xinference 0.11.1__py3-none-any.whl → 0.11.2.post1__py3-none-any.whl

xinference/model/llm/vllm/core.py

@@ -97,6 +97,8 @@ VLLM_SUPPORTED_MODELS = [
     "Yi-1.5",
     "code-llama",
     "code-llama-python",
+    "deepseek",
+    "deepseek-coder",
 ]
 VLLM_SUPPORTED_CHAT_MODELS = [
     "llama-2-chat",
@@ -125,6 +127,7 @@ VLLM_SUPPORTED_CHAT_MODELS = [
 ]
 if VLLM_INSTALLED and vllm.__version__ >= "0.3.0":
     VLLM_SUPPORTED_CHAT_MODELS.append("qwen1.5-chat")
+    VLLM_SUPPORTED_MODELS.append("codeqwen1.5")
     VLLM_SUPPORTED_CHAT_MODELS.append("codeqwen1.5-chat")
 
 if VLLM_INSTALLED and vllm.__version__ >= "0.3.2":
@@ -136,8 +139,8 @@ if VLLM_INSTALLED and vllm.__version__ >= "0.3.3":
 
 if VLLM_INSTALLED and vllm.__version__ >= "0.4.0":
     VLLM_SUPPORTED_CHAT_MODELS.append("qwen1.5-moe-chat")
-    VLLM_SUPPORTED_MODELS.append("c4ai-command-r-v01")
-    VLLM_SUPPORTED_MODELS.append("c4ai-command-r-v01-4bit")
+    VLLM_SUPPORTED_CHAT_MODELS.append("c4ai-command-r-v01")
+    VLLM_SUPPORTED_CHAT_MODELS.append("c4ai-command-r-v01-4bit")
 
 
 class VLLMModel(LLM):

xinference/model/rerank/core.py

@@ -46,7 +46,7 @@ def get_rerank_model_descriptions():
 class RerankModelSpec(CacheableModelSpec):
     model_name: str
     language: List[str]
-    type: Optional[str] = "normal"
+    type: Optional[str] = "unknown"
     model_id: str
     model_revision: Optional[str]
     model_hub: str = "huggingface"
@@ -118,6 +118,28 @@ class RerankModel:
         self._use_fp16 = use_fp16
         self._model = None
         self._counter = 0
+        if model_spec.type == "unknown":
+            model_spec.type = self._auto_detect_type(model_path)
+
+    @staticmethod
+    def _auto_detect_type(model_path):
+        """This method may not be stable due to the fact that the tokenizer name may be changed.
+        Therefore, we only use this method for unknown model types."""
+        from transformers import AutoTokenizer
+
+        type_mapper = {
+            "LlamaTokenizerFast": "LLM-based layerwise",
+            "GemmaTokenizerFast": "LLM-based",
+            "XLMRobertaTokenizerFast": "normal",
+        }
+
+        tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
+        rerank_type = type_mapper.get(type(tokenizer).__name__)
+        if rerank_type is None:
+            raise Exception(
+                f"Can't determine the rerank type based on the tokenizer {tokenizer}"
+            )
+        return rerank_type
 
     def load(self):
         if self._model_spec.type == "normal":

xinference/model/utils.py

@@ -19,6 +19,7 @@ from json import JSONDecodeError
 from pathlib import Path
 from typing import Any, Callable, Dict, Optional, Tuple, Union
 
+import huggingface_hub
 from fsspec import AbstractFileSystem
 
 from ..constants import XINFERENCE_CACHE_DIR, XINFERENCE_ENV_MODEL_SRC
@@ -27,6 +28,7 @@ from .core import CacheableModelSpec
 
 logger = logging.getLogger(__name__)
 MAX_ATTEMPTS = 3
+IS_NEW_HUGGINGFACE_HUB: bool = huggingface_hub.__version__ >= "0.23.0"
 
 
 def is_locale_chinese_simplified() -> bool:
@@ -76,6 +78,13 @@ def symlink_local_file(path: str, local_dir: str, relpath: str) -> str:
     return local_dir_filepath
 
 
+def create_symlink(download_dir: str, cache_dir: str):
+    for subdir, dirs, files in os.walk(download_dir):
+        for file in files:
+            relpath = os.path.relpath(os.path.join(subdir, file), download_dir)
+            symlink_local_file(os.path.join(subdir, file), cache_dir, relpath)
+
+
 def retry_download(
     download_func: Callable,
     model_name: str,
@@ -306,22 +315,23 @@ def cache(model_spec: CacheableModelSpec, model_description_type: type):
             model_spec.model_id,
             revision=model_spec.model_revision,
         )
-        for subdir, dirs, files in os.walk(download_dir):
-            for file in files:
-                relpath = os.path.relpath(os.path.join(subdir, file), download_dir)
-                symlink_local_file(os.path.join(subdir, file), cache_dir, relpath)
+        create_symlink(download_dir, cache_dir)
     else:
         from huggingface_hub import snapshot_download as hf_download
 
-        retry_download(
+        use_symlinks = {}
+        if not IS_NEW_HUGGINGFACE_HUB:
+            use_symlinks = {"local_dir_use_symlinks": True, "local_dir": cache_dir}
+        download_dir = retry_download(
             hf_download,
             model_spec.model_name,
             None,
             model_spec.model_id,
             revision=model_spec.model_revision,
-            local_dir=cache_dir,
-            local_dir_use_symlinks=True,
+            **use_symlinks,
         )
+        if IS_NEW_HUGGINGFACE_HUB:
+            create_symlink(download_dir, cache_dir)
     with open(meta_path, "w") as f:
         import json
 

xinference/thirdparty/deepseek_vl/models/processing_vlm.py

@@ -25,8 +25,8 @@ from PIL.Image import Image
 from transformers import LlamaTokenizerFast
 from transformers.processing_utils import ProcessorMixin
 
-from .image_processing_vlm import VLMImageProcessor
 from ..utils.conversation import get_conv_template
+from .image_processing_vlm import VLMImageProcessor
 
 
 class DictOutput(object):

xinference/thirdparty/deepseek_vl/models/siglip_vit.py

@@ -92,7 +92,7 @@ def _no_grad_trunc_normal_(tensor, mean, std, a, b):
 def trunc_normal_(tensor, mean=0.0, std=1.0, a=-2.0, b=2.0):
     # type: (torch.Tensor, float, float, float, float) -> torch.Tensor
     r"""The original timm.models.layers.weight_init.trunc_normal_ can not handle bfloat16 yet, here we first
-    convert the tensor to float32, apply the trunc_normal_() in float32, and then convert it back to its orignal dtype.
+    convert the tensor to float32, apply the trunc_normal_() in float32, and then convert it back to its original dtype.
     Fills the input Tensor with values drawn from a truncated normal distribution. The values are effectively drawn
     from the normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)`
     with values outside :math:`[a, b]` redrawn until they are within
@@ -305,7 +305,7 @@ class VisionTransformer(nn.Module):
             img_size: Input image size.
             patch_size: Patch size.
             in_chans: Number of image input channels.
-            num_classes: Mumber of classes for classification head.
+            num_classes: Number of classes for classification head.
             global_pool: Type of global pooling for final sequence (default: 'token').
             embed_dim: Transformer embedding dimension.
             depth: Depth of transformer.

xinference/thirdparty/llava/mm_utils.py

@@ -2,11 +2,12 @@ import base64
 from io import BytesIO
 
 import torch
-from .model import LlavaLlamaForCausalLM
-from .model.constants import IMAGE_TOKEN_INDEX
 from PIL import Image
 from transformers import AutoTokenizer, StoppingCriteria
 
+from .model import LlavaLlamaForCausalLM
+from .model.constants import IMAGE_TOKEN_INDEX
+
 
 def load_image_from_base64(image):
     return Image.open(BytesIO(base64.b64decode(image)))

xinference/thirdparty/llava/model/llava_arch.py

@@ -17,9 +17,9 @@ import os
 from abc import ABC, abstractmethod
 
 import torch
-from .constants import IGNORE_INDEX, IMAGE_TOKEN_INDEX, key_info
 
 from .clip_encoder.builder import build_vision_tower
+from .constants import IGNORE_INDEX, IMAGE_TOKEN_INDEX, key_info
 from .multimodal_projector.builder import build_vision_projector
 
 

xinference/thirdparty/omnilmm/chat.py

@@ -7,11 +7,6 @@ import torch
 from PIL import Image
 from transformers import AutoModel, AutoTokenizer
 
-from .model.omnilmm import OmniLMMForCausalLM
-from .model.utils import build_transform
-from .train.train_utils import omni_preprocess
-from .utils import disable_torch_init
-
 DEFAULT_IMAGE_TOKEN = "<image>"
 DEFAULT_IMAGE_PATCH_TOKEN = "<im_patch>"
 DEFAULT_IM_START_TOKEN = "<im_start>"
@@ -21,6 +16,10 @@ DEFAULT_IM_END_TOKEN = "<im_end>"
 def init_omni_lmm(model_path, device_map):
     from accelerate import init_empty_weights, load_checkpoint_and_dispatch
 
+    from .model.omnilmm import OmniLMMForCausalLM
+    from .model.utils import build_transform
+    from .utils import disable_torch_init
+
     torch.backends.cuda.matmul.allow_tf32 = True
     disable_torch_init()
     model_name = os.path.expanduser(model_path)
@@ -98,6 +97,8 @@ def expand_question_into_multimodal(
 
 
 def wrap_question_for_omni_lmm(question, image_token_len, tokenizer):
+    from .train.train_utils import omni_preprocess
+
     question = expand_question_into_multimodal(
         question,
         image_token_len,

{xinference-0.11.1.dist-info → xinference-0.11.2.post1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: xinference
-Version: 0.11.1
+Version: 0.11.2.post1
 Summary: Model Serving Made Easy
 Home-page: https://github.com/xorbitsai/inference
 Author: Qin Xuye
@@ -20,7 +20,7 @@ Classifier: Topic :: Software Development :: Libraries
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: xoscar >=0.3.0
-Requires-Dist: torch <2.3.0
+Requires-Dist: torch
 Requires-Dist: gradio >=3.39.0
 Requires-Dist: typer[all] <0.12.0
 Requires-Dist: pillow
@@ -31,7 +31,7 @@ Requires-Dist: requests
 Requires-Dist: pydantic
 Requires-Dist: fastapi
 Requires-Dist: uvicorn
-Requires-Dist: huggingface-hub <0.23.0,>=0.19.4
+Requires-Dist: huggingface-hub >=0.19.4
 Requires-Dist: typing-extensions
 Requires-Dist: fsspec <=2023.10.0,>=2023.1.0
 Requires-Dist: s3fs
@@ -50,7 +50,7 @@ Provides-Extra: all
 Requires-Dist: chatglm-cpp >=0.3.0 ; extra == 'all'
 Requires-Dist: llama-cpp-python !=0.2.58,>=0.2.25 ; extra == 'all'
 Requires-Dist: transformers >=4.34.1 ; extra == 'all'
-Requires-Dist: torch <2.3.0 ; extra == 'all'
+Requires-Dist: torch ; extra == 'all'
 Requires-Dist: accelerate >=0.27.2 ; extra == 'all'
 Requires-Dist: sentencepiece ; extra == 'all'
 Requires-Dist: transformers-stream-generator ; extra == 'all'
@@ -70,7 +70,7 @@ Requires-Dist: torchvision ; extra == 'all'
 Requires-Dist: FlagEmbedding ; extra == 'all'
 Requires-Dist: auto-gptq ; (sys_platform != "darwin") and extra == 'all'
 Requires-Dist: autoawq ; (sys_platform != "darwin") and extra == 'all'
-Requires-Dist: vllm <0.4.2,>=0.2.6 ; (sys_platform == "linux") and extra == 'all'
+Requires-Dist: vllm >=0.2.6 ; (sys_platform == "linux") and extra == 'all'
 Requires-Dist: sglang[all] ; (sys_platform == "linux") and extra == 'all'
 Provides-Extra: benchmark
 Requires-Dist: psutil ; extra == 'benchmark'
@@ -92,6 +92,7 @@ Requires-Dist: black ; extra == 'dev'
 Requires-Dist: openai >1 ; extra == 'dev'
 Requires-Dist: opencv-contrib-python ; extra == 'dev'
 Requires-Dist: langchain ; extra == 'dev'
+Requires-Dist: langchain-community ; extra == 'dev'
 Requires-Dist: orjson ; extra == 'dev'
 Requires-Dist: sphinx-tabs ; extra == 'dev'
 Requires-Dist: sphinx-design ; extra == 'dev'
@@ -122,7 +123,7 @@ Provides-Extra: sglang
 Requires-Dist: sglang[all] ; extra == 'sglang'
 Provides-Extra: transformers
 Requires-Dist: transformers >=4.34.1 ; extra == 'transformers'
-Requires-Dist: torch <2.3.0 ; extra == 'transformers'
+Requires-Dist: torch ; extra == 'transformers'
 Requires-Dist: accelerate >=0.27.2 ; extra == 'transformers'
 Requires-Dist: sentencepiece ; extra == 'transformers'
 Requires-Dist: transformers-stream-generator ; extra == 'transformers'
@@ -138,7 +139,7 @@ Requires-Dist: peft ; extra == 'transformers'
 Requires-Dist: auto-gptq ; (sys_platform != "darwin") and extra == 'transformers'
 Requires-Dist: autoawq ; (sys_platform != "darwin") and extra == 'transformers'
 Provides-Extra: vllm
-Requires-Dist: vllm <0.4.2,>=0.2.6 ; extra == 'vllm'
+Requires-Dist: vllm >=0.2.6 ; extra == 'vllm'
 
 <div align="center">
 <img src="./assets/xorbits-logo.png" width="180px" alt="xorbits" />

{xinference-0.11.1.dist-info → xinference-0.11.2.post1.dist-info}/RECORD

@@ -1,6 +1,6 @@
 xinference/__init__.py,sha256=0LgIveLP6CXxoIaSrxhlFyOh0lOqPgJBVcBe0tkWJjc,987
 xinference/_compat.py,sha256=SQAjZMGxtBIce45qtW7ob7RWzA0zhv2yB3AxT0rb0uU,1778
-xinference/_version.py,sha256=Kr4n850ATJdQwIPIw0zSvWb9kh32gnRz1MJkkJRrTPA,498
+xinference/_version.py,sha256=mDJV0U5DkMVMiU2azZlEJ4w5LINeGsZFYONO9K3d4d0,504
 xinference/conftest.py,sha256=Qus4KWCeaKS7c5UgNCTpPNucD2bjV8P7u1_qRosgGno,9743
 xinference/constants.py,sha256=Bu_fOJUGAvvqF_6FY5OzOHl7fQ1Nomek3LY17xr9oz4,2882
 xinference/device_utils.py,sha256=zswJiws3VyTIaNO8z-MOcsJH_UiPoePPiKK5zoNrjTA,3285
@@ -9,7 +9,7 @@ xinference/isolation.py,sha256=NstVRcO3dG4umHExICXAHlzVKwH8ch8MBwKwE-KFkE0,1826
 xinference/types.py,sha256=BFKUGHb0jKkAA1dczSf8pPlFutRE7-JtRp6C3oVSJ7Q,13626
 xinference/utils.py,sha256=VSOJMFd9H7kce98OtJZbcDjjpfzRpHAFs8WU0xXPBM8,717
 xinference/api/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
-xinference/api/restful_api.py,sha256=8x0bQDE17GwQ631vDSjAgEemFZ85l93qPc4t_qUD_kc,59403
+xinference/api/restful_api.py,sha256=Wj1LWFTQiwH7FEheaG1pA4wpVOCtWHnbTdRHUa6Qcek,60530
 xinference/api/oauth2/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
 xinference/api/oauth2/auth_service.py,sha256=74JzB42fbbmBu4Q1dW3A9Fp_N7167KgRGB42Z0NHjAM,6119
 xinference/api/oauth2/types.py,sha256=K923sv_XySIUtM2Eozl9IG082IJcDOS5SFLrPZ5ELBg,996
@@ -18,9 +18,9 @@ xinference/client/__init__.py,sha256=Gc4HOzAy_1cic5kXlso7hahYgw89CKvZSJDicEU461k
 xinference/client/common.py,sha256=iciZRs5YjM2gYsXnwACPMaiBZp4_XpawWwfym0Iyu40,1617
 xinference/client/handlers.py,sha256=3gd9C7u4URbcVdR6Eyv8cpEZ175Ll4q_jGL07CnEIpg,648
 xinference/client/restful/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
-xinference/client/restful/restful_client.py,sha256=qM0lPYhyUG7vK1OzBAmkfC3CGf6GimnPhe0kFPedzDE,42798
+xinference/client/restful/restful_client.py,sha256=zqfZHliUeiDHvKKo1SrXH3rAvyxp0o_TaMwuCjV2pjQ,43589
 xinference/core/__init__.py,sha256=Fe5tYCHDbYJ7PhxJhQ68VbfgKgOsAuslNPr4wPhFMJM,612
-xinference/core/cache_tracker.py,sha256=41utiulASohGLlBqMMSmrAiH7ieGXgahEyyT9xOVh6w,4277
+xinference/core/cache_tracker.py,sha256=OMY_0HZkb-R5EhHHWppIGqPhC3v9Iwdjp9oema_6eFc,4798
 xinference/core/chat_interface.py,sha256=B-qG7RF7HOquhKaPAJSnHA3Dqov5QZQA2yO-txt1IPs,17380
 xinference/core/event.py,sha256=dTXv-zg-sAqlY1rFLvyor9D8WEXZvnUH7NigegpziO8,1648
 xinference/core/image_interface.py,sha256=G2iK24auEN4MrLkPlu1CAA_gf-BQrGQTjazi_FYqIxE,8825
@@ -28,23 +28,23 @@ xinference/core/metrics.py,sha256=ScmTG15Uq3h_ob72ybZSMWdnk8P4sUZFcm60f4ikSXc,26
 xinference/core/model.py,sha256=KAD9a2fYY97MfByNRnmia2oZR-UK7B-zW8gWl_tUJgs,18121
 xinference/core/resource.py,sha256=FQ0aRt3T4ZQo0P6CZZf5QUKHiCsr5llBvKb1f7wfnxg,1611
 xinference/core/status_guard.py,sha256=fF5hisvfn6es9DV6Z6RRD6V_S_uLcb8lHM6PArGgb04,2820
-xinference/core/supervisor.py,sha256=-NaMfDEQMfb0Xlj-_67aOunoxrmIEKM9pqbQ6YCoiiw,40663
-xinference/core/utils.py,sha256=tUpUJUQv1zkE9i7fw1pAFfFdcB3PC6DvKJn4Bmmq75E,6008
-xinference/core/worker.py,sha256=kx1ukLOdFM7kyD-E2EtGkmUe7ln7h-IffjinyKOdaLc,32629
+xinference/core/supervisor.py,sha256=bREFDEeO6rF9hyWOuf9mA6IHo6Ur2BDGgPmnuysaN9o,41969
+xinference/core/utils.py,sha256=LqPrez5dGELRQDSwOD5EP8XHb-aUKAdyszS-QpNouuw,6401
+xinference/core/worker.py,sha256=oGsqHQ5yJ4Ry-tAL06x6jkLxkQKbj0LxWtvvOy-SRdM,32780
 xinference/deploy/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
-xinference/deploy/cmdline.py,sha256=pd4j9lf376B_v1jdJY5E3TVcK51CslC4w2UV8u8H-T4,41042
+xinference/deploy/cmdline.py,sha256=J1NSfnQ-iJ87OPb_kNOfONgMZNqCow1Klm91uLS4_Lk,44582
 xinference/deploy/local.py,sha256=vlAvhcl8utP1DjW4MJpBgD4JLHQV-1Xebmdd8j9M8IM,3946
 xinference/deploy/supervisor.py,sha256=fMHeEGigQ72PD9JEFmZ5Xudn25Uj4DhD2OVIlAu_YpA,2978
 xinference/deploy/utils.py,sha256=PYdxLRARG-oZoQZtC-1t2Xve6ehjfuEITwAuDglDMIA,5377
 xinference/deploy/worker.py,sha256=Av3qU1b0tdxfkds3Mc2Qiqy9c_xSD0Tp3cToWoXqTpo,2966
 xinference/deploy/test/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
-xinference/deploy/test/test_cmdline.py,sha256=SMQwrzcxSHDjEldrv9dmC2e_4_ofT14TJLjWMTsho4M,6804
+xinference/deploy/test/test_cmdline.py,sha256=W83eR1nQGE7-DaRItEoFU0JkVH1sz_bc8TnSpc07zHc,7436
 xinference/locale/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
 xinference/locale/utils.py,sha256=w-G1DAJGw1UUQVVtq6khOZn7ZjobUmTw6qwHMm2eWIs,1312
 xinference/locale/zh_CN.json,sha256=YA55G9s1p05Bt5RBoDo5SV12dd-CMJI0ABap6RpCp4M,1097
 xinference/model/__init__.py,sha256=IRC3ojiqYkVLIK_xsIxYeKypEeeTTdrovnVzK_4L4eg,663
 xinference/model/core.py,sha256=nENyctnKtEBMakd_BpTbgCtLUeci3ak7QIwEk-pq4IM,3546
-xinference/model/utils.py,sha256=qqCaje-dJvSarVzeGgmwKnq85e82JCLPVq2yCfAFZlo,14586
+xinference/model/utils.py,sha256=L-pGH8HwIPwrJyRdnLk1OVPoDNdSlAyKp0eEY47J_fw,14974
 xinference/model/audio/__init__.py,sha256=0EVzX6b4pcOO63NAcNpYWTVYVa7w7yG5cPpGxOY9MXw,2347
 xinference/model/audio/core.py,sha256=ypbIvbueTFKeulYt7aJX7FfU4y3Hn3DzxkhhjKO6Dxw,4373
 xinference/model/audio/custom.py,sha256=Li6VpTmpZ17YXk_bwN2-tUKRAJwNcW-O4OwrJefzC2o,4966
@@ -65,12 +65,13 @@ xinference/model/image/model_spec_modelscope.json,sha256=KMI-2YOJoLUtN93LZiqqETW
 xinference/model/image/utils.py,sha256=gxg8jJ2nYaDknzCcSC53WCy1slbB5aWU14AbJbfm6Z4,906
 xinference/model/image/stable_diffusion/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
 xinference/model/image/stable_diffusion/core.py,sha256=ib_ZeSg7hzynmRqSnhjtrVuhoLOgZPrR1ZH2LjBmH2E,6063
-xinference/model/llm/__init__.py,sha256=Za1yBXmoKlvBPmqQhBV8b9xyIjRjRwfp-Hgb-ttUrck,9790
+xinference/model/llm/__init__.py,sha256=kl2z63rNpgkoFQcg536SUt4TKub2wuDB8c7weYKeeoU,9874
 xinference/model/llm/core.py,sha256=CZJrbW3HaV8vJj5mxSnBD8ssABALG_xyXyPS0hHVBPY,7982
-xinference/model/llm/llm_family.json,sha256=xOA3G-z1hsso46QaM5wL0tgZKyb28aqJ1oXcIPjossU,138441
-xinference/model/llm/llm_family.py,sha256=IGaVV1hbJfdnalBCKHtYYxb_whJUYgMgt6HBdFcdOcQ,34918
-xinference/model/llm/llm_family_modelscope.json,sha256=naDSqUlOBXkF1tYw7OrM2UuHfIRnpNJ4KrUE11wIvCI,85921
-xinference/model/llm/utils.py,sha256=JxvhC1u4Ku3tQxXrINmSWLhCoKPlywSzkKi-wMWKmFk,31079
+xinference/model/llm/llm_family.json,sha256=otMgqAExkRqPC0yV_si0yc6sNhPEIxQjDrSg31Er9F8,152586
+xinference/model/llm/llm_family.py,sha256=aQZPE2gj5YfII2mhgbaSs1MfllKnUXd8xbflKr_YdFs,37357
+xinference/model/llm/llm_family_modelscope.json,sha256=5cyM-y0HtG_fBfC289IU97iasfjZZgUefn5UmVmg7TY,91282
+xinference/model/llm/memory.py,sha256=PTD8m6TCZVU1zrwc9wepX9cUjCqAXBENj6X7tjua0to,10207
+xinference/model/llm/utils.py,sha256=0SindpTW6dUWn17E-Ne9scnSfPOovb53sIuc9zxIFfo,31653
 xinference/model/llm/ggml/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
 xinference/model/llm/ggml/chatglm.py,sha256=qqCxVvPp4CZq0z6MuwTcVUMEkQJDjjXm6naJ6WfGOl0,16208
 xinference/model/llm/ggml/llamacpp.py,sha256=7Fvt2h7AJ8P6xVRztcw0aJgteagjCeOZJ7RniN6VYqs,13345
@@ -81,9 +82,10 @@ xinference/model/llm/pytorch/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKt
 xinference/model/llm/pytorch/baichuan.py,sha256=JBHldfdjUlSV44HB_td6402CwryDm61rlzw4D_flKHI,2820
 xinference/model/llm/pytorch/chatglm.py,sha256=meFb4P1MXw63IhoDQwmFGczTXa3vEpv1L1Eo9XIifG8,9624
 xinference/model/llm/pytorch/compression.py,sha256=U0vMJ-JaBt4oC2LffgWg6HbPj1CeUi_YdwVbjDd0mRA,8112
-xinference/model/llm/pytorch/core.py,sha256=kWBtqxtKH7n-Cy6YbD9DuQJNVEqGgCrIQiNvC_m9bAI,19937
+xinference/model/llm/pytorch/core.py,sha256=VGJfZ92e3SftQXrOwxiYMUSF1YnWSUU8Z8WHr6yzpJ8,19984
 xinference/model/llm/pytorch/deepseek_vl.py,sha256=T9DKP4cULvRaHSiU08lOWd_j6mt8b3ZIBByneZ0jY8U,11498
 xinference/model/llm/pytorch/falcon.py,sha256=POSP7vzRJaM5PjvX8dh60jNDXgnCwktwSmeZ7kypQU0,4499
+xinference/model/llm/pytorch/intern_vl.py,sha256=MlP7vcp0qu7ehSg3Z7_qe18aiepi3KKjN9N9P-qVTwM,13166
 xinference/model/llm/pytorch/internlm2.py,sha256=vjspoc2VHbuD1JaUtjt0sOq9MwvRr2OD3_tKQhBVUPc,7244
 xinference/model/llm/pytorch/llama_2.py,sha256=HMhUmn4oYW2maeSMIr1yY7jlAOMD0OVAxnF0dnRWmio,3710
 xinference/model/llm/pytorch/omnilmm.py,sha256=4r6pipch1LU1FPA80sOCE7Z0k3TO_J8CIT7pmVmWKEM,5664
@@ -94,9 +96,9 @@ xinference/model/llm/pytorch/yi_vl.py,sha256=MljT7tpgFIhL6n5rdoS3hmq_u0rtHRE6cxX
 xinference/model/llm/sglang/__init__.py,sha256=-sjSIQ4K6w-TEzx49kVaWeWC443fnZqODU91GCQ_JNo,581
 xinference/model/llm/sglang/core.py,sha256=RGHy6t9n0c4zL6Uha8P7t-qPvisPyulFVHw-8Aq8CJ0,14046
 xinference/model/llm/vllm/__init__.py,sha256=h_JgzSqV5lP6vQ6XX_17kE4IY4BRnvKta_7VLQAL1ms,581
-xinference/model/llm/vllm/core.py,sha256=XJpp9pPzYJwoFtTiFsETYRhf_Yemzux9_hvS3om0aDo,21543
+xinference/model/llm/vllm/core.py,sha256=_cM-NEnDtpGVHMgHxr-uxV0sA-xXUjDsL5i__cpHQEU,21639
 xinference/model/rerank/__init__.py,sha256=BXIL1uu3ZpZHX9bODhW9lxKUXudZE7-OkXFmmM5rpMU,2817
-xinference/model/rerank/core.py,sha256=UVfue73hHE9UL5c-X7OajZfTR_mLTv673RLFWZAfWV4,9665
+xinference/model/rerank/core.py,sha256=BLIIStjxUFghSFoxCimet88ghqGwmVaskOYdVRxKdpI,10572
 xinference/model/rerank/custom.py,sha256=NKk7jA7p4xkuwS5WoOs2SY2wdnoAVpyCjBTvv317bBw,3917
 xinference/model/rerank/model_spec.json,sha256=LCiiCdNz4NYt9vKVnHffk3ZpwvgzzHxe4zsaxOqxL18,1367
 xinference/model/rerank/model_spec_modelscope.json,sha256=vSSC0aWy_DHnNDzzBcMWr2pqdISDmPS95FtD_YfMmn4,1275
@@ -107,19 +109,19 @@ xinference/thirdparty/deepseek_vl/models/__init__.py,sha256=gVJoBKpRfny6w_QpTrTh
 xinference/thirdparty/deepseek_vl/models/clip_encoder.py,sha256=tn-uTCAcb63WOX6cB0rl2ynOsum23xy1tAvBqPbIHHo,8197
 xinference/thirdparty/deepseek_vl/models/image_processing_vlm.py,sha256=Nk3rK0eCwl-hoYeOrv1iiG7Zv3KlZFF2AAzVXN2wRT0,6796
 xinference/thirdparty/deepseek_vl/models/modeling_vlm.py,sha256=guGUqPuN_sjanlyP9D7N58tZwTw8CTy9AN-54_uEM9E,5705
-xinference/thirdparty/deepseek_vl/models/processing_vlm.py,sha256=CpyiwwJL0vvClTwM5mdbrgZeKc7UOS7qeRVRcAurhwc,13076
+xinference/thirdparty/deepseek_vl/models/processing_vlm.py,sha256=xytEqLoEKoKis-RBHjdQKHvhsSmT9YuSbaVX6F3CBUg,13076
 xinference/thirdparty/deepseek_vl/models/projector.py,sha256=BXPx4jj4jIDdZYZJ_mo3SmzgaTAVC7BxQMuU1wEzdIA,3625
 xinference/thirdparty/deepseek_vl/models/sam.py,sha256=pHdqAwPSurRyRuwmMQmriThz8eyPFiaofaLGGwpLDuU,20344
-xinference/thirdparty/deepseek_vl/models/siglip_vit.py,sha256=k5ac1YZK4an46m0AHhydlI6169GXNpUjMmrYr2vQ-Bc,24087
+xinference/thirdparty/deepseek_vl/models/siglip_vit.py,sha256=G0NdFVBvU4FxLu5I20rw--Ypjlx_2TvjRPW3Dhany7I,24088
 xinference/thirdparty/deepseek_vl/utils/__init__.py,sha256=u8C--egJMT_DDm0JG3frXGSQNNv6cfIAumPhghzkxhk,1091
 xinference/thirdparty/deepseek_vl/utils/conversation.py,sha256=7oITA8TsIuOfumVZ8hcoDpA6dm1jz7n8-VXPe_7-4YA,11702
 xinference/thirdparty/deepseek_vl/utils/io.py,sha256=HQqLoONZF4nl1E9oMrm7_2LwU8ZSh1C9htm7vLN4YfI,2751
 xinference/thirdparty/llava/__init__.py,sha256=UlCNtyBVX645CB6S6LfyYkTfQVO18_a8e9SmR7cHExA,41
 xinference/thirdparty/llava/conversation.py,sha256=_OKwx8mkikwfsuTVR3BtvolqNcwY1X3DP7X9TlmR31g,7758
-xinference/thirdparty/llava/mm_utils.py,sha256=QRMsFAeOV5sLoWdx-XDsfuoLcv8_rNgDLpkfNby_p0M,3882
+xinference/thirdparty/llava/mm_utils.py,sha256=f0JFOhbdFyLjA8W6VRDsGn5Xwnz4fWjY4Gyt3Apzk7k,3883
 xinference/thirdparty/llava/model/__init__.py,sha256=KlZDBvQbrUBrHix1PQTmAmq5dJHCipCKOlcpLZUJZso,60
 xinference/thirdparty/llava/model/constants.py,sha256=fFKjeLBhxYo3UGBhA6DbS0LGApXmxJqeQ0iMPaPh480,140
-xinference/thirdparty/llava/model/llava_arch.py,sha256=la9YRdFHh9bCki7m7UXZFvXdlPki096gHdewRYWXo3A,17198
+xinference/thirdparty/llava/model/llava_arch.py,sha256=I2SkH4mTNdCsXgaXGAOKsABTfJzSXj9KC2IYOOjOw08,17198
 xinference/thirdparty/llava/model/llava_llama.py,sha256=cXGSrL2bcy2FIAxSCQ25BhSbk6JYOs0uvPD7_zH7ZBI,5678
 xinference/thirdparty/llava/model/clip_encoder/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 xinference/thirdparty/llava/model/clip_encoder/builder.py,sha256=D1d2_hW2V1kT4jn_-spbYH3CRysyqZGPKsvb6Aai3QY,315
@@ -127,7 +129,7 @@ xinference/thirdparty/llava/model/clip_encoder/clip_encoder.py,sha256=ieNXVeV1UF
 xinference/thirdparty/llava/model/multimodal_projector/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 xinference/thirdparty/llava/model/multimodal_projector/builder.py,sha256=URP051uhGLWqC6ERsqjn2MiQDNYzOoQ6EJEcB-Z0RzM,1933
 xinference/thirdparty/omnilmm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-xinference/thirdparty/omnilmm/chat.py,sha256=z7mDC5ZsTIZDVq5lM3S6pI_Oa4Hnw1bDE8LHAki0OuI,6767
+xinference/thirdparty/omnilmm/chat.py,sha256=oi0I-yVkvvnW_Vqj-iQzKaPUbGJ__EBn1we9AHlaGn4,6784
 xinference/thirdparty/omnilmm/constants.py,sha256=bq6C4oIANqwTtt5G7sjJUdRlLyTa-_urOHMJN7VBmU4,84
 xinference/thirdparty/omnilmm/conversation.py,sha256=m0lzft61Sw1Z4RxEpucweeQvV0PZEyrvh0mXL7Psh8I,13276
 xinference/thirdparty/omnilmm/utils.py,sha256=VUMXIgq1tx3qwDmS2VewsXra6XqCUArQB0ik0_qk7x0,3989
@@ -15421,9 +15423,9 @@ xinference/web/ui/node_modules/yargs-parser/package.json,sha256=BSwbOzgetKXMK4u0
 xinference/web/ui/node_modules/yocto-queue/package.json,sha256=6U1XHQPGXJTqsiFvT953ORihUtXTblZy4fXBWP9qxC0,725
 xinference/web/ui/node_modules/yup/package.json,sha256=xRFSROB9NKxqSWHEVFvSTsPs9Ll074uo8OS1zEw0qhA,1206
 xinference/web/ui/node_modules/yup/node_modules/type-fest/package.json,sha256=JTv2zTTVgxQ2H82m1-6qEpdMv08lHjFx4Puf_MsbB_Q,1134
-xinference-0.11.1.dist-info/LICENSE,sha256=QwcOLU5TJoTeUhuIXzhdCEEDDvorGiC6-3YTOl4TecE,11356
-xinference-0.11.1.dist-info/METADATA,sha256=hBtPsJXz35UCfyMXv7FPwPVxQYV-WdSKX0Mf5Ze9p-8,15533
-xinference-0.11.1.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-xinference-0.11.1.dist-info/entry_points.txt,sha256=-lDyyzqWMFQF0Rgm7VxBNz0V-bMBMQLRR3pvQ-Y8XTY,226
-xinference-0.11.1.dist-info/top_level.txt,sha256=L1rQt7pl6m8tmKXpWVHzP-GtmzAxp663rXxGE7qnK00,11
-xinference-0.11.1.dist-info/RECORD,,
+xinference-0.11.2.post1.dist-info/LICENSE,sha256=QwcOLU5TJoTeUhuIXzhdCEEDDvorGiC6-3YTOl4TecE,11356
+xinference-0.11.2.post1.dist-info/METADATA,sha256=gwgZIL72lulJsk-4NasdYTQc5vxs5ibTFRjrf9TMgJc,15548
+xinference-0.11.2.post1.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+xinference-0.11.2.post1.dist-info/entry_points.txt,sha256=-lDyyzqWMFQF0Rgm7VxBNz0V-bMBMQLRR3pvQ-Y8XTY,226
+xinference-0.11.2.post1.dist-info/top_level.txt,sha256=L1rQt7pl6m8tmKXpWVHzP-GtmzAxp663rXxGE7qnK00,11
+xinference-0.11.2.post1.dist-info/RECORD,,