diff --git a/requirements.txt b/requirements.txt index 3e71feab0..fbc39d9bb 100644 --- a/requirements.txt +++ b/requirements.txt @@ -2,5 +2,5 @@ ray>=2.48.0 pandas>=2.2.3 numba>=0.58.0 -numpy>=1.26.0 +numpy>=1.26.0,<=2.3 transformers>=4.1,<4.56.0 diff --git a/vllm_gaudi/extension/features.py b/vllm_gaudi/extension/features.py index 719e9abd0..649f50d99 100644 --- a/vllm_gaudi/extension/features.py +++ b/vllm_gaudi/extension/features.py @@ -104,7 +104,7 @@ def get_features(): All(VersionRange(">=1.24.0.171"), MinPackageVersion("neural_compressor_pt", "3.7")), env_var_type=boolean), Value('per_token_kv_scaling_support', - All(VersionRange(">=1.24.0.350"), MinPackageVersion("neural_compressor_pt", "3.7")), + All(VersionRange(">=1.24.0.350"), MinPackageVersion("neural_compressor_pt", "3.4")), env_var_type=boolean), Value('moe_chunk', "", env_var='VLLM_MOE_CHUNK', env_var_type=list_of(int)), Value('moe_token_boundary', "", env_var='VLLM_MOE_TOKEN_BOUNDARY', env_var_type=list_of(int)),