xinference 0.10.1__py3-none-any.whl → 0.10.2__py3-none-any.whl

xinference/model/rerank/model_spec_modelscope.json

@@ -1,6 +1,7 @@
 [
   {
     "model_name": "bge-reranker-base",
+    "type": "normal",
     "language": ["en", "zh"],
     "model_id": "Xorbits/bge-reranker-base",
     "model_revision": "v0.0.1",
@@ -8,16 +9,39 @@
   },
   {
     "model_name": "bge-reranker-large",
+    "type": "normal",
     "language": ["en", "zh"],
     "model_id": "Xorbits/bge-reranker-large",
     "model_revision": "v0.0.1",
     "model_hub": "modelscope"
   },
-    {
+  {
     "model_name": "bce-reranker-base_v1",
+    "type": "normal",
     "language": ["en", "zh"],
     "model_id": "maidalun/bce-reranker-base_v1",
     "model_revision": "v0.0.1",
     "model_hub": "modelscope"
+  },
+  {
+    "model_name": "bge-reranker-v2-m3",
+    "type": "normal",
+    "language": ["en", "zh", "multilingual"],
+    "model_id": "AI-ModelScope/bge-reranker-v2-m3",
+    "model_hub": "modelscope"
+  },
+  {
+    "model_name": "bge-reranker-v2-gemma",
+    "type": "LLM-based",
+    "language": ["en", "zh", "multilingual"],
+    "model_id": "AI-ModelScope/bge-reranker-v2-gemma",
+    "model_hub": "modelscope"
+  },
+  {
+    "model_name": "bge-reranker-v2-minicpm-layerwise",
+    "type": "LLM-based layerwise",
+    "language": ["en", "zh", "multilingual"],
+    "model_id": "zfffff/bge-reranker-v2-minicpm-layerwise",
+    "model_hub": "modelscope"
   }
 ]

xinference/model/utils.py

@@ -17,7 +17,7 @@ import os
 import shutil
 from json import JSONDecodeError
 from pathlib import Path
-from typing import Any, Callable, Dict, Optional, Tuple
+from typing import Any, Callable, Dict, Optional, Tuple, Union
 
 from fsspec import AbstractFileSystem
 
@@ -415,3 +415,14 @@ def select_device(device):
             raise ValueError(f"{device} is unavailable in your environment")
 
     return device
+
+
+def convert_float_to_int_or_str(model_size: float) -> Union[int, str]:
+    """convert float to int or string
+
+    if float can be presented as int, convert it to int, otherwise convert it to string
+    """
+    if int(model_size) == model_size:
+        return int(model_size)
+    else:
+        return str(model_size)

xinference/types.py

@@ -450,3 +450,58 @@ class CreateChatCompletion(
     CreateChatCompletionOpenAI,
 ):
     pass
+
+
+class LoRA:
+    def __init__(self, lora_name: str, local_path: str):
+        self.lora_name = lora_name
+        self.local_path = local_path
+
+    def to_dict(self):
+        return {
+            "lora_name": self.lora_name,
+            "local_path": self.local_path,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict):
+        return cls(
+            lora_name=data["lora_name"],
+            local_path=data["local_path"],
+        )
+
+
+class PeftModelConfig:
+    def __init__(
+        self,
+        peft_model: Optional[List[LoRA]] = None,
+        image_lora_load_kwargs: Optional[Dict] = None,
+        image_lora_fuse_kwargs: Optional[Dict] = None,
+    ):
+        self.peft_model = peft_model
+        self.image_lora_load_kwargs = image_lora_load_kwargs
+        self.image_lora_fuse_kwargs = image_lora_fuse_kwargs
+
+    def to_dict(self):
+        return {
+            "lora_list": [lora.to_dict() for lora in self.peft_model]
+            if self.peft_model
+            else None,
+            "image_lora_load_kwargs": self.image_lora_load_kwargs,
+            "image_lora_fuse_kwargs": self.image_lora_fuse_kwargs,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict):
+        peft_model_list = data.get("lora_list", None)
+        peft_model = (
+            [LoRA.from_dict(lora_dict) for lora_dict in peft_model_list]
+            if peft_model_list is not None
+            else None
+        )
+
+        return cls(
+            peft_model=peft_model,
+            image_lora_load_kwargs=data.get("image_lora_load_kwargs"),
+            image_lora_fuse_kwargs=data.get("image_lora_fuse_kwargs"),
+        )

xinference/utils.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+
 import torch
 
 

xinference/web/ui/build/asset-manifest.json

@@ -1,11 +1,11 @@
 {
   "files": {
-    "main.js": "./static/js/main.76ef2b17.js",
+    "main.js": "./static/js/main.26fdbfbe.js",
     "static/media/icon.webp": "./static/media/icon.4603d52c63041e5dfbfd.webp",
     "index.html": "./index.html",
-    "main.76ef2b17.js.map": "./static/js/main.76ef2b17.js.map"
+    "main.26fdbfbe.js.map": "./static/js/main.26fdbfbe.js.map"
   },
   "entrypoints": [
-    "static/js/main.76ef2b17.js"
+    "static/js/main.26fdbfbe.js"
   ]
 }

xinference/web/ui/build/index.html

@@ -1 +1 @@
-<!doctype html><html lang="en"><head><meta charset="utf-8"/><link rel="icon" href="./favicon.svg"/><meta name="viewport" content="width=device-width,initial-scale=1"/><meta name="theme-color" content="#000000"/><meta name="description" content="Web site created using create-react-app"/><link rel="apple-touch-icon" href="./logo192.png"/><link rel="manifest" href="./manifest.json"/><title>Xinference</title><script defer="defer" src="./static/js/main.76ef2b17.js"></script></head><body><noscript>You need to enable JavaScript to run this app.</noscript><div id="root"></div></body></html>
+<!doctype html><html lang="en"><head><meta charset="utf-8"/><link rel="icon" href="./favicon.svg"/><meta name="viewport" content="width=device-width,initial-scale=1"/><meta name="theme-color" content="#000000"/><meta name="description" content="Web site created using create-react-app"/><link rel="apple-touch-icon" href="./logo192.png"/><link rel="manifest" href="./manifest.json"/><title>Xinference</title><script defer="defer" src="./static/js/main.26fdbfbe.js"></script></head><body><noscript>You need to enable JavaScript to run this app.</noscript><div id="root"></div></body></html>