xinference 1.2.1__py3-none-any.whl → 1.3.0__py3-none-any.whl

xinference/model/llm/transformers/utils.py

@@ -193,16 +193,14 @@ def _get_pad_param(seq_len_idx: int, pad_len: int) -> Tuple:
 
 def _merge_kv_cache(
     xinf_model_obj: "PytorchModel",
-    past_kv: Tuple[Tuple[torch.Tensor]],
-    new_kv: Tuple[Tuple[torch.Tensor]],
-):
+    past_cache: DynamicCache,
+    new_cache: DynamicCache,
+) -> DynamicCache:
     from torch.nn.functional import pad
 
     _, seq_len_idx = xinf_model_obj.get_batch_size_and_seq_len_indexes_from_kv()
-    past_cache = DynamicCache.from_legacy_cache(past_kv)
-    new_cache = DynamicCache.from_legacy_cache(new_kv)
-    past_seq_len = past_kv[0][0].shape[seq_len_idx]
-    new_seq_len = new_kv[0][0].shape[seq_len_idx]
+    past_seq_len = past_cache[0][0].shape[seq_len_idx]
+    new_seq_len = new_cache[0][0].shape[seq_len_idx]
     if past_seq_len != new_seq_len:
         padding_target = new_cache if past_seq_len > new_seq_len else past_cache
         padding_len = abs(past_seq_len - new_seq_len)
@@ -219,8 +217,12 @@ def _merge_kv_cache(
     for idx in range(len(past_cache)):
         k1, k2 = new_cache.key_cache[idx], past_cache.key_cache[idx]
         v1, v2 = new_cache.value_cache[idx], past_cache.value_cache[idx]
-        ret_kv.update(torch.cat((k1, k2), 0), torch.cat((v1, v2), 0), idx)
-    return ret_kv.to_legacy_cache()
+        ret_kv.update(
+            torch.cat((k1, k2), 0).contiguous(),
+            torch.cat((v1, v2), 0).contiguous(),
+            idx,
+        )
+    return ret_kv
 
 
 def get_batch_size_and_seq_len_from_kv_cache(kv, xinf_model_obj: "PytorchModel"):
@@ -228,6 +230,15 @@ def get_batch_size_and_seq_len_from_kv_cache(kv, xinf_model_obj: "PytorchModel")
     return kv[0][0].shape[bs_idx], kv[0][0].shape[seq_len_idx] + 1
 
 
+def convert_to_cache_cls(cache) -> DynamicCache:
+    """
+    Compatible with some old models
+    """
+    if isinstance(cache, tuple):
+        return DynamicCache.from_legacy_cache(cache)
+    return cache
+
+
 @torch.inference_mode()
 def _batch_inference_one_step_internal(
     xinf_model_obj: "PytorchModel",
@@ -269,7 +280,7 @@ def _batch_inference_one_step_internal(
         out = model(**prefill_kws, use_cache=True)
 
         logits = out.logits
-        past_key_values = out.past_key_values
+        past_key_values = convert_to_cache_cls(out.past_key_values)
 
         for i, r in enumerate(prefill_reqs):
             (
@@ -317,7 +328,7 @@ def _batch_inference_one_step_internal(
         )
         out = model(**inf_kws, use_cache=True, past_key_values=past_key_values)
         logits = out.logits
-        past_key_values = out.past_key_values
+        past_key_values = convert_to_cache_cls(out.past_key_values)
 
         for i, r in enumerate(valid_req_list):
             (

xinference/model/llm/utils.py

@@ -11,16 +11,28 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import base64
 import functools
 import json
 import logging
 import os
+import re
 import time
 import typing
 import uuid
 from io import BytesIO
-from typing import AsyncGenerator, Dict, Iterator, List, Optional, Tuple, cast
+from typing import (
+    Any,
+    AsyncGenerator,
+    Dict,
+    Iterable,
+    Iterator,
+    List,
+    Optional,
+    Tuple,
+    cast,
+)
 
 import requests
 from PIL import Image
@@ -42,6 +54,7 @@ from .llm_family import (
     _get_cache_dir,
     get_cache_status,
 )
+from .reasoning_parsers.abs_reasoning_parsers import ReasoningParser
 
 logger = logging.getLogger(__name__)
 
@@ -64,6 +77,18 @@ LLAMA3_TOOL_CALL_FAMILY = [
     "llama-3.1-instruct",
 ]
 
+DEEPSEEK_TOOL_CALL_FAMILY = [
+    "deepseek-r1-distill-qwen",
+    "deepseek-r1-distill-llama",
+]
+
+TOOL_CALL_FAMILY = (
+    QWEN_TOOL_CALL_FAMILY
+    + GLM4_TOOL_CALL_FAMILY
+    + LLAMA3_TOOL_CALL_FAMILY
+    + DEEPSEEK_TOOL_CALL_FAMILY
+)
+
 QWEN_TOOL_CALL_SYMBOLS = ["<tool_call>", "</tool_call>"]
 
 
@@ -297,6 +322,7 @@ class ChatModelMixin:
     def _to_chat_completion_chunks(
         cls,
         chunks: Iterator[CompletionChunk],
+        reasoning_parse: Optional[ReasoningParser] = None,
     ) -> Iterator[ChatCompletionChunk]:
         for i, chunk in enumerate(chunks):
             if i == 0:
@@ -308,41 +334,102 @@ class ChatModelMixin:
             else:
                 yield cls._to_chat_completion_chunk(chunk)
 
+    @classmethod
+    def _tools_to_messages_for_deepseek(
+        cls, messages: List[dict], tools: Iterable[dict]
+    ):
+        # deepseek integrates tool calls into messages
+        # we follow the chat template rule to integrate tools into messages
+        tool_call_message: Dict[str, Any] = {
+            "role": "assistant",
+            "content": None,
+            "tool_calls": [],
+        }
+
+        for tool in tools:
+            function_name = tool["function"]["name"]
+            parameters = tool["function"].get("parameters", {}).get("properties", {})
+            function_args_json = json.dumps(parameters)
+
+            tool_call_message["tool_calls"].append(
+                {
+                    "type": "function",
+                    "function": {
+                        "name": function_name,
+                        "arguments": function_args_json,
+                    },
+                }
+            )
+
+        messages.append(tool_call_message)
+
     @classmethod
     async def _async_to_chat_completion_chunks(
         cls,
         chunks: AsyncGenerator[CompletionChunk, None],
+        reasoning_parser: Optional[ReasoningParser] = None,
     ) -> AsyncGenerator[ChatCompletionChunk, None]:
         i = 0
+        previous_text = ""
+        current_text = ""
         async for chunk in chunks:
             if i == 0:
-                yield cls._get_first_chat_completion_chunk(chunk)
-            # usage
-            choices = chunk.get("choices")
-            if not choices:
-                yield cls._get_final_chat_completion_chunk(chunk)
+                chunk = cls._get_first_chat_completion_chunk(chunk)
+            elif not chunk.get("choices"):
+                # usage
+                chunk = cls._get_final_chat_completion_chunk(chunk)
             else:
-                yield cls._to_chat_completion_chunk(chunk)
+                chunk = cls._to_chat_completion_chunk(chunk)
+            if reasoning_parser is not None:
+                choices = chunk.get("choices")
+                for choice in choices:
+                    delta = choice.get("delta")
+                    if not delta:
+                        continue
+                    current_text = previous_text + delta.get("content")
+                    choice[
+                        "delta"
+                    ] = reasoning_parser.extract_reasoning_content_streaming(
+                        previous_text=previous_text,
+                        current_text=current_text,
+                        delta=delta,
+                    )
+                    previous_text = current_text
+            yield chunk
             i += 1
 
     @staticmethod
-    def _to_chat_completion(completion: Completion) -> ChatCompletion:
-        return {
-            "id": "chat" + completion["id"],
-            "object": "chat.completion",
-            "created": completion["created"],
-            "model": completion["model"],
-            "choices": [
+    def _to_chat_completion(
+        completion: Completion, reasoning_parser: Optional[ReasoningParser] = None
+    ) -> ChatCompletion:
+        choices = []
+        for i, choice in enumerate(completion["choices"]):
+            content = choice["text"]
+            reasoning_content = None
+
+            if reasoning_parser is not None:
+                reasoning_content, content = reasoning_parser.extract_reasoning_content(
+                    choice
+                )
+
+            message = {"role": "assistant", "content": content}
+
+            # add only reasoning_content is None
+            if reasoning_content is not None:
+                message["reasoning_content"] = reasoning_content
+
+            choices.append(
                 {
                     "index": i,
-                    "message": {
-                        "role": "assistant",
-                        "content": choice["text"],
-                    },
+                    "message": message,
                     "finish_reason": choice["finish_reason"],
                 }
-                for i, choice in enumerate(completion["choices"])
-            ],
+            )
+        return {
+            "id": "chat" + completion["id"],
+            "object": "chat.completion",
+            "created": completion["created"],
+            "model": choices,
             "usage": completion["usage"],
         }
 
@@ -401,6 +488,61 @@ class ChatModelMixin:
         except Exception:
             return [(text, None, None)]
 
+    @classmethod
+    def _eval_deepseek_chat_arguments(cls, c) -> List[Tuple]:
+        """
+        Parses tool calls from deepseek-r1 format and removes duplicates.
+
+        Returns:
+        List[Tuple[Optional[str], Optional[str], Optional[dict]]]
+        - (None, function_name, arguments) if successfully parsed.
+        - (content, None, None) if parsing failed (content is raw JSON text).
+
+        Example input:
+        <｜tool▁call｜>get_current_weather
+        ```json
+        {"location": "tokyo", "unit": "fahrenheit"}
+        ```
+
+        Output:
+        [
+            (None, "get_current_weather", {"location": "tokyo", "unit": "fahrenheit"})
+        ]
+        """
+
+        text = c["choices"][0]["text"]
+
+        pattern = r"<｜tool▁call｜>(\w+)\s*```json\s*(.*?)\s*```"
+        matches = re.findall(pattern, text, re.DOTALL)
+
+        if not matches:
+            return [(text, None, None)]
+
+        tool_calls = set()  # Used for deduplication
+        results = []
+
+        for function_name, args_json in matches:
+            try:
+                arguments = json.loads(args_json)
+                # Convert dictionary to frozenset for deduplication
+                arguments_hashable = frozenset(arguments.items())
+                tool_call_tuple = (None, function_name, arguments)
+            except json.JSONDecodeError:
+                tool_call_tuple = (
+                    args_json,
+                    None,
+                    None,
+                )  # If parsing fails, treat as raw content
+                arguments_hashable = None  # No need for hashing
+
+            # Avoid duplicate entries
+            dedup_key = (function_name, arguments_hashable)
+            if dedup_key not in tool_calls:
+                tool_calls.add(dedup_key)
+                results.append(tool_call_tuple)
+
+        return results
+
     @classmethod
     def _eval_tool_arguments(cls, model_family, c):
         family = model_family.model_family or model_family.model_name
@@ -410,6 +552,8 @@ class ChatModelMixin:
             result = cls._eval_qwen_chat_arguments(c)
         elif family in LLAMA3_TOOL_CALL_FAMILY:
             result = cls._eval_llama3_chat_arguments(c)
+        elif family in DEEPSEEK_TOOL_CALL_FAMILY:
+            result = cls._eval_deepseek_chat_arguments(c)
         else:
             raise Exception(
                 f"Model {model_family.model_name} is not support tool calls."

xinference/model/llm/vllm/core.py

@@ -43,7 +43,10 @@ from ....types import (
 )
 from .. import LLM, LLMFamilyV1, LLMSpecV1
 from ..llm_family import CustomLLMFamilyV1
+from ..reasoning_parsers import deepseek_r1_reasoning_parser  # noqa: F401
+from ..reasoning_parsers.abs_reasoning_parsers import ReasoningParserManager
 from ..utils import (
+    DEEPSEEK_TOOL_CALL_FAMILY,
     QWEN_TOOL_CALL_FAMILY,
     QWEN_TOOL_CALL_SYMBOLS,
     ChatModelMixin,
@@ -71,6 +74,7 @@ class VLLMModelConfig(TypedDict, total=False):
     limit_mm_per_prompt: Optional[Dict[str, int]]
     guided_decoding_backend: Optional[str]
     scheduling_policy: Optional[str]
+    reasoning_content: bool
 
 
 class VLLMGenerateConfig(TypedDict, total=False):
@@ -175,6 +179,8 @@ if VLLM_INSTALLED and vllm.__version__ >= "0.5.1":
     VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-v2-chat")
     VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-v2-chat-0628")
     VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-v2.5")
+    VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-v3")
+    VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-r1")
 
 if VLLM_INSTALLED and vllm.__version__ >= "0.5.3":
     VLLM_SUPPORTED_CHAT_MODELS.append("gemma-2-it")
@@ -185,9 +191,11 @@ if VLLM_INSTALLED and vllm.__version__ > "0.5.3":
     VLLM_SUPPORTED_MODELS.append("llama-3.1")
     VLLM_SUPPORTED_CHAT_MODELS.append("llama-3.1-instruct")
     VLLM_SUPPORTED_CHAT_MODELS.append("llama-3.3-instruct")
+    VLLM_SUPPORTED_CHAT_MODELS.append("deepseek-r1-distill-llama")
 
 if VLLM_INSTALLED and vllm.__version__ >= "0.6.1":
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("internvl2")
+    VLLM_SUPPORTED_VISION_MODEL_LIST.append("InternVL2.5")
 
 if VLLM_INSTALLED and vllm.__version__ >= "0.6.2":
     VLLM_SUPPORTED_CHAT_MODELS.append("minicpm3-4b")
@@ -198,6 +206,15 @@ if VLLM_INSTALLED and vllm.__version__ >= "0.6.3":
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("qwen2-vl-instruct")
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("QvQ-72B-Preview")
 
+if VLLM_INSTALLED and vllm.__version__ >= "0.7.0":
+    VLLM_SUPPORTED_CHAT_MODELS.append("internlm3-instruct")
+
+if VLLM_INSTALLED and vllm.__version__ >= "0.7.2":
+    VLLM_SUPPORTED_VISION_MODEL_LIST.append("qwen2.5-vl-instruct")
+
+if VLLM_INSTALLED and vllm.__version__ >= "0.7.3":
+    VLLM_SUPPORTED_CHAT_MODELS.append("qwen-2.5-instruct-1m")
+
 
 class VLLMModel(LLM):
     def __init__(
@@ -226,6 +243,7 @@ class VLLMModel(LLM):
         self.lora_modules = peft_model
         self.lora_requests: List[LoRARequest] = []
         self._xavier_config = None
+        self.reasoning_parser = None
 
     def set_xavier_config(self, value: Optional[Dict]):
         self._xavier_config = value  # type: ignore
@@ -254,6 +272,16 @@ class VLLMModel(LLM):
             multiprocessing.set_start_method("fork", force=True)
 
         self._model_config = self._sanitize_model_config(self._model_config)
+        reasoning_content = self._model_config.pop("reasoning_content")
+
+        # Initialize reasoning parser if model has reasoning ability
+        if "reasoning" in self.model_family.model_ability and reasoning_content:
+            module_name = self.model_family.model_family or self.model_family.model_name
+            self.reasoning_parser = ReasoningParserManager.get_parser(module_name)
+            self.reasoning_parser = self.reasoning_parser(
+                self.model_family.reasoning_start_tag,
+                self.model_family.reasoning_end_tag,
+            )
         if self.lora_modules is None:
             self.lora_requests = []
         else:
@@ -360,6 +388,7 @@ class VLLMModel(LLM):
         model_config.setdefault("quantization", None)
         model_config.setdefault("max_model_len", None)
         model_config.setdefault("guided_decoding_backend", "outlines")
+        model_config.setdefault("reasoning_content", False)
         # Add scheduling policy if vLLM version is 0.6.3 or higher
         if vllm.__version__ >= "0.6.3":
             model_config.setdefault("scheduling_policy", "fcfs")
@@ -807,8 +836,11 @@ class VLLMChatModel(VLLMModel, ChatModelMixin):
         tools = generate_config.pop("tools", []) if generate_config else None
         model_family = self.model_family.model_family or self.model_family.model_name
         full_context_kwargs = {}
-        if tools and model_family in QWEN_TOOL_CALL_FAMILY:
-            full_context_kwargs["tools"] = tools
+        if tools:
+            if model_family in QWEN_TOOL_CALL_FAMILY:
+                full_context_kwargs["tools"] = tools
+            elif model_family in DEEPSEEK_TOOL_CALL_FAMILY:
+                self._tools_to_messages_for_deepseek(messages, tools)
         assert self.model_family.chat_template is not None
         full_prompt = self.get_full_context(
             messages, self.model_family.chat_template, **full_context_kwargs
@@ -824,7 +856,7 @@ class VLLMChatModel(VLLMModel, ChatModelMixin):
             assert isinstance(agen, AsyncGenerator)
             if tools:
                 return self._async_to_tool_completion_chunks(agen)
-            return self._async_to_chat_completion_chunks(agen)
+            return self._async_to_chat_completion_chunks(agen, self.reasoning_parser)
         else:
             c = await self.async_generate(
                 full_prompt, generate_config, request_id=request_id
@@ -832,7 +864,7 @@ class VLLMChatModel(VLLMModel, ChatModelMixin):
             assert not isinstance(c, AsyncGenerator)
             if tools:
                 return self._tool_calls_completion(self.model_family, self.model_uid, c)
-            return self._to_chat_completion(c)
+            return self._to_chat_completion(c, self.reasoning_parser)
 
 
 class VLLMVisionModel(VLLMModel, ChatModelMixin):

xinference/model/llm/vllm/xavier/executor.py

@@ -14,7 +14,7 @@
 from typing import TYPE_CHECKING, List, Optional, Set, Tuple, Union
 
 import xoscar as xo
-from vllm.executor.gpu_executor import GPUExecutorAsync
+from vllm.executor.mp_distributed_executor import MultiprocessingDistributedExecutor
 from vllm.model_executor.layers.sampler import SamplerOutput
 from vllm.sequence import ExecuteModelRequest, PoolerOutput
 from vllm.utils import is_pin_memory_available
@@ -24,7 +24,7 @@ if TYPE_CHECKING:
     from .scheduler import XavierScheduler
 
 
-class XavierExecutor(GPUExecutorAsync):
+class XavierExecutor(MultiprocessingDistributedExecutor):
     scheduler: Optional[List["XavierScheduler"]] = None
 
     def _init_executor(self) -> None:

xinference/model/llm/vllm/xavier/scheduler.py

@@ -21,7 +21,7 @@ import xoscar as xo
 from vllm.config import CacheConfig, LoRAConfig, SchedulerConfig
 from vllm.core.block.interfaces import Block
 from vllm.core.interfaces import BlockSpaceManager
-from vllm.core.scheduler import Scheduler, SchedulerOutputs
+from vllm.core.scheduler import ScheduledSequenceGroup, Scheduler, SchedulerOutputs
 from vllm.sequence import (
     SequenceData,
     SequenceGroup,
@@ -216,7 +216,7 @@ class XavierScheduler(Scheduler):
         """Xinference Change!!!
         Additional data structures required by Xavier.
         """
-        scheduled_seq_groups = []
+        scheduled_seq_groups: List[ScheduledSequenceGroup] = []
         has_transferring = False
 
         # Create input data structures.
@@ -288,7 +288,7 @@ class XavierScheduler(Scheduler):
                     has_transferring = True
                     continue
                 else:
-                    scheduled_seq_groups.append(seq_group)
+                    scheduled_seq_groups.append(scheduled_seq_group)
 
             if self.cache_config.enable_prefix_caching:
                 common_computed_block_nums = (

xinference/thirdparty/internvl/conversation.py

@@ -227,7 +227,7 @@ class Conversation:
 
             return ret
         elif self.sep_style == SeparatorStyle.INTERNVL_ZH:
-            seps = [self.sep, self.sep2]
+            seps = [self.sep2, self.sep]
             ret = self.system_message + seps[0]
             for i, (role, message) in enumerate(self.messages):
                 if message:
@@ -330,6 +330,19 @@ def get_conv_template(name: str) -> Conversation:
     return conv_templates[name].copy()
 
 
+# InternVL-Chat-V1-1 template
+register_conv_template(
+    Conversation(
+        name='internvl_zh',
+        system_template='',
+        roles=('<human>', '<bot>'),
+        sep_style=SeparatorStyle.INTERNVL_ZH,
+        sep='</s>',
+        sep2=' ',
+    )
+)
+
+
 # Both Hermes-2 and internlm2-chat are chatml-format conversation templates. The difference
 # is that during training, the preprocessing function for the Hermes-2 template doesn't add
 # <s> at the beginning of the tokenized sequence, while the internlm2-chat template does.
@@ -344,12 +357,6 @@ register_conv_template(
         roles=('<|im_start|>user\n', '<|im_start|>assistant\n'),
         sep_style=SeparatorStyle.MPT,
         sep='<|im_end|>',
-        stop_token_ids=[
-            2,
-            6,
-            7,
-            8,
-        ],
         stop_str='<|endoftext|>',
     )
 )
@@ -365,11 +372,6 @@ register_conv_template(
         roles=('<|im_start|>user\n', '<|im_start|>assistant\n'),
         sep_style=SeparatorStyle.MPT,
         sep='<|im_end|>',
-        stop_token_ids=[
-            2,
-            92543,
-            92542
-        ]
     )
 )
 
@@ -384,10 +386,17 @@ register_conv_template(
         roles=('<|user|>\n', '<|assistant|>\n'),
         sep_style=SeparatorStyle.MPT,
         sep='<|end|>',
-        stop_token_ids=[
-            2,
-            32000,
-            32007
-        ]
     )
 )
+
+
+register_conv_template(
+    Conversation(
+        name='internvl2_5',
+        system_template='<|im_start|>system\n{system_message}',
+        system_message='你是书生·万象，英文名是InternVL，是由上海人工智能实验室、清华大学及多家合作单位联合开发的多模态大语言模型。',
+        roles=('<|im_start|>user\n', '<|im_start|>assistant\n'),
+        sep_style=SeparatorStyle.MPT,
+        sep='<|im_end|>\n',
+    )
+)

xinference/types.py

@@ -177,6 +177,7 @@ class Completion(TypedDict):
 
 class ChatCompletionMessage(TypedDict):
     role: str
+    reasoning_content: NotRequired[str]
     content: Optional[str]
     user: NotRequired[str]
     tool_calls: NotRequired[List]
@@ -199,6 +200,7 @@ class ChatCompletion(TypedDict):
 
 class ChatCompletionChunkDelta(TypedDict):
     role: NotRequired[str]
+    reasoning_content: NotRequired[str]
     content: NotRequired[str]
     tool_calls: NotRequired[List[ToolCalls]]
 

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

@@ -1,14 +1,14 @@
 {
   "files": {
-    "main.css": "./static/css/main.51a587ff.css",
-    "main.js": "./static/js/main.b0936c54.js",
+    "main.css": "./static/css/main.f8177338.css",
+    "main.js": "./static/js/main.ad42919c.js",
     "static/media/icon.webp": "./static/media/icon.4603d52c63041e5dfbfd.webp",
     "index.html": "./index.html",
-    "main.51a587ff.css.map": "./static/css/main.51a587ff.css.map",
-    "main.b0936c54.js.map": "./static/js/main.b0936c54.js.map"
+    "main.f8177338.css.map": "./static/css/main.f8177338.css.map",
+    "main.ad42919c.js.map": "./static/js/main.ad42919c.js.map"
   },
   "entrypoints": [
-    "static/css/main.51a587ff.css",
-    "static/js/main.b0936c54.js"
+    "static/css/main.f8177338.css",
+    "static/js/main.ad42919c.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.b0936c54.js"></script><link href="./static/css/main.51a587ff.css" rel="stylesheet"></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.ad42919c.js"></script><link href="./static/css/main.f8177338.css" rel="stylesheet"></head><body><noscript>You need to enable JavaScript to run this app.</noscript><div id="root"></div></body></html>

xinference/web/ui/build/static/css/main.f8177338.css

@@ -0,0 +1,2 @@
+.container{cursor:pointer;display:block}.container,.descriptionCard{border-radius:20px;height:300px;position:relative;width:300px}.descriptionCard{left:-1px;padding:20px;top:-1px}.cardTitle{display:flex;justify-content:space-between}.iconButtonBox{align-items:center;display:flex}.drawerCard{min-height:100%;padding:20px 80px 0;position:relative;width:60vw}.p{-webkit-line-clamp:4;-webkit-box-orient:vertical;display:-webkit-box;font-size:14px;overflow:hidden;padding:0 10px;text-overflow:ellipsis;word-break:break-word}.pasteText{color:#1976d2;cursor:pointer;font-size:18px!important;margin-inline:10px}.pasteText:hover{color:#1976d2b3}.copyToCommandLine{color:#1976d2;cursor:pointer;font-size:16px!important}.copyToCommandLine:hover{color:#1976d2b3}.formContainer{height:80%;overflow:scroll;padding:0 10px 160px}.buttonsContainer{align-items:center;bottom:50px;display:flex;justify-content:space-between;left:100px;position:absolute;right:100px}.buttonContainer{background-color:initial;border-width:0;width:45%}.buttonItem{border:1px solid #e5e7eb;border-radius:4px;cursor:pointer;padding:5px;width:100%}.buttonItem:hover{border-color:#888}.instructionText{color:#666;font-size:12px;font-style:italic;margin:30px 0;text-align:center}.iconRow{bottom:20px;justify-content:space-between;left:20px;position:absolute;right:20px}.iconItem,.iconRow{align-items:center;display:flex}.iconItem{flex-direction:column;margin:20px}.boldIconText{font-size:1.2em;font-weight:700}.muiIcon{font-size:1.5em}.smallText{font-size:.8em}.dialogBox{background-color:#fff;height:607px;margin:32px;overflow-x:scroll;width:1241px}.dialogTitle{color:#000;display:flex;justify-content:space-between;padding:20px 20px 7px}.dialogTitle-model_name{font-size:18px;font-weight:700}.pathBox{cursor:pointer;overflow:hidden;text-overflow:ellipsis;white-space:nowrap;width:160px}.pathBox2{width:300px}.empty{color:#555;font-size:20px;left:50%;position:absolute;top:30%;-webkit-transform:translate(-50%);transform:translate(-50%)}.deleteDialog{align-items:center;display:flex}.warningIcon{color:#ed6c02;margin-right:10px}.jsonDialog{background-color:#fff;border-radius:8px;color:#000;display:flex;flex-direction:column;padding:10px 30px}.jsonDialog-title{align-items:center;display:flex;justify-content:space-between;margin:10px 0 20px}.title-name{font-size:16px;font-weight:700}.main-box{height:500px;width:700px}.textarea-box{border:1px solid #ddd;border-radius:5px;color:#444;height:100%;padding:5px 10px;resize:none;width:100%}.but-box{display:flex;justify-content:end;margin-top:20px}.drawer{bottom:0;left:0;opacity:0;position:fixed;right:0;top:0;transition:visibility .3s ease,opacity .3s ease;visibility:hidden;z-index:1000}.drawer.open{opacity:1;visibility:visible}.drawer-overlay{background-color:rgba(0,0,0,.5);left:0;z-index:999}.drawer-content,.drawer-overlay{bottom:0;position:absolute;right:0;top:0}.drawer-content{background-color:#fff;box-shadow:-2px 0 10px rgba(0,0,0,.1);overflow-y:auto;-webkit-transform:translateX(100%);transform:translateX(100%);transition:-webkit-transform .3s ease;transition:transform .3s ease;transition:transform .3s ease,-webkit-transform .3s ease;z-index:1000}.drawer.open .drawer-content{-webkit-transform:translateX(0);transform:translateX(0)}.copyText{color:#666;cursor:pointer;font-size:14px!important}.copyText:hover{color:#1976d2}.formBox{max-height:80vh;max-width:50vw;min-width:50vw;overflow:auto;padding:40px 20px 0 0;position:relative;transition:all .4s ease-in-out}.broaden{max-width:100%;min-width:100%;padding-right:0}.show-json{align-items:center;display:flex;right:60px;top:90px}.icon,.show-json{position:absolute}.icon{cursor:pointer;margin-left:20px;right:-40px}.icon:hover{color:#1976d2}.arrow{font-size:24px!important}.jsonBox{min-height:80vh;position:relative;transition:all .4s ease-in-out;width:100%}.hide{overflow:hidden;-webkit-transform:translate(30vw);transform:translate(30vw);width:0}.checkboxWrapper{align-items:center;display:flex;flex-wrap:wrap;width:100%}.jsonBox-header{align-items:center;display:flex;justify-content:space-between}.jsonBox-title{font-weight:700;line-height:40px}.textarea{background-color:initial;border:1px solid #ddd;border-radius:5px;color:#444;height:calc(100% - 40px);padding:5px 10px;resize:none;width:100%}.addBtn{margin-left:20px!important}.item{border:1px solid #ddd;border-radius:10px;margin:10px 50px 0;overflow:hidden;padding:20px;position:relative}.item:hover .deleteBtn{-webkit-transform:translateX(-50px);transform:translateX(-50px)}.deleteBtn{background-color:#1976d2;border-radius:25px;height:50px;line-height:70px;position:absolute;right:20px;text-align:center;top:calc(50% - 25px);-webkit-transform:translateX(80px);transform:translateX(80px);transition:all .3s ease-in-out;width:50px}.deleteBtn:hover{box-shadow:0 0 10px #aaa;cursor:pointer}.deleteIcon{color:#fff;font-size:28px!important}.chat_template_box{align-items:start;display:flex;gap:10px}.chat_template_test{width:30%}.chat_template_test_mainBox{border:1px solid #ccc;border-radius:4px;height:137px;overflow:scroll;padding:10px}.chat_template_test_tip{color:rgba(0,0,0,.6);font-size:10px;margin:4px 14px 0}.test_res_box{border:1px solid #ddd;border-radius:4px;margin-top:5px;min-height:55px;padding:10px}.css-19qh8xo-MuiInputBase-input-MuiOutlinedInput-input.Mui-disabled{-webkit-text-fill-color:#000!important}
+/*# sourceMappingURL=main.f8177338.css.map*/

xinference/web/ui/build/static/css/main.f8177338.css.map

@@ -0,0 +1 @@
+{"version":3,"file":"static/css/main.f8177338.css","mappings":"AAAA,WAKE,cAAe,CAJf,aAMF,CACA,4BAFE,kBAAmB,CAFnB,YAAa,CAFb,iBAAkB,CAClB,WAaF,CARA,iBAGE,SAAU,CAGV,YAAa,CAJb,QAMF,CACA,WACE,YAAa,CACb,6BACF,CACA,eAEE,kBAAmB,CADnB,YAEF,CACA,YAGE,eAAgB,CADhB,mBAAoB,CADpB,iBAAkB,CAGlB,UACF,CACA,GAEE,oBAAqB,CACrB,2BAA4B,CAF5B,mBAAoB,CAMpB,cAAe,CAHf,eAAgB,CAIhB,cAAiB,CAHjB,sBAAuB,CACvB,qBAGF,CACA,WAEE,aAAc,CACd,cAAe,CAFf,wBAA0B,CAG1B,kBACF,CACA,iBACE,eACF,CACA,mBAEE,aAAc,CACd,cAAe,CAFf,wBAGF,CACA,yBACE,eACF,CACA,eACE,UAAW,CACX,eAAgB,CAEhB,oBACF,CACA,kBAOE,kBAAmB,CALnB,WAAY,CAGZ,YAAa,CACb,6BAA8B,CAH9B,UAAW,CAFX,iBAAkB,CAGlB,WAIF,CACA,iBAGE,wBAA6B,CAD7B,cAAiB,CADjB,SAGF,CACA,YAOE,wBAAqB,CAHrB,iBAAkB,CAHlB,cAAe,CAEf,WAAY,CADZ,UAMF,CACA,kBACE,iBACF,CACA,iBAEE,UAAc,CADd,cAAe,CAEf,iBAAkB,CAClB,aAAc,CACd,iBACF,CACA,SAEE,WAAY,CAIZ,6BAA8B,CAH9B,SAAU,CAFV,iBAAkB,CAGlB,UAIF,CACA,mBAFE,kBAAmB,CAFnB,YASF,CALA,UAEE,qBAAsB,CAEtB,WACF,CACA,cAEE,eAAgB,CADhB,eAEF,CACA,SACE,eACF,CACA,WACE,cACF,CACA,WAGE,qBAAsB,CADtB,YAAa,CAEb,WAAY,CACZ,iBAAkB,CAJlB,YAKF,CACA,aAIE,UAAW,CAHX,YAAa,CACb,6BAA8B,CAC9B,qBAEF,CACA,wBACE,cAAe,CACf,eACF,CACA,SAEE,cAAe,CACf,eAAgB,CAEhB,sBAAuB,CADvB,kBAAmB,CAHnB,WAKF,CACA,UACE,WACF,CACA,OAKE,UAAW,CADX,cAAe,CAFf,QAAS,CADT,iBAAkB,CAElB,OAAQ,CAGR,iCAA6B,CAA7B,yBACF,CACA,cAEE,kBAAmB,CADnB,YAEF,CACA,aAEE,aAAuB,CADvB,iBAEF,CACA,YAIE,qBAAsB,CAEtB,iBAAkB,CADlB,UAAW,CAJX,YAAa,CACb,qBAAsB,CACtB,iBAIF,CACA,kBAGE,kBAAmB,CAFnB,YAAa,CACb,6BAA8B,CAE9B,kBACF,CACA,YACE,cAAe,CACf,eACF,CACA,UAEE,YAAa,CADb,WAEF,CACA,cAIE,qBAAsB,CACtB,iBAAkB,CAElB,UAAW,CALX,WAAY,CACZ,gBAAiB,CAGjB,WAAY,CALZ,UAOF,CACA,SACE,YAAa,CACb,mBAAoB,CACpB,eACF,CACA,QAKE,QAAS,CACT,MAAO,CAEP,SAAU,CANV,cAAe,CAEf,OAAQ,CADR,KAAM,CAMN,+CAAmD,CAFnD,iBAAkB,CANlB,YASF,CACA,aAEE,SAAU,CADV,kBAEF,CACA,gBAME,+BAAoC,CADpC,MAAO,CAEP,WACF,CACA,gCALE,QAAS,CAHT,iBAAkB,CAElB,OAAQ,CADR,KAkBF,CAXA,gBAKE,qBAAuB,CAEvB,qCAA0C,CAG1C,eAAgB,CADhB,kCAA2B,CAA3B,0BAA2B,CAD3B,qCAA+B,CAA/B,6BAA+B,CAA/B,wDAA+B,CAF/B,YAKF,CACA,6BACE,+BAAwB,CAAxB,uBACF,CC/OA,UAEE,UAAW,CACX,cAAe,CAFf,wBAGF,CAEA,gBACE,aACF,CCRA,SAIE,eAAgB,CAFhB,cAAe,CACf,cAAe,CAEf,aAAc,CACd,qBAAsB,CALtB,iBAAkB,CAMlB,8BACF,CAEA,SACE,cAAe,CACf,cAAe,CACf,eACF,CAEA,WAEE,kBAAmB,CADnB,YAAa,CAIb,UAAW,CADX,QAEF,CAEA,iBALE,iBAUF,CALA,MAGE,cAAe,CACf,gBAAiB,CAFjB,WAGF,CAEA,YACE,aACF,CAEA,OACE,wBACF,CAEA,SAEE,eAAgB,CADhB,iBAAkB,CAGlB,8BAAgC,CADhC,UAEF,CAEA,MAGE,eAAgB,CADhB,iCAA6B,CAA7B,yBAA6B,CAD7B,OAGF,CAEA,iBAGE,kBAAmB,CAFnB,YAAa,CACb,cAAe,CAEf,UACF,CAEA,gBAGE,kBAAmB,CAFnB,YAAa,CACb,6BAEF,CAEA,eAEE,eAAgB,CADhB,gBAEF,CAEA,UAQE,wBAA6B,CAJ7B,qBAAsB,CACtB,iBAAkB,CAElB,UAAW,CALX,wBAAyB,CACzB,gBAAiB,CAGjB,WAAY,CALZ,UAQF,CAEA,QACE,0BACF,CAEA,MAEE,qBAAsB,CAGtB,kBAAmB,CAFnB,kBAAmB,CAGnB,eAAgB,CAFhB,YAAa,CAHb,iBAMF,CAEA,uBACE,mCAA4B,CAA5B,2BACF,CAEA,WAUE,wBAAyB,CADzB,kBAAmB,CAJnB,WAAY,CAGZ,gBAAiB,CAPjB,iBAAkB,CAClB,UAAW,CAKX,iBAAkB,CAJlB,oBAAqB,CAGrB,kCAA2B,CAA3B,0BAA2B,CAK3B,8BAAgC,CAPhC,UAQF,CAEA,iBAEE,wBAAyB,CADzB,cAEF,CAEA,YAEE,UAAW,CADX,wBAEF,CAEA,mBAEE,iBAAkB,CADlB,YAAa,CAEb,QACF,CAEA,oBACE,SACF,CAEA,4BAGE,qBAAsB,CACtB,iBAAkB,CAHlB,YAAa,CAIb,eAAgB,CAHhB,YAIF,CAEA,wBAGE,oBAAyB,CAFzB,cAAe,CACf,iBAEF,CAEA,cACE,qBAAsB,CAItB,iBAAkB,CADlB,cAAe,CAFf,eAAgB,CAChB,YAGF,CAEA,oEACE,sCACF","sources":["scenes/launch_model/styles/modelCardStyle.css","components/copyComponent/style.css","scenes/register_model/styles/registerModelStyle.css"],"sourcesContent":[".container {\n  display: block;\n  position: relative;\n  width: 300px;\n  height: 300px;\n  cursor: pointer;\n  border-radius: 20px;\n}\n.descriptionCard {\n  position: relative;\n  top: -1px;\n  left: -1px;\n  width: 300px;\n  height: 300px;\n  padding: 20px;\n  border-radius: 20px;\n}\n.cardTitle {\n  display: flex;\n  justify-content: space-between;\n}\n.iconButtonBox {\n  display: flex;\n  align-items: center;\n}\n.drawerCard {\n  position: relative;\n  padding: 20px 80px 0;\n  min-height: 100%;\n  width: 60vw;\n}\n.p {\n  display: -webkit-box;\n  -webkit-line-clamp: 4;\n  -webkit-box-orient: vertical;\n  overflow: hidden;\n  text-overflow: ellipsis;\n  word-break: break-word;\n  font-size: 14px;\n  padding: 0px 10px;\n}\n.pasteText {\n  font-size: 18px !important;\n  color: #1976d2;\n  cursor: pointer;\n  margin-inline: 10px;\n}\n.pasteText:hover {\n  color: #1976d2b3;\n}\n.copyToCommandLine {\n  font-size: 16px !important;\n  color: #1976d2;\n  cursor: pointer;\n}\n.copyToCommandLine:hover {\n  color: #1976d2b3;\n}\n.formContainer {\n  height: 80%;\n  overflow: scroll;\n  padding: 0 10px;\n  padding-bottom: 160px;\n}\n.buttonsContainer {\n  position: absolute;\n  bottom: 50px;\n  left: 100px;\n  right: 100px;\n  display: flex;\n  justify-content: space-between;\n  align-items: center;\n}\n.buttonContainer {\n  width: 45%;\n  border-width: 0px;\n  background-color: transparent;\n}\n.buttonItem {\n  cursor: pointer;\n  width: 100%;\n  padding: 5px;\n  border-radius: 4px;\n  border: 1px solid #e5e7eb;\n  border-width: 1px;\n  border-color: #e5e7eb;\n}\n.buttonItem:hover {\n  border-color: #888;\n}\n.instructionText {\n  font-size: 12px;\n  color: #666666;\n  font-style: italic;\n  margin: 30px 0;\n  text-align: center;\n}\n.iconRow {\n  position: absolute;\n  bottom: 20px;\n  left: 20px;\n  right: 20px;\n  display: flex;\n  justify-content: space-between;\n  align-items: center;\n}\n.iconItem {\n  display: flex;\n  flex-direction: column;\n  align-items: center;\n  margin: 20px;\n}\n.boldIconText {\n  font-weight: bold;\n  font-size: 1.2em;\n}\n.muiIcon {\n  font-size: 1.5em;\n}\n.smallText {\n  font-size: 0.8em;\n}\n.dialogBox {\n  width: 1241px;\n  height: 607px;\n  background-color: #fff;\n  margin: 32px;\n  overflow-x: scroll;\n}\n.dialogTitle {\n  display: flex;\n  justify-content: space-between;\n  padding: 20px 20px 7px;\n  color: #000;\n}\n.dialogTitle-model_name {\n  font-size: 18px;\n  font-weight: 700;\n}\n.pathBox {\n  width: 160px;\n  cursor: pointer;\n  overflow: hidden;\n  white-space: nowrap;\n  text-overflow: ellipsis;\n}\n.pathBox2 {\n  width: 300px;\n}\n.empty {\n  position: absolute;\n  left: 50%;\n  top: 30%;\n  font-size: 20px;\n  color: #555;\n  transform: translate(-50%, 0);\n}\n.deleteDialog {\n  display: flex;\n  align-items: center;\n}\n.warningIcon {\n  margin-right: 10px;\n  color: rgb(237, 108, 2);\n}\n.jsonDialog {\n  display: flex;\n  flex-direction: column;\n  padding: 10px 30px;\n  background-color: #fff;\n  color: #000;\n  border-radius: 8px;\n}\n.jsonDialog-title {\n  display: flex;\n  justify-content: space-between;\n  align-items: center;\n  margin: 10px 0 20px 0;\n}\n.title-name {\n  font-size: 16px;\n  font-weight: 700;\n}\n.main-box {\n  width: 700px;\n  height: 500px;\n}\n.textarea-box {\n  width: 100%;\n  height: 100%;\n  padding: 5px 10px;\n  border: 1px solid #ddd;\n  border-radius: 5px;\n  resize: none;\n  color: #444;\n}\n.but-box {\n  display: flex;\n  justify-content: end;\n  margin-top: 20px;\n}\n.drawer {\n  z-index: 1000;\n  position: fixed;\n  top: 0;\n  right: 0;\n  bottom: 0;\n  left: 0;\n  visibility: hidden;\n  opacity: 0;\n  transition: visibility 0.3s ease, opacity 0.3s ease;\n}\n.drawer.open {\n  visibility: visible;\n  opacity: 1;\n}\n.drawer-overlay {\n  position: absolute;\n  top: 0;\n  right: 0;\n  bottom: 0;\n  left: 0;\n  background-color: rgba(0, 0, 0, 0.5);\n  z-index: 999;\n}\n.drawer-content {\n  position: absolute;\n  top: 0;\n  right: 0;\n  bottom: 0;\n  background-color: white;\n  z-index: 1000;\n  box-shadow: -2px 0 10px rgba(0, 0, 0, 0.1);\n  transition: transform 0.3s ease;\n  transform: translateX(100%);\n  overflow-y: auto;\n}\n.drawer.open .drawer-content {\n  transform: translateX(0);\n}\n",".copyText {\n  font-size: 14px !important;\n  color: #666;\n  cursor: pointer;\n}\n\n.copyText:hover {\n  color: #1976d2;\n}\n",".formBox {\n  position: relative;\n  max-width: 50vw;\n  min-width: 50vw;\n  max-height: 80vh;\n  overflow: auto;\n  padding: 40px 20px 0 0;\n  transition: all 0.4s ease-in-out;\n}\n\n.broaden {\n  max-width: 100%;\n  min-width: 100%;\n  padding-right: 0;\n}\n\n.show-json {\n  display: flex;\n  align-items: center;\n  position: absolute;\n  top: 90px;\n  right: 60px;\n}\n\n.icon {\n  position: absolute;\n  right: -40px;\n  cursor: pointer;\n  margin-left: 20px;\n}\n\n.icon:hover {\n  color: #1976d2;\n}\n\n.arrow {\n  font-size: 24px !important;\n}\n\n.jsonBox {\n  position: relative;\n  min-height: 80vh;\n  width: 100%;\n  transition: all 0.4s ease-in-out;\n}\n\n.hide {\n  width: 0;\n  transform: translate(30vw, 0);\n  overflow: hidden;\n}\n\n.checkboxWrapper {\n  display: flex;\n  flex-wrap: wrap;\n  align-items: center;\n  width: 100%;\n}\n\n.jsonBox-header {\n  display: flex;\n  justify-content: space-between;\n  align-items: center;\n}\n\n.jsonBox-title {\n  line-height: 40px;\n  font-weight: 700;\n}\n\n.textarea {\n  width: 100%;\n  height: calc(100% - 40px);\n  padding: 5px 10px;\n  border: 1px solid #ddd;\n  border-radius: 5px;\n  resize: none;\n  color: #444;\n  background-color: transparent;\n}\n\n.addBtn {\n  margin-left: 20px !important;\n}\n\n.item {\n  position: relative;\n  border: 1px solid #ddd;\n  margin: 10px 50px 0;\n  padding: 20px;\n  border-radius: 10px;\n  overflow: hidden;\n}\n\n.item:hover .deleteBtn {\n  transform: translateX(-50px);\n}\n\n.deleteBtn {\n  position: absolute;\n  right: 20px;\n  top: calc(50% - 25px);\n  width: 50px;\n  height: 50px;\n  transform: translateX(80px);\n  text-align: center;\n  line-height: 70px;\n  border-radius: 25px;\n  background-color: #1976d2;\n  transition: all 0.3s ease-in-out;\n}\n\n.deleteBtn:hover {\n  cursor: pointer;\n  box-shadow: 0 0 10px #aaa;\n}\n\n.deleteIcon {\n  font-size: 28px !important;\n  color: #fff;\n}\n\n.chat_template_box {\n  display: flex;\n  align-items: start;\n  gap: 10px;\n}\n\n.chat_template_test {\n  width: 30%;\n}\n\n.chat_template_test_mainBox {\n  height: 137px;\n  padding: 10px;\n  border: 1px solid #ccc;\n  border-radius: 4px;\n  overflow: scroll;\n}\n\n.chat_template_test_tip {\n  font-size: 10px;\n  margin: 4px 14px 0;\n  color: rgba(0, 0, 0, 0.6);\n}\n\n.test_res_box {\n  border: 1px solid #ddd;\n  min-height: 55px;\n  padding: 10px;\n  margin-top: 5px;\n  border-radius: 4px;\n}\n\n.css-19qh8xo-MuiInputBase-input-MuiOutlinedInput-input.Mui-disabled {\n  -webkit-text-fill-color: #000 !important;\n}\n"],"names":[],"sourceRoot":""}