Skip to content

[Flax] Fix & Add scripts to push pretrained flax model weights to the huggingface hub #545

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

[Flax] Fix & Add scripts to push pretrained flax model weights to the huggingface hub #545

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f841d59
fix pretrained parameter assignment in text classification with flax
snoop2head Jan 1, 2025
190ff5b
add pretrained LM parameter uploading code to huggingface hub
snoop2head Jan 1, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
165 changes: 105 additions & 60 deletions examples/causal_language_modeling_flax.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1550,9 +1550,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (1/10 | Loss: 6.935000419616699, Learning Rate: 0.0002699999895412475)\n"
]
},
Expand All @@ -1576,9 +1576,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (1/10 | Loss: 7.108445644378662 | Perplexity: 1246.529052734375)\n"
]
},
Expand All @@ -1602,9 +1602,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (2/10 | Loss: 6.334000110626221, Learning Rate: 0.00023999999393709004)\n"
]
},
Expand All @@ -1628,9 +1628,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (2/10 | Loss: 6.567610740661621 | Perplexity: 738.8753662109375)\n"
]
},
Expand All @@ -1654,9 +1654,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (3/10 | Loss: 5.798000335693359, Learning Rate: 0.0002099999983329326)\n"
]
},
Expand All @@ -1680,9 +1680,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (3/10 | Loss: 6.278167247772217 | Perplexity: 557.9488525390625)\n"
]
},
Expand All @@ -1706,9 +1706,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (4/10 | Loss: 5.557000160217285, Learning Rate: 0.00018000000272877514)\n"
]
},
Expand All @@ -1732,9 +1732,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (4/10 | Loss: 6.062875270843506 | Perplexity: 451.3289794921875)\n"
]
},
Expand All @@ -1758,9 +1758,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (5/10 | Loss: 5.543000221252441, Learning Rate: 0.00014999999257270247)\n"
]
},
Expand All @@ -1784,9 +1784,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (5/10 | Loss: 5.920379161834717 | Perplexity: 392.97332763671875)\n"
]
},
Expand All @@ -1810,9 +1810,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (6/10 | Loss: 5.361000061035156, Learning Rate: 0.00011999999696854502)\n"
]
},
Expand All @@ -1836,9 +1836,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (6/10 | Loss: 5.821027755737305 | Perplexity: 356.4353942871094)\n"
]
},
Expand All @@ -1862,9 +1862,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (7/10 | Loss: 5.207000255584717, Learning Rate: 9.000000136438757e-05)\n"
]
},
Expand All @@ -1888,9 +1888,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (7/10 | Loss: 5.748736381530762 | Perplexity: 332.1453857421875)\n"
]
},
Expand All @@ -1914,9 +1914,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (8/10 | Loss: 5.124000072479248, Learning Rate: 5.999999848427251e-05)\n"
]
},
Expand All @@ -1940,9 +1940,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (8/10 | Loss: 5.703180313110352 | Perplexity: 317.5106201171875)\n"
]
},
Expand All @@ -1966,9 +1966,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (9/10 | Loss: 5.220000267028809, Learning Rate: 2.9999999242136255e-05)\n"
]
},
Expand All @@ -1992,9 +1992,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (9/10 | Loss: 5.674434185028076 | Perplexity: 308.7478942871094)\n"
]
},
Expand All @@ -2018,9 +2018,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Train... (10/10 | Loss: 4.992000102996826, Learning Rate: 0.0)\n"
]
},
Expand All @@ -2044,9 +2044,9 @@
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"\r",
"\r",
"\r\n",
"\r\n",
"\r\n",
"Eval... (10/10 | Loss: 5.66389274597168 | Perplexity: 305.58953857421875)\n",
"\n"
]
Expand Down Expand Up @@ -2098,6 +2098,51 @@
"\n",
"For a more in-detail comparison of runtimes please refer to [this](https://github.com/huggingface/transformers/tree/master/examples/flax/language-modeling#runtime-evaluation) table."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"You may upload the model to the huggingface hub. The model will be uploaded under `https://huggingface.co/<your-username>/<model_dir>`:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"unreplicated_params = flax.jax_utils.unreplicate(state.params)\n",
"model.params = jax.device_get(jax.tree_map(lambda x: x.astype(jnp.float32), unreplicated_params))\n",
"\n",
"model.push_to_hub(model_dir)\n",
"tokenizer.push_to_hub(model_dir)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"After uploading, you may use the model to generate a text using pipeline"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Use a pipeline as a high-level helper\n",
"from transformers import pipeline\n",
"\n",
"your_username = \"\" # huggingface user name\n",
"pipe = pipeline(\"text-generation\", model=f\"{your_username}/{model_dir}\")\n",
"\n",
"sample_prompt = \"Týðingin verður løgd til almennar\" # sample inputs\n",
"\n",
"# decoding with beam search of 5\n",
"pipe(sample_prompt, max_new_tokens=20, num_beams=5)"
]
}
],
"metadata": {
Expand Down
3 changes: 3 additions & 0 deletions examples/text_classification_flax.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1481,6 +1481,9 @@
},
"outputs": [],
"source": [
"unreplicated_params = flax.jax_utils.unreplicate(state.params)\n",
"model.params = jax.device_get(jax.tree_map(lambda x: x.astype(jnp.float32), unreplicated_params))\n",
"\n",
"model.push_to_hub(model_id, use_auth_token=hf_auth_token)\n",
"tokenizer.push_to_hub(model_id, use_auth_token=hf_auth_token)"
]
Expand Down