,76],execut:16,executionstat:26,factor:76,fairseq:6,fairseqconvert:38,falcon:11,featur:1,float16:73,framework:[0,13],frequent:4,from:[14,15],futur:4,gener:[5,8,17,29],generationresult:27,generationstepresult:28,german:6,get:13,get_cuda_device_count:45,get_log_level:46,get_supported_compute_typ:47,gpt:11,gptbigcod:11,gpu:[12,14,17],greedi:1,guid:13,hardwar:12,how:4,i:4,implement:4,implicit:73,index:13,inform:13,instal:14,instead:4,int16:73,int8:73,integ:73,intra:16,j:11,known:4,languag:6,languagemodelspec:59,layerspec:60,length:1,librari:14,limit:4,llama:11,load:73,m2m:[6,11],manag:15,marian:7,marianconvert:39,marianmt:11,mbart:6,memori:15,model:[0,6,16,48,73],modelspec:61,mpiinfo:30,mpt:11,mt:10,multithread:16,neox:11,nllb:11,op:16,openaigpt2convert:40,opennmt:[8,9],opennmtpyconvert:41,opennmttfconvert:42,opt:11,option:14,opu:10,opusmtconvert:43,origin:4,other:13,parallel:16,perform:17,pip:14,plan:4,point:73,portabl:0,posit:1,precis:0,project:4,prompt:5,provid:4,py:8,python:[14,15,72,74,76],pytorch:4,quantiz:[0,73],question:4,quickstart:74,random:1,recognit:75,reduc:0,reduct:76,relat:4,rotaryscalingtyp:62,sampl:1,scoringresult:31,search:[1,6],sequencetosequencemodelspec:63,serial:0,server:4,set_log_level:54,set_random_se:55,should:4,sourc:[14,76],spec:56,special:[5,11],specif:0,speech:75,start:13,storageview:32,stream:5,structur:0,support:[0,12,14,73],t5:11,task:13,tensor:16,tensorflow:4,text:[2,5,8,76],tf:9,thi:4,tip:17,token:[5,11],transform:11,transformer_lm:8,transformerdecodermodelspec:64,transformerdecoderspec:65,transformerencodermodelspec:66,transformerencoderspec:67,transformersconvert:44,transformerspec:68,translat:[4,11,17,34,76],translationresult:33,type:73,us:4,variabl:3,version:77,vocabulari:76,wav2vec2:49,wav2vec2bert:50,wav2vec2bertspec:69,wav2vec2spec:70,what:4,when:4,whisper:[11,51],whispergenerationresult:52,whispergenerationresultasync:53,whisperspec:71,why:4,wmt16:6,wmt19:6,wrapper:14,you:4}}) \ No newline at end of file diff --git a/speech_recognition.html b/speech_recognition.html index 8df9475c7..8054a719c 100644 --- a/speech_recognition.html +++ b/speech_recognition.html @@ -4,7 +4,7 @@ - Speech recognition — CTranslate2 4.4.0 documentation + Speech recognition — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    diff --git a/translation.html b/translation.html index 1066c0208..ed8797841 100644 --- a/translation.html +++ b/translation.html @@ -4,7 +4,7 @@ - Text translation — CTranslate2 4.4.0 documentation + Text translation — CTranslate2 4.5.0 documentation @@ -32,7 +32,7 @@ CTranslate2
    - 4.4 + 4.5
    diff --git a/versioning.html b/versioning.html index 09efc3866..69cf28510 100644 --- a/versioning.html +++ b/versioning.html @@ -4,7 +4,7 @@ - Versioning — CTranslate2 4.4.0 documentation + Versioning — CTranslate2 4.5.0 documentation @@ -31,7 +31,7 @@ CTranslate2
    - 4.4 + 4.5