ipex-llm 2.2.0b20250120__py3-none-win_amd64.whl → 2.2.0b20250122__py3-none-win_amd64.whl

ipex_llm/vllm/cpu/model_convert.py

@@ -14,259 +14,152 @@
 # limitations under the License.
 #
 import torch
+from typing import Optional, Union
+from vllm.distributed import tensor_model_parallel_gather, tensor_model_parallel_all_gather
 from vllm.logger import init_logger
-from vllm.model_executor.model_loader import get_model
-from vllm.model_executor.model_loader.utils import get_model_architecture
-from vllm.model_executor.models.llama import LlamaMLP, LlamaAttention
-from vllm.model_executor.models.qwen2 import Qwen2MLP, Qwen2Attention
-from vllm.model_executor.models.qwen import QWenMLP, QWenAttention
+from vllm.model_executor.models.llama import LlamaMLP, LlamaAttention, LlamaForCausalLM
+from vllm.model_executor.models.qwen2 import Qwen2MLP, Qwen2Attention, Qwen2ForCausalLM
+from vllm.model_executor.models.qwen import QWenMLP, QWenAttention, QWenLMHeadModel
 from vllm.model_executor.models.baichuan import BaiChuanMLP, BaiChuanAttention
-from vllm.model_executor.models.chatglm import GLMMLP, GLMAttention
-from vllm.attention import Attention, AttentionMetadata
-from vllm.lora.worker_manager import LRUCacheWorkerLoRAManager
+from vllm.model_executor.models.baichuan import BaiChuanBaseForCausalLM
+from vllm.model_executor.models.chatglm import GLMMLP, GLMAttention, ChatGLMForCausalLM
+from vllm.model_executor.model_loader import get_model
+from vllm.model_executor.layers.vocab_parallel_embedding import (
+    VocabParallelEmbedding)
+from vllm.attention import AttentionMetadata
 from vllm.config import DeviceConfig
-from vllm.logger import init_logger
-
-from vllm._C import ops
-from ipex_llm.utils.common import invalidInputError
-from typing import List, Optional, Tuple, Union
-
-logger = init_logger(__name__)
-
-
-def _MLP_forward(self, x):
-    gate_up = self.gate_up_proj(x)
-    x = self.act_fn(gate_up)
-    x = self.down_proj(x)
-    return x
-
-
-def _Attention_forward(
-    self,
-    positions: torch.Tensor,
-    hidden_states: torch.Tensor,
-    kv_cache: torch.Tensor,
-    attn_metadata: AttentionMetadata,
-) -> torch.Tensor:
-    qkv = self.qkv_proj(hidden_states).to(dtype=kv_cache.dtype)
-    q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
-    q, k = self.rotary_emb(positions, q, k)
-    attn_output = self.attn(q, k, v, kv_cache, attn_metadata, self.kv_scale)
-    output = self.o_proj(attn_output)
-    return output
-
-
-def _QWen_Attention_forward(
-    self,
-    positions: torch.Tensor,
-    hidden_states: torch.Tensor,
-    kv_cache: Tuple[torch.Tensor, torch.Tensor],
-    attn_metadata: AttentionMetadata,
-) -> torch.Tensor:
-    qkv = self.c_attn(hidden_states).to(dtype=kv_cache.dtype)
-    q, k, v = qkv.chunk(chunks=3, dim=-1)
-    q, k = self.rotary_emb(positions, q, k)
-    attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
-    output = self.c_proj(attn_output)
-    return output
-
-
-def _QWen_MLP_forward(self, x):
-    gate_up = self.gate_up_proj(x)
-    x = self.act_fn(gate_up)
-    x = self.c_proj(x)
-    return x
-
-
-def _Qwen2_Attention_forward(
-    self,
-    positions: torch.Tensor,
-    hidden_states: torch.Tensor,
-    kv_cache: torch.Tensor,
-    attn_metadata: AttentionMetadata,
-) -> torch.Tensor:
-    qkv = self.qkv_proj(hidden_states).to(dtype=kv_cache.dtype)
-    q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
-    q, k = self.rotary_emb(positions, q, k)
-    attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
-    output = self.o_proj(attn_output)
-    return output
-
-
-def _ChatGLM_MLP_forward(self, hidden_states):
-    # [s, b, 4hp]
-    intermediate_parallel = self.dense_h_to_4h(hidden_states)
-    intermediate_parallel = self.activation_func(intermediate_parallel)
-    # [s, b, h]
-    output = self.dense_4h_to_h(intermediate_parallel)
-    return output
-
-
-def _Baichuan_Attention_forward(
-    self,
-    positions: torch.Tensor,
-    hidden_states: torch.Tensor,
-    kv_cache: Tuple[torch.Tensor, torch.Tensor],
-    attn_metadata: AttentionMetadata,
-) -> torch.Tensor:
-    qkv = self.W_pack(hidden_states).to(dtype=kv_cache.dtype)
-    q, k, v = qkv.chunk(chunks=3, dim=-1)
-    if self.postion_embedding != "ALIBI":
-        q, k = self.rotary_emb(positions, q, k)
-    attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
-    output = self.o_proj(attn_output)
-    return output
+from typing import Tuple
+from ipex_llm.transformers.low_bit_linear import LowBitLinear
 
 
-def _ChatGLM_Attention_forward(
+def _sample_get_logits(
     self,
     hidden_states: torch.Tensor,
-    position_ids: torch.Tensor,
-    kv_cache: Tuple[torch.Tensor, torch.Tensor],
-    attn_metadata: AttentionMetadata,
+    lm_head: Union[VocabParallelEmbedding, LowBitLinear],
+    embedding_bias: Optional[torch.Tensor],
 ) -> torch.Tensor:
-    qkv = self.query_key_value(hidden_states).to(dtype=kv_cache.dtype)
-    q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
-    q, k = self.rotary_emb(position_ids, q, k)
-    context_layer = self.attn(
-        q,
-        k,
-        v,
-        kv_cache,
-        attn_metadata,
-    )
-    attn_output = self.dense(context_layer)
-    return attn_output
-
-_REPLACED_MLP_LAYERS = {
-    LlamaMLP: _MLP_forward,
-    Qwen2MLP: _MLP_forward,
-    BaiChuanMLP: _MLP_forward,
-    # QWenMLP: _QWen_MLP_forward,
-    GLMMLP: _ChatGLM_MLP_forward
-}
-
-_REPLACED_ATTENTION_LAYERS = {
-    LlamaAttention: _Attention_forward,
-    Qwen2Attention: _Qwen2_Attention_forward,
-    # QWenAttention: _QWen_Attention_forward,
-    BaiChuanAttention: _Baichuan_Attention_forward,
-    GLMAttention: _ChatGLM_Attention_forward
-}
-
-_IPEX_LLM_SUPPORTED_MODELS = [
-    "LlamaForCausalLM",
-    "BaichuanForCausalLM",
-    "ChatGLMForCausalLM",
-    "Qwen2ForCausalLM",
-]
-
-
-def _model_mlp_convert():
-    for module, replaced_func in _REPLACED_MLP_LAYERS.items():
-        setattr(module, "forward", replaced_func)
+    # HINT: we do not support other types of quantization for now
+    # TODO: we may encounter tie-word-embedding problems
+    if isinstance(lm_head, VocabParallelEmbedding):
+        logits = lm_head.linear_method.apply(lm_head,
+                                             hidden_states,
+                                             bias=embedding_bias)
+    else:
+        logits = lm_head(hidden_states)
+        if embedding_bias is not None:
+            logits += embedding_bias
+    if self.use_gather:
+        logits = tensor_model_parallel_gather(logits)
+    else:
+        logits = tensor_model_parallel_all_gather(logits)
+    if logits is not None:
+        logits = logits[:, : self.org_vocab_size]
+    return logits
 
 
-def _model_attention_convert():
-    for module, replaced_func in _REPLACED_ATTENTION_LAYERS.items():
-        setattr(module, "forward", replaced_func)
+def _model_sample_convert():
+    from vllm.model_executor.layers.logits_processor import LogitsProcessor
+    setattr(LogitsProcessor, "_get_logits", _sample_get_logits)
 
 
 def _ipex_llm_convert(load_in_low_bit):
-    if load_in_low_bit is None:
-        return
     from vllm.worker.cpu_model_runner import CPUModelRunner
-    import vllm.model_executor.model_loader as model_loader
+    from ipex_llm.vllm.cpu.ipex_llm_wrapper import get_ipex_llm_wrapper
+    from ipex_llm.vllm.cpu.ipex_llm_v1_wrapper import get_ipex_llm_v1_wrapper
+    import vllm.executor.ray_utils as ray_utils_v0
+    import vllm.v1.executor.ray_utils as ray_utils_v1
     setattr(CPUModelRunner, "load_model", get_load_function(load_in_low_bit))
-
-    from vllm.model_executor.layers.rotary_embedding import RotaryEmbedding
-    setattr(RotaryEmbedding, "forward", _ipex_llm_rotary_embedding_forward)
-    from vllm.model_executor.layers.layernorm import RMSNorm
-    setattr(RMSNorm, "forward", _ipex_llm_rmsnorm_forward)
-
-
-def _ipex_llm_rotary_embedding_forward(
-    self,
-    positions: torch.Tensor,
-    query: torch.Tensor,
-    key: torch.Tensor,
-    offsets: Optional[torch.Tensor] = None,
-) -> Tuple[torch.Tensor, torch.Tensor]:
-    self.cos_sin_cache = self.cos_sin_cache.to(positions.device, dtype=query.dtype)
-
-    # ops.rotary_embedding()/batched_rotary_embedding()
-    # are in-place operations that update the query and key tensors.
-    if offsets is not None:
-        ops.batched_rotary_embedding(positions, query, key, self.head_size,
-                                     self.cos_sin_cache,
-                                     self.is_neox_style, self.rotary_dim,
-                                     offsets)
-    else:
-        ops.rotary_embedding(positions, query, key, self.head_size,
-                             self.cos_sin_cache, self.is_neox_style)
-    return query, key
-
-
-def _ipex_llm_rmsnorm_forward(
-    self,
-    x: torch.Tensor,
-    residual: Optional[torch.Tensor] = None,
-) -> Union[torch.Tensor, Tuple[torch.Tensor, torch.Tensor]]:
-    x = x.to(dtype=self.weight.data.dtype)
-    if residual is not None:
-        residual = residual.to(dtype=self.weight.data.dtype)
-        ops.fused_add_rms_norm(
-            x,
-            residual,
-            self.weight.data,
-            self.variance_epsilon,
-        )
-        return x, residual
-    out = torch.empty_like(x)
-    ops.rms_norm(
-        out,
-        x,
-        self.weight.data,
-        self.variance_epsilon,
-    )
-    return out
+    setattr(ray_utils_v0, "RayWorkerWrapper", get_ipex_llm_wrapper(load_in_low_bit))
+    setattr(ray_utils_v1, "RayWorkerWrapper", get_ipex_llm_v1_wrapper(load_in_low_bit))
 
 
 def get_load_function(low_bit):
     def _ipex_llm_load_model(self) -> None:
-        model_class = get_model_architecture(self.model_config)[1]
-        cur_model_list = ", ".join(_IPEX_LLM_SUPPORTED_MODELS)
-        if low_bit != "bf16":
-            invalidInputError(model_class in _IPEX_LLM_SUPPORTED_MODELS,
-                              f"Currently IPEX-LLM vLLM convert only support {cur_model_list}.")
-        else:
-            if model_class not in _IPEX_LLM_SUPPORTED_MODELS:
-                logger.warning(
-                    f"Currently IPEX-LLM vLLM convert only support {cur_model_list}."
-                )
-                self.model = get_model(
-                    model_config=self.model_config,
-                    load_config=self.load_config,
-                    device_config=self.device_config,
-                    vision_language_config=self.vision_language_config,
-                    lora_config=self.lora_config,
-                    parallel_config=self.parallel_config,
-                    scheduler_config=self.scheduler_config)
-                return
-
-        # _model_mlp_convert()
-        # _model_attention_convert()
-
+        _model_sample_convert()
+
+        # from vllm.utils import measure_device_memory
+        # from vllm.utils import DeviceMemoryProfiler
+        # with DeviceMemoryProfiler() as m:
+        from dataclasses import replace
+        new_device_config = DeviceConfig("cpu")
+        new_vllm_config = replace(self.vllm_config, device_config=new_device_config)
         self.model = get_model(
-            model_config=self.model_config,
-            load_config=self.load_config,
-            device_config=self.device_config,
-            vision_language_config=self.vision_language_config,
-            lora_config=self.lora_config,
-            parallel_config=self.parallel_config,
-            scheduler_config=self.scheduler_config)
-
+            vllm_config=new_vllm_config
+        )
+        if "qwen" in self.vllm_config.model_config.model.lower() or \
+                "baichuan" in self.vllm_config.model_config.model.lower() or \
+                "codegeex4-all" in self.vllm_config.model_config.model.lower() or \
+                "chatglm" in self.vllm_config.model_config.model.lower():
+            self.model.apply(padding_mlp)
         from ipex_llm import optimize_model
-        optimize_model(self.model, low_bit=low_bit, torch_dtype=self.model_config.dtype)
+        import os
+        not_convert_last_mlp = os.getenv("IPEX_LLM_NOT_CONVERT_LAST_MLP", None)
+        if not_convert_last_mlp is not None:
+            # only use to avoid nan value in last mlp forward running glm4-9b-chat
+            modules = ["35.mlp", "36.mlp", "37.mlp", "38.mlp", "39.mlp"]
+        else:
+            modules = None
+        if "minicpm" in self.vllm_config.model_config.model.lower():
+            modules = ["vpm", "resampler"]
+        # only for minicpm_2_6
+        if "minicpm-v" in self.vllm_config.model_config.model.lower():
+            from ipex_llm.transformers.models.minicpmv import merge_qkv
+            self.model.vpm.apply(merge_qkv)
+        if "internvl2" in self.vllm_config.model_config.model.lower():
+            modules = ["vision_model", "mlp1"]
+
+        # print(self.vllm_config.model_config.dtype)
+        # print("---------------------------------------")
+        optimize_model(self.model, low_bit=low_bit, torch_dtype=self.vllm_config.model_config.dtype,
+                       modules_to_not_convert=modules)
+        self.model = self.model.to(device=self.vllm_config.device_config.device,
+                                   dtype=self.vllm_config.model_config.dtype)
+        # print(self.model)
+        # self.model_memory_usage = m.consumed_memory
+        # logger = init_logger(__name__)
+        # logger.info("Loading model weights took %.4f GB",
+        #             self.model_memory_usage / float(2**30))
 
     return _ipex_llm_load_model
+
+
+def padding_mlp(module: torch.nn.Module):
+    mlp_gate_up_name = None
+    mlp_down_name = None
+    if isinstance(module, Qwen2MLP):
+        mlp_gate_up_name = "gate_up_proj"
+        mlp_down_name = "down_proj"
+    elif isinstance(module, GLMMLP):
+        mlp_gate_up_name = "dense_h_to_4h"
+        mlp_down_name = "dense_4h_to_h"
+    elif isinstance(module, BaiChuanMLP):
+        mlp_gate_up_name = "gate_up_proj"
+        mlp_down_name = "down_proj"
+    else:
+        return
+    hidden_size = getattr(module, mlp_down_name).output_size
+    # devide by rank
+    intermediate_size = getattr(module, mlp_down_name).input_size_per_partition
+    padding_size = 256
+    padding_intermediate_size = \
+        (intermediate_size + padding_size - 1) // padding_size * padding_size
+    if intermediate_size % padding_size == 0:
+        return
+    gate_up_weight = getattr(module, mlp_gate_up_name).weight.data
+    new_gate_up_weight = torch.zeros([padding_intermediate_size * 2, hidden_size],
+                                     dtype=gate_up_weight.dtype, device=gate_up_weight.device)
+    # merge_gate_up_weight
+    new_gate_up_weight[:intermediate_size, :] = gate_up_weight[:intermediate_size, :]
+    new_gate_up_weight[padding_intermediate_size:padding_intermediate_size+intermediate_size, :] = gate_up_weight[intermediate_size:, :]  # noqa
+    getattr(module, mlp_gate_up_name).output_size_per_partition = padding_intermediate_size * 2
+    getattr(module, mlp_gate_up_name).output_size = padding_intermediate_size * 2
+    getattr(module, mlp_gate_up_name).weight = \
+        torch.nn.Parameter(new_gate_up_weight, requires_grad=False)
+
+    down_weight = getattr(module, mlp_down_name).weight.data
+    new_down_weight = torch.zeros([hidden_size, padding_intermediate_size],
+                                  dtype=down_weight.dtype, device=down_weight.device)
+    new_down_weight[:, :intermediate_size] = down_weight
+    getattr(module, mlp_down_name).input_size_per_partition = padding_intermediate_size
+    getattr(module, mlp_down_name).input_size = padding_intermediate_size
+    getattr(module, mlp_down_name).weight = torch.nn.Parameter(new_down_weight, requires_grad=False)

{ipex_llm-2.2.0b20250120.dist-info → ipex_llm-2.2.0b20250122.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ipex-llm
-Version: 2.2.0b20250120
+Version: 2.2.0b20250122
 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 ==2.6.0b20250120 ; extra == 'cpp'
+Requires-Dist: bigdl-core-cpp ==2.6.0b20250122 ; extra == 'cpp'
 Requires-Dist: setuptools ; extra == 'cpp'
 Provides-Extra: cpp-arl
-Requires-Dist: bigdl-core-cpp ==2.6.0b20250120 ; extra == 'cpp-arl'
+Requires-Dist: bigdl-core-cpp ==2.6.0b20250122 ; 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 ==2.6.0b20250120 ; (platform_system == "Windows") and extra == 'npu'
+Requires-Dist: bigdl-core-npu ==2.6.0b20250122 ; (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 ==2.6.0b20250120 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250120 ; extra == 'xpu'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250120 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-21 ==2.6.0b20250122 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250122 ; extra == 'xpu'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250122 ; 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 ==2.6.0b20250120 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250120 ; extra == 'xpu-2-1'
-Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250120 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-21 ==2.6.0b20250122 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-batch-21 ==2.6.0b20250122 ; extra == 'xpu-2-1'
+Requires-Dist: bigdl-core-xe-addons-21 ==2.6.0b20250122 ; 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 ==2.6.0b20250120 ; extra == 'xpu-2-6'
+Requires-Dist: bigdl-core-xe-all ==2.6.0b20250122 ; 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'
@@ -140,9 +140,9 @@ 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 ==2.6.0b20250120 ; extra == 'xpu-arc'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250120 ; extra == 'xpu-arc'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250120 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250122 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250122 ; extra == 'xpu-arc'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250122 ; 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'
@@ -163,9 +163,9 @@ 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 ==2.6.0b20250120 ; extra == 'xpu-arl'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250120 ; extra == 'xpu-arl'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250120 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250122 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250122 ; extra == 'xpu-arl'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250122 ; 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'
@@ -186,9 +186,9 @@ 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 ==2.6.0b20250120 ; extra == 'xpu-lnl'
-Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250120 ; extra == 'xpu-lnl'
-Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250120 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-23 ==2.6.0b20250122 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-batch-23 ==2.6.0b20250122 ; extra == 'xpu-lnl'
+Requires-Dist: bigdl-core-xe-addons-23 ==2.6.0b20250122 ; 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-2.2.0b20250120.dist-info → ipex_llm-2.2.0b20250122.dist-info}/RECORD

@@ -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=ElOM48DXunT7S4guvSaQH8xQ_JL7JaXnbvoUF5BfOtw,36352
-ipex_llm/libs/bloom.dll,sha256=TtYxuRMUc7ZsN-cP1AlKJBNclQ0wZduO1kUeSgcOAqo,507904
-ipex_llm/libs/gptneox-api.dll,sha256=A3LEWtq5-jYIZdKBVoktVoIgxnuCyFQCdckhfbypb2k,24576
-ipex_llm/libs/gptneox.dll,sha256=qqqL9d-FIzS13rix1cLivsymaNv29WMu7xTUHLZttRg,568320
-ipex_llm/libs/libbloom_avx.dll,sha256=fuBy9eI2euJLyHcMa5J8EOXa2_YwY3oko6DTEOD0zBE,536576
-ipex_llm/libs/libbloom_vnni.dll,sha256=kKbqwJVSOg8Hitn4GBYZSpEogY0f9ytXRi63McBrQuw,508416
-ipex_llm/libs/libgptneox_avx.dll,sha256=DVEf_qTfo9H8qc75SioP-UsgLbcB5FM52oq1EY3T6tg,596992
-ipex_llm/libs/libgptneox_vnni.dll,sha256=w1fZIx8HpodE1kCoTKwG_FOtJ4_iwSTvP81ER7g5Coo,568832
-ipex_llm/libs/libllama_avx.dll,sha256=lDJZiFY6oeGE4NsOJnH9j0JoicC2NFS3sJ1_B8I0-Qk,591360
-ipex_llm/libs/libllama_vnni.dll,sha256=Q0NqK_4ox6L5CtJw-_NDL3Bcq8d2wJxUgJF5pOjAx1w,563200
-ipex_llm/libs/libstarcoder_avx.dll,sha256=xtEKhbGxU3ve8CR6nCrA0X6eZE7kHfXt-ROTc3u93DQ,627712
-ipex_llm/libs/libstarcoder_vnni.dll,sha256=e4NEMrTTYVHkH3AmafYyFjD7buJvCzeBRpmni6v_J5Q,599552
-ipex_llm/libs/llama-api.dll,sha256=aqbrwZGMjY4u48DELpr4Z1upxVuUqoGErtYkPyff8_Y,25600
-ipex_llm/libs/llama.dll,sha256=5jehsfn5c_y6tYe6569oSWWMYGANu51d_44Z7hzS4Lw,562688
-ipex_llm/libs/main-bloom.exe,sha256=cGjimmFdrQPWSVf1d_wWcY1grIdX4atvRAA7Dt0ug_4,103424
-ipex_llm/libs/main-gptneox.exe,sha256=vdL1ZEsEPyOMfqJWxgnP_X92Wnc-zZY-tqcJ8_vRcmE,98816
-ipex_llm/libs/main-llama.exe,sha256=mPutujcN_vxxfQzohMZgwl5y2455A4zoSdXT7Pf1JsU,99840
-ipex_llm/libs/main-starcoder.exe,sha256=lY3E8TCuD73ONf4QEj-r-hZdCPX79jJq6WXqjkY7hV0,157696
-ipex_llm/libs/pipeline.dll,sha256=LD7iwczX5rvo9FjTwcX66i1TwqFZkid3d3MUtIvfBFg,72704
-ipex_llm/libs/quantize-bloom.exe,sha256=d08GthA3axJ51Dt3aSLu5B_dMQAb2Y5_F8ZmEtK_XkA,126464
-ipex_llm/libs/quantize-bloom_vnni.exe,sha256=sk83Oyef-RwWIMe4FW7kI_Zw5RDjlYvbitsL0LuZU7w,128000
-ipex_llm/libs/quantize-gptneox.exe,sha256=UwUFOyUoN6OCEUjN2lHOpQNyxlnwtMj8k4YfnnuckuM,104448
-ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=DoU6RNdMQBdr0FuYWy6dN0cmlt5emgt-VwiSK3KVkRI,104960
-ipex_llm/libs/quantize-llama.exe,sha256=iNS_ihom3QFqNHJLL_GR02-krdihtgIy9TKM6CiO4cY,110080
-ipex_llm/libs/quantize-llama_vnni.exe,sha256=7Uc-1v2Vrn3A_WwHSOPxKMJqbiH2y5JwZmHslsDxgko,110592
-ipex_llm/libs/quantize-starcoder.exe,sha256=pJ25creC-Ry6RDZU1K-OKEvNj0GwSFHLOM8CpNiSi-o,127488
-ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=HjF-4S6_0R_yuGiu42RVyrt1-rytl5GjRp30ZvFumMs,128512
-ipex_llm/libs/starcoder-api.dll,sha256=bqmEpK1r-zNyNWlz0Ezc151jRt3Ad9QsRYqDrse0xko,21504
-ipex_llm/libs/starcoder.dll,sha256=_VzX6itysqS0taFtSn5X1ZmeCIj9aO1lozr-jatxYhI,599040
+ipex_llm/libs/bloom-api.dll,sha256=quxNPcqDy7sOIEfeu_fPdJmVrkWuqGWTsO8bmQWW030,36352
+ipex_llm/libs/bloom.dll,sha256=CAX0tJlqYwMnmlYwXjQ2M7WeUmvWsRid51pMItEc0l4,507904
+ipex_llm/libs/gptneox-api.dll,sha256=ZRB3wAjxWce7RivyvgqGuwg_Fn4uJc1uAmBszhg63EI,24576
+ipex_llm/libs/gptneox.dll,sha256=fNbCKX4W844dEnTEdoMPVtQb3A8IGi4MOg44EAV4cYI,568320
+ipex_llm/libs/libbloom_avx.dll,sha256=UZ94a_6kmXZLAdbUyq0uo3KHMj3UoLntdKAlXDL73zU,536576
+ipex_llm/libs/libbloom_vnni.dll,sha256=b1LdhjaBZfghPNjLdtiSMdNQqgj19DgOaqszo5dXZ-4,508416
+ipex_llm/libs/libgptneox_avx.dll,sha256=RlX5bX5OV_oCuelT1mLdU5oZy1V8jtG_ewUjjrDGrgE,596992
+ipex_llm/libs/libgptneox_vnni.dll,sha256=NLIdy93bX-FWbP-3gYxawMKMyDcem7zuC-ytzx7z4aI,568832
+ipex_llm/libs/libllama_avx.dll,sha256=BwTPw65FSFWexwbAEyk8-LTGdsiTHIykdXMu5vrBENY,591360
+ipex_llm/libs/libllama_vnni.dll,sha256=T8-4VsUbB7NJYjdbyw_ZDGYp71GYfKPllOKNX4L81I4,563200
+ipex_llm/libs/libstarcoder_avx.dll,sha256=pEX0b5rCiVw-UBqOz8PSa04TXY6md94h1j8bCdj9sn8,627712
+ipex_llm/libs/libstarcoder_vnni.dll,sha256=U_JlHGgtYdcPXdkiacm4dWBqouke4Mf8lKAHXW7UcFE,599552
+ipex_llm/libs/llama-api.dll,sha256=zUqOq0weiUTPsbaUSiW6yuKzkp2fu1lYDUShzkZDKeg,25600
+ipex_llm/libs/llama.dll,sha256=nWO8ekh8KduP0M5jV6ABmdSyNN_DCF5zOwMOiCDB_I0,562688
+ipex_llm/libs/main-bloom.exe,sha256=Fn8MRfqGuw1zASWW0SEgbjBIMl-D1jLPTt9VQ6mc8Co,103424
+ipex_llm/libs/main-gptneox.exe,sha256=7_vPfyJ6xbP3xBvf3jwpk7EqCs0jNTTMqG5xeOaDoSE,98816
+ipex_llm/libs/main-llama.exe,sha256=RAkzWejQI2b_0RmYVum0QTxRvLq74V13SRMOqE2wetU,99840
+ipex_llm/libs/main-starcoder.exe,sha256=FvpnlhBOLlkrw9bgpU9eDRnzk0AHWQluwq4u-Y6hwUw,157696
+ipex_llm/libs/pipeline.dll,sha256=Y6JosovQjHe8fnBJiAPrf6Gd-5uR9larFZFJMHVP0vk,72704
+ipex_llm/libs/quantize-bloom.exe,sha256=PXpJ-8-UgthckwMV-15Cj4pf3_DRh_PIkyVHPNJ40F4,126464
+ipex_llm/libs/quantize-bloom_vnni.exe,sha256=5x9Oxz2cURavrKui0YCJ6CQTNodu0_h14o2TjpyDKOs,128000
+ipex_llm/libs/quantize-gptneox.exe,sha256=9TawQ7WBCnJkDiTwq-RyukNofB_TxXAQSXu8JGiU-9A,104448
+ipex_llm/libs/quantize-gptneox_vnni.exe,sha256=WOtH_wUcm_0p6tlJ5ugilThl8zB9C9dpYtdYpxMxoag,104960
+ipex_llm/libs/quantize-llama.exe,sha256=ztv6PtFIbOHgBaDSfmP10PrEAoLMHYqsGHLDeN0GwXg,110080
+ipex_llm/libs/quantize-llama_vnni.exe,sha256=OERE5kHKLpHxyUa8aD8tNk_covf0x26ZsWbq9ubHbdE,110592
+ipex_llm/libs/quantize-starcoder.exe,sha256=SK7-9t9Q_qRTSKlChEDTFcvQmdkEDIvuwSAl9IofnRA,127488
+ipex_llm/libs/quantize-starcoder_vnni.exe,sha256=A6q48FyFaZRiRSSgDdTO7Ep1vdi4vvsZHgoZVYTkyYE,128512
+ipex_llm/libs/starcoder-api.dll,sha256=0TAn1ce4BRkQz9oIZ8nJGb3pW1GM_YNT-pE1Hd9njes,21504
+ipex_llm/libs/starcoder.dll,sha256=PRvZN0qYiX8C1adihh-ZU2VgwCvMYIX04yzJgnXp88s,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
@@ -87,25 +87,25 @@ 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=RVW8PIKpLrdMbvqKuUtWswyECTvDP7IM-o6I-Ez3TZ4,99554
+ipex_llm/transformers/convert.py,sha256=Ss4q1bsMgGMU2hauZJxUsEA2noPByR-fALrJGPN1sEk,99520
 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=2FhbDODYaw0D2RXfYCx3B9NyDep2D50tzQm6pYDxpzQ,40964
-ipex_llm/transformers/model.py,sha256=64eRLtbFftku9kcN5xSNerbZ3jAFUjCbNlM8Ly54ygE,41079
+ipex_llm/transformers/low_bit_linear.py,sha256=mFJRKU60ZVHm-V7gDsJYIz-ryntZi15XhS0eqSUPag4,41136
+ipex_llm/transformers/model.py,sha256=cQJNlAkdfoWmVbWd-TS2hf-Do41mMO9orPvG3FO4Nns,40855
 ipex_llm/transformers/modelling_bigdl.py,sha256=7JpNVMuyq_OmtNUaMFMXdxPWZp2q0QHC02QeA-VTPOw,6709
 ipex_llm/transformers/npu_model.py,sha256=X8ZtvZJpzz64XrSPhUYXXZmdJcbZ9X6G3Vlzw-zgN1Q,39749
-ipex_llm/transformers/patches.py,sha256=halPWm__ORh2fRFSIFPiCNg3LQBfrRkTPtmtRpBJCZQ,1286
+ipex_llm/transformers/patches.py,sha256=G9KcXxo42H1HJEDaroq4JbBN5P0P0lty7U7kk7-g4tw,991
 ipex_llm/transformers/pipeline_parallel.py,sha256=uNZpOXljNmdoEYnP8U-VFiN4dRZb2piQbIf2bG9LQnE,49051
 ipex_llm/transformers/qlora.py,sha256=jtPGsvWFjbTUGzDBCdfftnCis_0nJQNRpACSwXUbbGU,14943
 ipex_llm/transformers/relora.py,sha256=-dYzUV0P-IhO2jFdnzN9-v_sFzJpRj3ZwN9eCJzOoCw,16567
 ipex_llm/transformers/speculative.py,sha256=0XNLgc9dGswJHVPrXo4iM7pPxkWwfFfJMECcivJSnIc,63368
 ipex_llm/transformers/streamer.py,sha256=RrVlLblzCOtABRUpaMXAyaMnCGgLUtAi_YesLumRbww,4842
 ipex_llm/transformers/training_patch.py,sha256=oxMkUtqyvqJiprw6dE3skkYfD1HOmUlH9N0hBkbn0G0,10799
-ipex_llm/transformers/utils.py,sha256=9IRSqfDokf8QFW9T47R--i3RL1E-_O31HO7IJf7H6pg,16748
+ipex_llm/transformers/utils.py,sha256=JBekwpPD-CyMxt1OzvVsp7tu26pSA4v2mjuaUbqrAgI,16995
 ipex_llm/transformers/xpu_customize_fwd.py,sha256=wFpIhs5F6tkNs8gBOrLxWdhLzO3EDHovVkERPIAoAvg,7611
 ipex_llm/transformers/xpu_ops.py,sha256=vw4cUwvqUqDr45d-WMIkCpM2oiHfjN-VjF0bjMSF4kY,4830
 ipex_llm/transformers/awq/__init__.py,sha256=Du5gu3-eeAkeDO_dEMBTzrDBA66DSN3uL3-rn8WGXQw,875
@@ -233,10 +233,14 @@ ipex_llm/utils/common/lazyimport.py,sha256=AOxkmsRnqpr9zEGA5_0baqrWGhdWBmIgyKO8c
 ipex_llm/utils/common/log4Error.py,sha256=8UgIpEJYQasQO4gMOWO22nsOgr14w1emAJy4ts1VOb0,1763
 ipex_llm/vllm/__init__.py,sha256=zBSG6nzrVF5QnpR6_f7kPhBFeowTE9gaZ7D5m98E7_w,585
 ipex_llm/vllm/cpu/__init__.py,sha256=zBSG6nzrVF5QnpR6_f7kPhBFeowTE9gaZ7D5m98E7_w,585
-ipex_llm/vllm/cpu/model_convert.py,sha256=BoIQqIJpm-L0ZygnlBMiC3pHOwU3VLBNtf5JP0cAUyM,9217
-ipex_llm/vllm/cpu/engine/__init__.py,sha256=mzPVAyZdbvfzBQi-wxZh1sbme_NElPMmtrJ9C2zh8Us,747
-ipex_llm/vllm/cpu/engine/engine.py,sha256=PpFWHL5tFg49DNRtjyj0lGNzFdk_XjJpKvdWEDz6VAE,7106
-ipex_llm/vllm/cpu/entrypoints/openai/api_server.py,sha256=N1DUxhLVE9hgIMe99jhVGPdCfHAx6dfRqhLZm-D6cNk,6917
+ipex_llm/vllm/cpu/ipex_llm_v1_wrapper.py,sha256=_hCFJKXW6_Pekss8brhIIFukErOILN8BNlUtP9OFNEo,825
+ipex_llm/vllm/cpu/ipex_llm_wrapper.py,sha256=wENN6x7qZPXiUKkkFnh7E2uqeePsBxy0xsXjhPDQ88g,892
+ipex_llm/vllm/cpu/model_convert.py,sha256=FJT1rxSZjpmsFYBeu7n_s91rdQwi-FoL95lYbGqXhh8,7853
+ipex_llm/vllm/cpu/engine/__init__.py,sha256=pY_CpyuZd72fr6s32ejeKHKFW0K4vUU2rzZjWqm5YiA,783
+ipex_llm/vllm/cpu/engine/engine.py,sha256=MzVIncDlokHhWC8nj05dio0KXh_I0qob_mIlwt6v31g,9906
+ipex_llm/vllm/cpu/entrypoints/api_server.py,sha256=RDe3GrGMw47c7dVZL2IMmTHtGegnzhms7I1yOnBbwjE,30156
+ipex_llm/vllm/cpu/entrypoints/openai/api_server.py,sha256=l0tHnCaD96WD2k3zLTrYtLrcigf1eqYvszs79fYgoK8,29812
+ipex_llm/vllm/cpu/entrypoints/openai/cli_args.py,sha256=hB398yYtKauASRzevctScdbFIjiiSGMAe1bwEuIHrhY,10893
 ipex_llm/vllm/xpu/__init__.py,sha256=zBSG6nzrVF5QnpR6_f7kPhBFeowTE9gaZ7D5m98E7_w,585
 ipex_llm/vllm/xpu/ipex_llm_wrapper.py,sha256=_CbhvBuf_KPnmLfngYKtJl5gPAHVsG2mWth3wSeaH3M,892
 ipex_llm/vllm/xpu/model_convert.py,sha256=tDgaOliyBJdpcCctFNlMd9RK8fA7usLQi2RadgOj--A,7599
@@ -244,11 +248,11 @@ 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-2.2.0b20250120.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
-ipex_llm-2.2.0b20250120.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
-ipex_llm-2.2.0b20250120.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
-ipex_llm-2.2.0b20250120.dist-info/METADATA,sha256=q7KGrzwCH-U7MiiWVeH6NibIoEag5WgHfMnbhWjdSio,12879
-ipex_llm-2.2.0b20250120.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
-ipex_llm-2.2.0b20250120.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
-ipex_llm-2.2.0b20250120.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
-ipex_llm-2.2.0b20250120.dist-info/RECORD,,
+ipex_llm-2.2.0b20250122.data/scripts/ipex-llm-init.bat,sha256=HPtCYuDYwEatq7dAwOvdfVcHYCpAVdbj75K1qh0vQek,2578
+ipex_llm-2.2.0b20250122.data/scripts/llm-chat.ps1,sha256=6qrs-hGVAV8IKh7Jx8nq_XrnZcjd7qGU5wndArM7Yag,2769
+ipex_llm-2.2.0b20250122.data/scripts/llm-cli.ps1,sha256=3qBtTLs_EjYDnM8YyCpJhzLnGCKTEGssu9UNqfkjVXs,3009
+ipex_llm-2.2.0b20250122.dist-info/METADATA,sha256=9Z4u6y119-W33jLX1WPjrNGbR3RtgbUfzeXDJNbK2W8,12879
+ipex_llm-2.2.0b20250122.dist-info/WHEEL,sha256=6iYPr8vTHsyDK75jr9X0V3I9wPSVmtwr_8fdATBciGk,98
+ipex_llm-2.2.0b20250122.dist-info/entry_points.txt,sha256=TiUyBB2MRmfF3ko-pyAEzqeBCRnyhu27bNOAsWPp3e8,61
+ipex_llm-2.2.0b20250122.dist-info/top_level.txt,sha256=CGCMHM-SyqUabU4h8RqJ2KTYckQUO3LvIWwmUQ6Qbzw,9
+ipex_llm-2.2.0b20250122.dist-info/RECORD,,