ipex-llm 2.3.0b20250427__py3-none-manylinux2010_x86_64.whl → 2.3.0b20250501__py3-none-manylinux2010_x86_64.whl

ipex_llm/vllm/xpu/model_convert.py

@@ -48,7 +48,7 @@ def _sample_get_logits(
         logits = lm_head(hidden_states)
         if embedding_bias is not None:
             logits += embedding_bias
-    if self.use_gather:
+    if self.use_all_gather:
         logits = tensor_model_parallel_gather(logits)
     else:
         logits = tensor_model_parallel_all_gather(logits)
@@ -63,6 +63,8 @@ def _model_sample_convert():
 
 
 def _ipex_llm_convert(load_in_low_bit):
+    # import pdb
+    # pdb.set_trace()
     from vllm.worker.xpu_model_runner import XPUModelRunner
     from ipex_llm.vllm.xpu.ipex_llm_wrapper import get_ipex_llm_wrapper
     from ipex_llm.vllm.xpu.ipex_llm_v1_wrapper import get_ipex_llm_v1_wrapper
@@ -99,7 +101,8 @@ def get_load_function(low_bit):
                         "codegeex4-all" in self.vllm_config.model_config.model.lower() or
                         "chatglm" in self.vllm_config.model_config.model.lower()) and \
                         "gptq" not in self.model_config.model.lower() and \
-                        "awq" not in self.model_config.model.lower():
+                        "awq" not in self.model_config.model.lower() and \
+                        "qwen3" not in self.model_config.model.lower():
                     self.model.apply(padding_mlp)
                 from ipex_llm import optimize_model
                 not_convert_last_mlp = os.getenv("IPEX_LLM_NOT_CONVERT_LAST_MLP", None)

{ipex_llm-2.3.0b20250427.dist-info → ipex_llm-2.3.0b20250501.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ipex-llm
-Version: 2.3.0b20250427
+Version: 2.3.0b20250501
 Summary: Large Language Model Develop Toolkit
 Home-page: https://github.com/intel-analytics/ipex-llm
 Author: BigDL Authors
@@ -27,7 +27,7 @@ Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine
 Requires-Dist: torch ==2.1.2+cpu ; (platform_system == "Linux") and extra == 'all'
 Requires-Dist: torch ==2.1.2 ; (platform_system == "Windows") and extra == 'all'
 Provides-Extra: cpp
-Requires-Dist: bigdl-core-cpp ==2.7.0b20250427 ; extra == 'cpp'
+Requires-Dist: bigdl-core-cpp ==2.7.0b20250501 ; extra == 'cpp'
 Requires-Dist: setuptools ; extra == 'cpp'
 Requires-Dist: onednn-devel ==2025.0.1 ; (platform_system == "Windows") and extra == 'cpp'
 Requires-Dist: onednn ==2025.0.1 ; (platform_system == "Windows") and extra == 'cpp'
@@ -60,7 +60,7 @@ Requires-Dist: transformers ==4.40.0 ; extra == 'npu'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'npu'
 Requires-Dist: torch ==2.1.2+cpu ; (platform_system == "Linux") and extra == 'npu'
 Requires-Dist: torch ==2.1.2 ; (platform_system == "Windows") and extra == 'npu'
-Requires-Dist: bigdl-core-npu ==2.7.0b20250427 ; (platform_system == "Windows") and extra == 'npu'
+Requires-Dist: bigdl-core-npu ==2.7.0b20250501 ; (platform_system == "Windows") and extra == 'npu'
 Provides-Extra: serving
 Requires-Dist: py-cpuinfo ; extra == 'serving'
 Requires-Dist: fschat[model_worker,webui] ==0.2.36 ; extra == 'serving'
@@ -80,9 +80,9 @@ Requires-Dist: setuptools <70.0.0 ; extra == 'xpu'
 Requires-Dist: torch ==2.1.0a0 ; extra == 'xpu'
 Requires-Dist: torchvision ==0.16.0a0 ; extra == 'xpu'
 Requires-Dist: intel-extension-for-pytorch ==2.1.10+xpu ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-21 ==2.7.0b20250427 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.7.0b20250427 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.7.0b20250427 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-21 ==2.7.0b20250501 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.7.0b20250501 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.7.0b20250501 ; extra == 'xpu'
 Provides-Extra: xpu-2-1
 Requires-Dist: py-cpuinfo ; extra == 'xpu-2-1'
 Requires-Dist: protobuf ; extra == 'xpu-2-1'
@@ -97,9 +97,9 @@ Requires-Dist: setuptools <70.0.0 ; extra == 'xpu-2-1'
 Requires-Dist: torch ==2.1.0a0 ; extra == 'xpu-2-1'
 Requires-Dist: torchvision ==0.16.0a0 ; extra == 'xpu-2-1'
 Requires-Dist: intel-extension-for-pytorch ==2.1.10+xpu ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-21 ==2.7.0b20250427 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.7.0b20250427 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.7.0b20250427 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-21 ==2.7.0b20250501 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.7.0b20250501 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.7.0b20250501 ; extra == 'xpu-2-1'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'xpu-2-1'
 Requires-Dist: dpcpp-cpp-rt ==2024.0.2 ; (platform_system == "Windows") and extra == 'xpu-2-1'
 Requires-Dist: mkl-dpcpp ==2024.0.0 ; (platform_system == "Windows") and extra == 'xpu-2-1'
@@ -117,7 +117,7 @@ Requires-Dist: setuptools ; extra == 'xpu-2-6'
 Requires-Dist: torch ==2.6.0+xpu ; extra == 'xpu-2-6'
 Requires-Dist: torchvision ==0.21.0+xpu ; extra == 'xpu-2-6'
 Requires-Dist: torchaudio ==2.6.0+xpu ; extra == 'xpu-2-6'
-Requires-Dist: bigdl-core-xe-all ==2.7.0b20250427 ; extra == 'xpu-2-6'
+Requires-Dist: bigdl-core-xe-all ==2.7.0b20250501 ; extra == 'xpu-2-6'
 Requires-Dist: onednn-devel ==2025.0.1 ; extra == 'xpu-2-6'
 Requires-Dist: onednn ==2025.0.1 ; extra == 'xpu-2-6'
 Requires-Dist: dpcpp-cpp-rt ==2025.0.2 ; extra == 'xpu-2-6'
@@ -132,7 +132,7 @@ Requires-Dist: tokenizers ==0.15.2 ; extra == 'xpu-2-6-arl'
 Requires-Dist: accelerate ==0.23.0 ; extra == 'xpu-2-6-arl'
 Requires-Dist: tabulate ; extra == 'xpu-2-6-arl'
 Requires-Dist: setuptools ; extra == 'xpu-2-6-arl'
-Requires-Dist: bigdl-core-xe-all ==2.7.0b20250427 ; extra == 'xpu-2-6-arl'
+Requires-Dist: bigdl-core-xe-all ==2.7.0b20250501 ; extra == 'xpu-2-6-arl'
 Requires-Dist: onednn-devel ==2025.0.1 ; extra == 'xpu-2-6-arl'
 Requires-Dist: onednn ==2025.0.1 ; extra == 'xpu-2-6-arl'
 Requires-Dist: dpcpp-cpp-rt ==2025.0.2 ; extra == 'xpu-2-6-arl'

{ipex_llm-2.3.0b20250427.dist-info → ipex_llm-2.3.0b20250501.dist-info}/RECORD

@@ -94,7 +94,7 @@ ipex_llm/serving/fastchat/tgi_api_protocol.py,sha256=brT3k3-V0NJrU4fRqUwWjC0O3iO
 ipex_llm/serving/fastchat/tgi_api_server.py,sha256=agNTAEiZPSuj3dEdIdYKwkoY0cXOUDX06DiM9VP2knQ,24418
 ipex_llm/serving/fastchat/vllm_worker.py,sha256=ZLz2Q9GxJO6r_LOiP6epgCRjBGk-K4EB1SNEWSJp5DA,11091
 ipex_llm/transformers/__init__.py,sha256=BreA3EY6hkNq0rVixb_sUuTLzMrcWXTt3yRsshCPHHQ,1214
-ipex_llm/transformers/convert.py,sha256=niCLZiEQbiMz8UkfYDNwYl2hE1caj93FUR1qjgVh9ZM,107538
+ipex_llm/transformers/convert.py,sha256=fmunTuZCTgTnZZXfDaCNHlO4TMDBGUgqPRKuOLp7x9Y,107586
 ipex_llm/transformers/convert_ipex.py,sha256=_nSnUTQy-yfkKaqGdqnBdWztZf3NGmnbZ0TKaDrF4X4,14617
 ipex_llm/transformers/embedding.py,sha256=bdgk59DvD4ZZyxRzewXOR7g56nThgO6uhIwk8QL7f-s,9299
 ipex_llm/transformers/kv.py,sha256=src_HcVDKFwQ1V8hdTrFQw5RIwUewM9VOR47GVTPJG4,21187
@@ -258,16 +258,16 @@ ipex_llm/vllm/cpu/entrypoints/openai/cli_args.py,sha256=hB398yYtKauASRzevctScdbF
 ipex_llm/vllm/xpu/__init__.py,sha256=zBSG6nzrVF5QnpR6_f7kPhBFeowTE9gaZ7D5m98E7_w,585
 ipex_llm/vllm/xpu/ipex_llm_v1_wrapper.py,sha256=pd939vFomKIg9Qn2NO4u0OF6hPgvQpqcfJSxqBzcqhA,825
 ipex_llm/vllm/xpu/ipex_llm_wrapper.py,sha256=_CbhvBuf_KPnmLfngYKtJl5gPAHVsG2mWth3wSeaH3M,892
-ipex_llm/vllm/xpu/model_convert.py,sha256=fhZ6QsoosWnzlPNMVdvz-XVVD1TMKWHpOGCNECn29Uw,9424
-ipex_llm/vllm/xpu/engine/__init__.py,sha256=pY_CpyuZd72fr6s32ejeKHKFW0K4vUU2rzZjWqm5YiA,783
-ipex_llm/vllm/xpu/engine/engine.py,sha256=NvCMbp0X8NVrOqbwm4FTvXOptTRLzu9jQsy37ZHnTk8,9493
-ipex_llm/vllm/xpu/entrypoints/openai/api_server.py,sha256=IjiSze9vzBCAkLu_VwIcJwuO1jyFna7DLrj6aSL7RaQ,35220
+ipex_llm/vllm/xpu/model_convert.py,sha256=lfPntjW0zD3fLvBOlaGdUj6Jj5HH06cotlfYqCvnXt0,9544
+ipex_llm/vllm/xpu/engine/__init__.py,sha256=sOvwLx_Zj0jiRCGj9W3DgGTfcSU3hABYhgIQI7T6cxU,879
+ipex_llm/vllm/xpu/engine/engine.py,sha256=XAprw7VifjfnR915TZOaKcxe3QCFsVBgxzS8qOdn1yg,14462
+ipex_llm/vllm/xpu/entrypoints/openai/api_server.py,sha256=uWHyyHLw-B8wXBnQw9_MCG81tKK9Jb0dyq1xfYHgoNw,45905
 ipex_llm/vllm/xpu/entrypoints/openai/cli_args.py,sha256=hB398yYtKauASRzevctScdbFIjiiSGMAe1bwEuIHrhY,10893
-ipex_llm-2.3.0b20250427.data/scripts/ipex-llm-init,sha256=fLQsT2dRL6H5bThb4GuIWotAuqoLsIxFwA-0c2qmaO8,6672
-ipex_llm-2.3.0b20250427.data/scripts/llm-chat,sha256=TdUnUmNapzuoe1c8IzrdVOQwWEg8IqsMSBRlOD3daZM,2249
-ipex_llm-2.3.0b20250427.data/scripts/llm-cli,sha256=RXGPlLElHxcKzoUxljEMBIAXbzCDysXL-Nxw-xF-7LU,2457
-ipex_llm-2.3.0b20250427.dist-info/METADATA,sha256=kVIebSEkjI0G5kzPWUE_jctdevrjOh9pOokBbm4vp5E,8865
-ipex_llm-2.3.0b20250427.dist-info/WHEEL,sha256=PPJcBMAZibF_2GFE9NmOJGqiaSMPiNFbJd6QaJjdA6Y,109
-ipex_llm-2.3.0b20250427.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
-ipex_llm-2.3.0b20250427.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
-ipex_llm-2.3.0b20250427.dist-info/RECORD,,
+ipex_llm-2.3.0b20250501.data/scripts/ipex-llm-init,sha256=fLQsT2dRL6H5bThb4GuIWotAuqoLsIxFwA-0c2qmaO8,6672
+ipex_llm-2.3.0b20250501.data/scripts/llm-chat,sha256=TdUnUmNapzuoe1c8IzrdVOQwWEg8IqsMSBRlOD3daZM,2249
+ipex_llm-2.3.0b20250501.data/scripts/llm-cli,sha256=RXGPlLElHxcKzoUxljEMBIAXbzCDysXL-Nxw-xF-7LU,2457
+ipex_llm-2.3.0b20250501.dist-info/METADATA,sha256=_j0pOrs0d1xMDqileIZ5jm_HBlQCeviWiDeiybQ54-A,8865
+ipex_llm-2.3.0b20250501.dist-info/WHEEL,sha256=PPJcBMAZibF_2GFE9NmOJGqiaSMPiNFbJd6QaJjdA6Y,109
+ipex_llm-2.3.0b20250501.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
+ipex_llm-2.3.0b20250501.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
+ipex_llm-2.3.0b20250501.dist-info/RECORD,,