ipex-llm 2.2.0b20250114__py3-none-win_amd64.whl → 2.2.0b20250115__py3-none-win_amd64.whl

ipex_llm/ggml/quantize.py

@@ -53,6 +53,7 @@ ggml_tensor_qtype = {"sym_int4": 2,   # q4_0 in ggml
                      "sym_int4_rtn": 31,
                      "sym_int8_rtn": 32,
                      "asym_int4_rtn": 33,
+                     "woq_int4": 34,
                      }
 
 # mixed precison from llama.cpp

ipex_llm/transformers/convert.py

@@ -995,8 +995,9 @@ def _optimize_pre(model, qtype=None):
         from ipex_llm.transformers.models.gemma2 import merge_qkv
         model.apply(merge_qkv)
     elif model.config.model_type == "llama":
-        from ipex_llm.transformers.models.llama import merge_qkv
+        from ipex_llm.transformers.models.llama import merge_qkv, pre_compute_inv_freq
         model.apply(merge_qkv)
+        model.apply(pre_compute_inv_freq)
     elif model.config.model_type == "mllama":
         from ipex_llm.transformers.models.mllama import merge_qkv
         model.apply(merge_qkv)

ipex_llm/transformers/low_bit_linear.py

@@ -84,6 +84,7 @@ FP6_K = ggml_tensor_qtype["fp6_k"]
 SYM_INT4_RTN = ggml_tensor_qtype["sym_int4_rtn"]
 SYM_INT8_RTN = ggml_tensor_qtype["sym_int8_rtn"]
 ASYM_INT4_RTN = ggml_tensor_qtype["asym_int4_rtn"]
+WOQ_INT4 = ggml_tensor_qtype["woq_int4"]
 RTN_DTYPE = {
     SYM_INT4_RTN: torch.uint8,
     ASYM_INT4_RTN: torch.uint8,
@@ -187,7 +188,7 @@ def ggml_q_format_convet_cpu2xpu(tensor: torch.Tensor, num_elem: int, qtype: int
     src = ctypes.c_void_p(tensor.data.data_ptr())
 
     if qtype in [SYM_INT4, ASYM_INT4, SYM_INT8, NF4, NF3, FP4, FP6, FP8E4, FP8E5,
-                 Q4_K, Q6_K, FP6_K]:
+                 Q4_K, Q6_K, FP6_K, WOQ_INT4]:
         dst_tensor = torch.empty_like(tensor)
     elif qtype == ggml_tensor_qtype["sym_int5"]:
         QK = ggml.ggml_qk_size(qtype)
@@ -213,7 +214,7 @@ def ggml_q_format_convet_xpu2cpu(tensor: torch.Tensor, num_elem: int, qtype: int
     src = ctypes.c_void_p(tensor.data.data_ptr())
 
     if qtype in [SYM_INT4, ASYM_INT4, SYM_INT8, NF4, NF3, FP4, FP6, FP8E4, FP8E5,
-                 Q4_K, Q6_K, FP6_K]:
+                 Q4_K, Q6_K, FP6_K, WOQ_INT4]:
         dst_tensor = torch.empty_like(tensor)
     elif qtype == ggml_tensor_qtype["sym_int5"]:
         QK = ggml.ggml_qk_size(ggml_tensor_qtype["asym_int5"])

ipex_llm/transformers/models/llama.py

@@ -119,6 +119,13 @@ def merge_qkv(module: torch.nn.Module):
     merge_qkv_base(module, LlamaAttention)
 
 
+def pre_compute_inv_freq(module: torch.nn.Module):
+    if module.__class__.__name__ == "LlamaLinearScalingRotaryEmbedding":
+        if hasattr(module, "scaling_factor"):
+            module.register_buffer("inv_freq_scaled", None, persistent=False)
+            module.inv_freq_scaled = module.inv_freq / module.scaling_factor
+
+
 def llama_attention_forward(
     self,
     hidden_states: torch.Tensor,
@@ -147,8 +154,12 @@ def llama_attention_forward(
         import xe_addons
         if hasattr(self, "rotary_emb"):
             # transformers < 4.46
-            xe_addons.rotary_half_inplaced(self.rotary_emb.inv_freq, position_ids,
-                                           query_states, key_states)
+            if hasattr(self.rotary_emb, "inv_freq_scaled"):
+                xe_addons.rotary_half_inplaced(self.rotary_emb.inv_freq_scaled, position_ids,
+                                               query_states, key_states)
+            else:
+                xe_addons.rotary_half_inplaced(self.rotary_emb.inv_freq, position_ids,
+                                               query_states, key_states)
         else:
             # transformers >= 4.46
             cos, sin = position_embeddings

ipex_llm/transformers/models/utils.py

@@ -21,7 +21,7 @@ from ipex_llm.utils.common import invalidInputError
 from ipex_llm.ggml.quantize import ggml_tensor_qtype
 from ipex_llm.transformers.utils import get_xpu_device_name
 from ipex_llm.transformers.low_bit_linear import SYM_INT4, SYM_INT8, FP8E5, IQ2_XXS, FP4, FP8E4,\
-    FP6, ASYM_INT4
+    FP6, ASYM_INT4, WOQ_INT4
 
 FP8_KV_ALLOC_LENGTH = 512
 KV_CACHE_ALLOC_BLOCK_LENGTH = int(os.environ.get("KV_CACHE_ALLOC_BLOCK_LENGTH", 256))
@@ -33,7 +33,7 @@ GELU = 1
 
 def decoding_fast_path_qtype_check(proj):
     qtype = getattr(proj, "qtype", None)
-    return qtype in [SYM_INT4, FP8E5, FP4]
+    return qtype in [SYM_INT4, FP8E5, FP4, WOQ_INT4]
 
 
 def init_kv_cache(batch_size, num_heads, head_dim, current_length, max_length, dtype, device):
@@ -248,7 +248,7 @@ def mlp_fusion_check(x, qtype, training):
         return False
     if x.device.type != 'xpu':
         return False
-    if qtype not in [SYM_INT4, FP8E5, FP4, IQ2_XXS, FP6]:
+    if qtype not in [SYM_INT4, FP8E5, FP4, IQ2_XXS, FP6, WOQ_INT4]:
         return False
     if training or x.requires_grad:
         return False
@@ -263,7 +263,7 @@ def use_xmx(x: torch.Tensor, qtype: int):
     device = get_xpu_device_name(x.device)
     return (
         device in ["arc", "pvc"]
-        and qtype in [SYM_INT4, SYM_INT8, FP8E4, FP8E5]
+        and qtype in [SYM_INT4, SYM_INT8, FP8E4, FP8E5, WOQ_INT4]
         and (
             (device == "pvc" and 1 < x.size(0) <= 16)
             or

{ipex_llm-2.2.0b20250114.dist-info → ipex_llm-2.2.0b20250115.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ipex-llm
-Version: 2.2.0b20250114
+Version: 2.2.0b20250115
 Summary: Large Language Model Develop Toolkit
 Home-page: https://github.com/intel-analytics/ipex-llm
 Author: BigDL Authors
@@ -27,10 +27,10 @@ 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 ==2.6.0b20250114 ; extra == 'cpp'
+Requires-Dist: bigdl-core-cpp ==2.6.0b20250115 ; extra == 'cpp'
 Requires-Dist: setuptools ; extra == 'cpp'
 Provides-Extra: cpp-arl
-Requires-Dist: bigdl-core-cpp ==2.6.0b20250114 ; extra == 'cpp-arl'
+Requires-Dist: bigdl-core-cpp ==2.6.0b20250115 ; extra == 'cpp-arl'
 Requires-Dist: setuptools ; extra == 'cpp-arl'
 Requires-Dist: onednn-devel ==2024.1.1 ; (platform_system == "Windows") and extra == 'cpp-arl'
 Requires-Dist: onednn ==2024.1.1 ; (platform_system == "Windows") and extra == 'cpp-arl'
@@ -67,7 +67,7 @@ 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 ==2.6.0b20250114 ; (platform_system == "Windows") and extra == 'npu'
+Requires-Dist: bigdl-core-npu ==2.6.0b20250115 ; (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'
@@ -87,9 +87,9 @@ 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 ==2.6.0b20250114 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250114 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250114 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-21 ==2.6.0b20250115 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250115 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250115 ; extra == 'xpu'
 Provides-Extra: xpu-2-1
 Requires-Dist: py-cpuinfo ; extra == 'xpu-2-1'
 Requires-Dist: protobuf ; extra == 'xpu-2-1'
@@ -104,9 +104,9 @@ 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 ==2.6.0b20250114 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250114 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250114 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-21 ==2.6.0b20250115 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250115 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250115 ; 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'
@@ -124,7 +124,7 @@ 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 ==2.6.0b20250114 ; extra == 'xpu-2-6'
+Requires-Dist: bigdl-core-xe-all ==2.6.0b20250115 ; extra == 'xpu-2-6'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'xpu-2-6'
 Provides-Extra: xpu-arc
 Requires-Dist: py-cpuinfo ; extra == 'xpu-arc'
@@ -137,9 +137,9 @@ Requires-Dist: tokenizers ==0.15.2 ; extra == 'xpu-arc'
 Requires-Dist: accelerate ==0.23.0 ; extra == 'xpu-arc'
 Requires-Dist: tabulate ; extra == 'xpu-arc'
 Requires-Dist: setuptools ; extra == 'xpu-arc'
-Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250114 ; extra == 'xpu-arc'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250114 ; extra == 'xpu-arc'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250114 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250115 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250115 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250115 ; extra == 'xpu-arc'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'xpu-arc'
 Requires-Dist: torch ==2.3.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-arc'
 Requires-Dist: torchvision ==0.18.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-arc'
@@ -160,9 +160,9 @@ Requires-Dist: tokenizers ==0.15.2 ; extra == 'xpu-arl'
 Requires-Dist: accelerate ==0.23.0 ; extra == 'xpu-arl'
 Requires-Dist: tabulate ; extra == 'xpu-arl'
 Requires-Dist: setuptools ; extra == 'xpu-arl'
-Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250114 ; extra == 'xpu-arl'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250114 ; extra == 'xpu-arl'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250114 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250115 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250115 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250115 ; extra == 'xpu-arl'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'xpu-arl'
 Requires-Dist: torch ==2.3.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-arl'
 Requires-Dist: torchvision ==0.18.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-arl'
@@ -183,9 +183,9 @@ Requires-Dist: tokenizers ==0.15.2 ; extra == 'xpu-lnl'
 Requires-Dist: accelerate ==0.23.0 ; extra == 'xpu-lnl'
 Requires-Dist: tabulate ; extra == 'xpu-lnl'
 Requires-Dist: setuptools ; extra == 'xpu-lnl'
-Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250114 ; extra == 'xpu-lnl'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250114 ; extra == 'xpu-lnl'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250114 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250115 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250115 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250115 ; extra == 'xpu-lnl'
 Requires-Dist: intel-openmp ; (platform_machine == "x86_64" or platform_machine == "AMD64") and extra == 'xpu-lnl'
 Requires-Dist: torch ==2.3.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-lnl'
 Requires-Dist: torchvision ==0.18.1+cxx11.abi ; (platform_system == "Linux") and extra == 'xpu-lnl'

{ipex_llm-2.2.0b20250114.dist-info → ipex_llm-2.2.0b20250115.dist-info}/RECORD

@@ -9,7 +9,7 @@ ipex_llm/cli/prompts/chat-with-llm.txt,sha256=PpSyd4FQQd-T7ptfXL9jZp7dgstevu1fsx
 ipex_llm/ggml/__init__.py,sha256=FzapYBUiTdZf0LzlN9hfJI-HE1OTi_2dzaYELJ9Mw8s,1272
 ipex_llm/ggml/convert.py,sha256=xfWH1E_hivbsxVo8h00STjH1Rlu9-dZQkCLLeIs1TWA,5286
 ipex_llm/ggml/convert_model.py,sha256=t-tGK9w8ZRi9dlDLTutput3ZBKj3ji94WUJi2KG8hkA,5955
-ipex_llm/ggml/quantize.py,sha256=TOoSe1w7tqAoV_fuqQhXIYb1S02A6udeud6IX0kLSic,6330
+ipex_llm/ggml/quantize.py,sha256=Cvk1R771rRDhSW7BRWcmb4ImY6TWDl_u9Vkdh7rYSuM,6367
 ipex_llm/ggml/model/__init__.py,sha256=T-EbRT6GJ_8RCu-iLmSzcftOimXSPQf2d5X72AUAy2Y,874
 ipex_llm/ggml/model/bloom/__init__.py,sha256=291QHI19FMw7Z1oaKBAf2YJ0M51iYqWC4IT1ejI-OGg,900
 ipex_llm/ggml/model/bloom/bloom.py,sha256=fUxgZd_Uc4RXaMC_naYdjekwNprM1TpURmQ8VbocShc,17975
@@ -41,35 +41,35 @@ ipex_llm/langchain/llms/transformerspipelinellm.py,sha256=vm522YPPwWxxAPVvQBtxRf
 ipex_llm/langchain/vllm/__init__.py,sha256=T-EbRT6GJ_8RCu-iLmSzcftOimXSPQf2d5X72AUAy2Y,874
 ipex_llm/langchain/vllm/vllm.py,sha256=6dxc-ZISZQrJilEa_HA827l75Dv9rcHpY_G6FdJ8BVs,7793
 ipex_llm/libs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ipex_llm/libs/bloom-api.dll,sha256=_INZo-YdawiFT4W8ycNSvfZGX7PyacE1FdWqdZ8ACBs,36352
-ipex_llm/libs/bloom.dll,sha256=rJwoSwgB6yXZBD98zjINiMMriV_HrsGtV-qUE7Lvfko,506880
-ipex_llm/libs/gptneox-api.dll,sha256=s5DlqOac9kaIpVCHI2R2y_GMz3ywntM2Z5JxC8CdByc,24576
-ipex_llm/libs/gptneox.dll,sha256=QH5m60fW8LZ6eDe-e66TIajdVNvh0KBjgFVlkJ-Hg_8,567296
-ipex_llm/libs/libbloom_avx.dll,sha256=3pTsql6-EovdJ4V7_RhYPHFNA2sNmdd3o6JiX_ZRB8A,535040
-ipex_llm/libs/libbloom_vnni.dll,sha256=_s9QDl8PwDk2i0HlAXAc4zNPSD-QP8CYuGlFNlW3kqw,506880
-ipex_llm/libs/libgptneox_avx.dll,sha256=5OFhG019GnKBSunLFj8Hdl3y74vYN9U166yH5jzq02o,595456
-ipex_llm/libs/libgptneox_vnni.dll,sha256=cO_uJK9ryOZCq1Mkis0JeeoXL7ELO4_5wwR2CsnLduM,567808
-ipex_llm/libs/libllama_avx.dll,sha256=rnzeKhvjPsR9yAqA2iGzeyTzJHsGqkzUhlWsLTKO-L0,589824
-ipex_llm/libs/libllama_vnni.dll,sha256=MpFGU7izhp3cjAV3dilxSawboR4h8YToQs-zLmVp09U,561664
-ipex_llm/libs/libstarcoder_avx.dll,sha256=_2mioTJ9PKfBnE7ePf_DXbf1nnv0kQvPWfH6Em8F2Dc,626688
-ipex_llm/libs/libstarcoder_vnni.dll,sha256=uCzy3zuwkJ9kpMsfjASte0AEAfANfMfy2QPO2_sU7iA,598528
-ipex_llm/libs/llama-api.dll,sha256=G9AOyrQqXM0sNyvuenC2AsltfyOGu7PUVP49Cd51Alw,25600
-ipex_llm/libs/llama.dll,sha256=2RsqXu5FuTFYV0-xfAT6p4W0rTpn_CLgPfY7_FOf6H0,561152
-ipex_llm/libs/main-bloom.exe,sha256=Oip9lCr0Z9E9JtNmmSfcpeIpa9u8QqljDoA6J26iyOs,103424
-ipex_llm/libs/main-gptneox.exe,sha256=HPSbPmvCZG7hsC6Cl5Aq2KIXwJ9x1C1rQKy-bXfsPvI,98816
-ipex_llm/libs/main-llama.exe,sha256=Ae1q1zudFru-aw8scEAByyfTNJysBJ8ok4gPM9QHgeg,99840
-ipex_llm/libs/main-starcoder.exe,sha256=_aCei2KKCjkY4eN8M-4uNBnmjSPCRCV4l5HzPuXKVyg,157696
-ipex_llm/libs/pipeline.dll,sha256=YNiP5VgCJurVeo7w-Ry3rfSW5a7XiXu_auJP6iPRNsE,72704
-ipex_llm/libs/quantize-bloom.exe,sha256=-BPVLMa469h23MFGald10g70gCzuAxsxYORFaov3iRs,126464
-ipex_llm/libs/quantize-bloom_vnni.exe,sha256=zOw0ZBktkMBZT1XuDrbmoAdNr1MgD-Sv7_0SjROSqUc,127488
-ipex_llm/libs/quantize-gptneox.exe,sha256=8-ZWZMSejey3NZYEgICAGB1_Ypb0g_sSSE7oEVRJBqU,104448
-ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=6q4jDAT4eRgVmlb-bTA2N2HUmtWv1X2D6CjzYxtayCI,104960
-ipex_llm/libs/quantize-llama.exe,sha256=0lKm4Iq-CfHORbiQHcANIzsthrDt5PovjMYMIeRTk9I,109568
-ipex_llm/libs/quantize-llama_vnni.exe,sha256=wMP_pOkyNOPvHSAuWFr5f07SRLKbrkgbGo41O3hv5a4,110592
-ipex_llm/libs/quantize-starcoder.exe,sha256=mbh4bLwkWbtddyPdCz0jYNUf8vMFPZxFShOxtCqMzV4,127488
-ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=AxO1y6ERwky9UJpjylpQ43U_xgxdX8PAdKXbyx1xsYI,128512
-ipex_llm/libs/starcoder-api.dll,sha256=_6xipNB_cWOy0iF7YmC10akPFNIpV_ymVDc5HcdUZtg,21504
-ipex_llm/libs/starcoder.dll,sha256=d3a-7_HMs7Mivq2719k7Unc5TEvLg2GwVIr7FgwiO7U,598016
+ipex_llm/libs/bloom-api.dll,sha256=m1QkAO-pzX8Tl4-7h7ECPkcQd55Mva_0YiCQHjvoCuc,36352
+ipex_llm/libs/bloom.dll,sha256=4wcMNz_M55USU1gK76cfwM8-UL4KQyBtqEPRi6w4VQA,507904
+ipex_llm/libs/gptneox-api.dll,sha256=ZMlvwz2bYbkonJ8na-AMhq-i-lhkwR4A2Y9SEKndcuU,24576
+ipex_llm/libs/gptneox.dll,sha256=qt-FMb8mLgt6yGydHET__9XVrt58Jzpvg4aSa1SO6oI,568320
+ipex_llm/libs/libbloom_avx.dll,sha256=Z95j-YTrQ8iHx9PoSxCeVFrYSmzT1mJxa7dTFMz598A,536576
+ipex_llm/libs/libbloom_vnni.dll,sha256=Kvl-Zy0k2dyoH4hksAfHfAQYYSFYMuuyeXXel8DjVq8,508416
+ipex_llm/libs/libgptneox_avx.dll,sha256=twEmfEw5TQYZuDkcKzxeXnrWpeutV4jKTTqcqLWNMSE,596992
+ipex_llm/libs/libgptneox_vnni.dll,sha256=3z656g1unGyfGRziVxIqjBMj4o_e2FYEzbebwNd7Kl4,568832
+ipex_llm/libs/libllama_avx.dll,sha256=h8VRZJ8kafy2NDJ2gRWxK4oHOWjn6vzkwDWD2Gj1nlQ,591360
+ipex_llm/libs/libllama_vnni.dll,sha256=CGSZXEmwnnJCrNRtARHAh6u8Xpdp5F-o8U2s3vhLCBU,563200
+ipex_llm/libs/libstarcoder_avx.dll,sha256=WXzhTIWIXhfMkkR8LHz_1v_V1m7PjGtO2tr6VyizyCc,627712
+ipex_llm/libs/libstarcoder_vnni.dll,sha256=NDIt9M04LDLQlTpCso01xLWxSbXf16yhJw5vZs4O7c0,599552
+ipex_llm/libs/llama-api.dll,sha256=K7f5zkVxmuhXF3hvqpMbvX4YDYx333UDjho-Dx8cSuM,25600
+ipex_llm/libs/llama.dll,sha256=NLjjB6-g2SG6eKl7Z5T1HfNo6vlCi9Ksv-4s0CgHWF4,562688
+ipex_llm/libs/main-bloom.exe,sha256=TyxnTvGmAiMCk1CSgjHrSB8heuxn01-C4v6tDoOrtUk,103424
+ipex_llm/libs/main-gptneox.exe,sha256=ObDn-P-3gLZYbu3H-CLK5sGFsQYqByuBpzRdk3dWqTU,98816
+ipex_llm/libs/main-llama.exe,sha256=qOtiJIf1S3sT-U576WuTP5ewBHjg2BRNd9iOm2Aw97g,99840
+ipex_llm/libs/main-starcoder.exe,sha256=so9jBqjslD_lH-dQX7_ud_BQLz2uostxVmRE5dCvQAc,157696
+ipex_llm/libs/pipeline.dll,sha256=V8grkx3xoVRzYkrs78LDqAPUZeKPIx3n73CjLG5cGMo,72704
+ipex_llm/libs/quantize-bloom.exe,sha256=J0ev15UaUIDCw966zJUyFqFcKtTqih0dlB2TdVqam7U,126464
+ipex_llm/libs/quantize-bloom_vnni.exe,sha256=b0aY-CcFLY7ric3fAOsiTG3yiKNGrobiZUw7JDNvZ-M,128000
+ipex_llm/libs/quantize-gptneox.exe,sha256=FcjfWHCHA5Frv3XYLIsBrGmbcfzzX_QY1mELCxsUDRk,104448
+ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=An_lGuTuftDfrSaMrLu5zpszFMalFjHIW7LnNQWUdfI,104960
+ipex_llm/libs/quantize-llama.exe,sha256=ftxWVqV1sj39hbXYjcHYzUdDsonZ_efNyY3ZkGl4Sko,110080
+ipex_llm/libs/quantize-llama_vnni.exe,sha256=IJfegrxBAL1HHj7PdAJ0HnGorGBJCkzlM_zIkvIduQ8,110592
+ipex_llm/libs/quantize-starcoder.exe,sha256=WxhZfJAPczJGLN6LBkxFBslNk5kjp3Oluq88PEQa6MU,127488
+ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=7N1OYZ3lqqfaMLP4M3tKg54f4dPtVzXqfDMd5l8nnLI,128512
+ipex_llm/libs/starcoder-api.dll,sha256=YUXlf8YqEa9M8x7Oe8G2AdctfOa32aJ4OMpUFfAMVuA,21504
+ipex_llm/libs/starcoder.dll,sha256=UwOP_9b_izkojuAt7xQnMQkDpTwmT0_uOxN0P36_5hQ,599040
 ipex_llm/llamaindex/__init__.py,sha256=T-EbRT6GJ_8RCu-iLmSzcftOimXSPQf2d5X72AUAy2Y,874
 ipex_llm/llamaindex/llms/__init__.py,sha256=KP1lEdGqDuxPoxL1ZSH25Pm2kKMPJBWUTLR0ckSLMIU,1139
 ipex_llm/llamaindex/llms/bigdlllm.py,sha256=FQBzq1KOjfc6uofTXAha3O7TqpJkNfOFepXQmOVlbnI,26314
@@ -87,14 +87,14 @@ 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=pJHs6GZXHIObVE4BUCuej-6BKBZZg9pYWKPrkhWSfB4,1192
-ipex_llm/transformers/convert.py,sha256=i2IOmDnQBKNtcfUL95l8w4rNBMiU4SqX_9uz_LtPHMI,98086
+ipex_llm/transformers/convert.py,sha256=_aTdg1c15aUXJD4ktgcaDcqrjyEIlsoPLbMT5zvXr6U,98150
 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=k4TU18LlA-Sbq9WNNQnfuzu3RSFBwFhmaV3BcGN5bAo,19191
 ipex_llm/transformers/lisa.py,sha256=F5WxbtXQ7RdKulj83h_2DnEIgKiKGZf7zvOmg6QBl2s,3289
 ipex_llm/transformers/loader.py,sha256=AwjV5RpI2t2bedlv7ZhLm8cfd-QJZm5hny-XyjIvdnk,6876
 ipex_llm/transformers/lookup.py,sha256=b6OlZ9OV10R9qeWw8mVryVpDxszkjwLkldvi7GPMJY8,19614
-ipex_llm/transformers/low_bit_linear.py,sha256=1diyN_5OTViAmi8CRIUh5j1s5Z9j3Umcz2MEiideV4U,40863
+ipex_llm/transformers/low_bit_linear.py,sha256=27xrIh-0-KAyepaT9lHz3W4kkjoRRo08zVO4C8LRJkc,40924
 ipex_llm/transformers/model.py,sha256=64eRLtbFftku9kcN5xSNerbZ3jAFUjCbNlM8Ly54ygE,41079
 ipex_llm/transformers/modelling_bigdl.py,sha256=7JpNVMuyq_OmtNUaMFMXdxPWZp2q0QHC02QeA-VTPOw,6709
 ipex_llm/transformers/npu_model.py,sha256=X8ZtvZJpzz64XrSPhUYXXZmdJcbZ9X6G3Vlzw-zgN1Q,39749
@@ -154,7 +154,7 @@ ipex_llm/transformers/models/gptbigcode.py,sha256=cP1_qGWoa43R2WacAMblShjku4Qupc
 ipex_llm/transformers/models/gptneox.py,sha256=loRh1x_5S6BCeOr_s5xr-N_1SQHL3Y5IiUBAEyoMUqQ,6172
 ipex_llm/transformers/models/internlm.py,sha256=JZFrI2HXsIAfM-6pA2RO0wcXopOliC1FggLMzNzaDZ4,17404
 ipex_llm/transformers/models/internvl.py,sha256=Vx0vENIEQLX2M6P398mw5TOhpks0U8xf8rtRQvy94go,8154
-ipex_llm/transformers/models/llama.py,sha256=NzpyQve_RC9ez1W-jWPLGZ80k_S1I5Rx5saAzCsDIoI,8558
+ipex_llm/transformers/models/llama.py,sha256=rqrNjuZb_jeb9MKx0z-FSVoGx8YDBxQzPJ9ZUvYhgx0,9138
 ipex_llm/transformers/models/minicpm.py,sha256=eaPNVNrep0_xGoELhZd886ff0ceoKqB6cusdAhd52eE,10145
 ipex_llm/transformers/models/minicpm3.py,sha256=11cYl8KM2hoIJNMAOZMxiwCu6dMhup9ric_OEn8-VrQ,9363
 ipex_llm/transformers/models/minicpmv.py,sha256=PP05b5iTnrMpiseCn8iJcxKJDnfq7WqXp9Mrch0kKZ0,9876
@@ -174,7 +174,7 @@ ipex_llm/transformers/models/rwkv5.py,sha256=OkRNj1pCAZg1z2Fw-I0DEnxLEdZyPeRSQ6m
 ipex_llm/transformers/models/sd.py,sha256=VvHV5u-0k2MgHu3NL9113hPj7DgfxqctuKzEEeNfRDU,5981
 ipex_llm/transformers/models/stablelm.py,sha256=fj-XtOnR6kggnFUQTMPCOOzolkPztN06WAv8QW-XRnI,7054
 ipex_llm/transformers/models/starcoder2.py,sha256=ONKvD7JCkRM0DI-R56x28QFBJ7CjD5hOZBQ_3WfOcNk,6626
-ipex_llm/transformers/models/utils.py,sha256=WYBc26vSiy_CzV07z-eT5ts90Kko2yUmS3DDZtfGcRk,15065
+ipex_llm/transformers/models/utils.py,sha256=Rdn9T4zk6Hz8ybJp6kvlyfPwgHwdxEZ8R4zGtMeozWg,15105
 ipex_llm/transformers/models/yuan.py,sha256=JYAn_ZaSGK0NBJLEIxCACfAq084a66GFJkdd5NbpmMA,7732
 ipex_llm/transformers/npu_models/__init__.py,sha256=ulEUGLjaP48LCrVeury3UxLjXxKzRi0UpSG4bYu-7f8,585
 ipex_llm/transformers/npu_models/baichuan.py,sha256=fJtd7fBrttySghRUgfZTAdxLjsSNC-XL08HISsXigLE,4685
@@ -243,11 +243,11 @@ ipex_llm/vllm/xpu/engine/__init__.py,sha256=pY_CpyuZd72fr6s32ejeKHKFW0K4vUU2rzZj
 ipex_llm/vllm/xpu/engine/engine.py,sha256=k4-D27WS_Gk3mA--w3HWAjPjb4Aiu043MVPi0ZoAUBc,5984
 ipex_llm/vllm/xpu/entrypoints/openai/api_server.py,sha256=GshTZFB8e4PWvqckfbmTOU6b0oLkNn7A-vzLuG9--j8,21544
 ipex_llm/vllm/xpu/entrypoints/openai/cli_args.py,sha256=2rENA2ucynMaIjiZBEh2ez1o5vR32GaP514t39CD7KM,8676
-ipex_llm-2.2.0b20250114.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
-ipex_llm-2.2.0b20250114.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
-ipex_llm-2.2.0b20250114.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
-ipex_llm-2.2.0b20250114.dist-info/METADATA,sha256=CsE0v0mVfDVChkqM_ovKZkcfFq2zcg1yL3xthEnO7JM,12705
-ipex_llm-2.2.0b20250114.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
-ipex_llm-2.2.0b20250114.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
-ipex_llm-2.2.0b20250114.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
-ipex_llm-2.2.0b20250114.dist-info/RECORD,,
+ipex_llm-2.2.0b20250115.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
+ipex_llm-2.2.0b20250115.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
+ipex_llm-2.2.0b20250115.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
+ipex_llm-2.2.0b20250115.dist-info/METADATA,sha256=IFEK6gXdz44NE1sgaflpY0KyvKZXn9rdnuAET380aUo,12705
+ipex_llm-2.2.0b20250115.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
+ipex_llm-2.2.0b20250115.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
+ipex_llm-2.2.0b20250115.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
+ipex_llm-2.2.0b20250115.dist-info/RECORD,,