sglang 0.3.5.post2__py3-none-any.whl → 0.3.6__py3-none-any.whl

sglang/test/runners.py

@@ -58,6 +58,28 @@ def get_top_logprobs(logits, k):
     return logprobs
 
 
+def _get_sentence_transformer_embedding_model(model_path, torch_dtype):
+    from sentence_transformers import SentenceTransformer
+    from sentence_transformers.util import is_sentence_transformer_model
+
+    if is_sentence_transformer_model(model_path):
+        model = SentenceTransformer(
+            model_path,
+            model_kwargs={"torch_dtype": torch_dtype},
+        )
+    else:  # if no pre-trained sentence-transformers model
+        from sentence_transformers import models
+
+        word_embedding_model = models.Transformer(model_path).to(dtype=torch_dtype)
+        pooling_model = models.Pooling(
+            word_embedding_model.get_word_embedding_dimension(),
+            pooling_mode="lasttoken",
+        )
+        model = SentenceTransformer(modules=[word_embedding_model, pooling_model])
+
+    return model.cuda()
+
+
 @dataclass
 class ModelOutput:
     output_strs: List[str] = None
@@ -114,12 +136,9 @@ class HFRunner:
                 low_cpu_mem_usage=True,
             ).cuda()
         elif self.model_type == "embedding":
-            from sentence_transformers import SentenceTransformer
-
-            self.model = SentenceTransformer(
-                model_path,
-                model_kwargs={"torch_dtype": torch_dtype},
-            ).cuda()
+            self.model = _get_sentence_transformer_embedding_model(
+                model_path, torch_dtype
+            )
         elif self.model_type == "reward":
             from transformers import AutoModelForSequenceClassification
 

sglang/test/srt/sampling/penaltylib/utils.py

@@ -1,7 +1,7 @@
 import dataclasses
 import enum
-import typing
 import unittest
+from typing import Dict, List, Optional, Set, Tuple, Type
 
 import torch
 
@@ -16,7 +16,7 @@ from sglang.srt.sampling.penaltylib.orchestrator import (
 class MockSamplingParams:
     frequency_penalty: float = 0.0
     min_new_tokens: int = 0
-    stop_token_ids: typing.List[int] = None
+    stop_token_ids: List[int] = None
     presence_penalty: float = 0.0
     repetition_penalty: float = 1.0
 
@@ -24,12 +24,12 @@ class MockSamplingParams:
 @dataclasses.dataclass
 class MockTokenizer:
     eos_token_id: int
-    additional_stop_token_ids: typing.Optional[typing.List[int]] = None
+    additional_stop_token_ids: Optional[List[int]] = None
 
 
 @dataclasses.dataclass
 class MockReq:
-    origin_input_ids: typing.List[int]
+    origin_input_ids: List[int]
     sampling_params: MockSamplingParams
     tokenizer: MockTokenizer
 
@@ -42,8 +42,8 @@ class StepType(enum.Enum):
 @dataclasses.dataclass
 class Step:
     type: StepType
-    token_ids: typing.List[int]
-    expected_tensors: typing.Dict[str, torch.Tensor]
+    token_ids: List[int]
+    expected_tensors: Dict[str, torch.Tensor]
     # assume initial logits are all 1
     expected_logits: torch.Tensor
 
@@ -52,7 +52,7 @@ class Step:
 class Subject:
     sampling_params: MockSamplingParams
     # first step must be input, which will be converted to Req
-    steps: typing.List[Step]
+    steps: List[Step]
     eos_token_id: int = -1
 
     def __post_init__(self):
@@ -66,7 +66,7 @@ class Subject:
                     f"Expected tensors keys must be the same for all steps. Got {self.steps[i].expected_tensors.keys()} for key={i} and {self.steps[0].expected_tensors.keys()}"
                 )
 
-    def tensor_keys(self, i: int = 0) -> typing.Set[str]:
+    def tensor_keys(self, i: int = 0) -> Set[str]:
         return set(self.steps[i].expected_tensors.keys())
 
     def to_req(self) -> MockReq:
@@ -80,7 +80,7 @@ class Subject:
 @dataclasses.dataclass
 class Case:
     enabled: bool
-    test_subjects: typing.List[Subject]
+    test_subjects: List[Subject]
 
     def __post_init__(self):
         # each test_subjects.steps should have the same expected_tensors.keys()
@@ -90,12 +90,12 @@ class Case:
                     f"Expected tensors keys must be the same for all test_subjects. Got {self.test_subjects[i].tensor_keys()} for key={i} and {self.test_subjects[0].tensor_keys()}"
                 )
 
-    def tensor_keys(self, i: int = 0) -> typing.List[str]:
+    def tensor_keys(self, i: int = 0) -> List[str]:
         return set(self.test_subjects[i].tensor_keys())
 
 
 class BaseBatchedPenalizerTest(unittest.TestCase):
-    Penalizer: typing.Type[_BatchedPenalizer]
+    Penalizer: Type[_BatchedPenalizer]
     device = "cuda"
     vocab_size = 5
 
@@ -115,7 +115,7 @@ class BaseBatchedPenalizerTest(unittest.TestCase):
         """
         return torch.tensor(data, **kwargs, device=self.device)
 
-    def create_test_subjects(self) -> typing.List[Subject]:
+    def create_test_subjects(self) -> List[Subject]:
         raise NotImplementedError()
 
     def create_test_cases(self):
@@ -127,7 +127,7 @@ class BaseBatchedPenalizerTest(unittest.TestCase):
 
     def _create_penalizer(
         self, case: Case
-    ) -> typing.Tuple[BatchedPenalizerOrchestrator, _BatchedPenalizer]:
+    ) -> Tuple[BatchedPenalizerOrchestrator, _BatchedPenalizer]:
         orchestrator = BatchedPenalizerOrchestrator(
             vocab_size=self.vocab_size,
             batch=_BatchLike(reqs=[subject.to_req() for subject in case.test_subjects]),
@@ -287,22 +287,24 @@ class BaseBatchedPenalizerTest(unittest.TestCase):
                         if i < len(subject.steps)
                     ]
 
-                    inputs: typing.List[typing.List[int]] = []
-                    outputs: typing.List[typing.List[int]] = []
+                    inputs: List[List[int]] = []
+                    outputs: List[List[int]] = []
                     for subject in filtered_subjects:
                         step = subject.steps[i]
                         if step.type == StepType.INPUT:
-                            inputs.append(step.token_ids)
-                            outputs.append([])
+                            raise NotImplementedError()
                         else:
                             inputs.append([])
                             outputs.append(step.token_ids)
 
-                    if any(inputs):
-                        orchestrator.cumulate_input_tokens(inputs)
-
                     if any(outputs):
-                        orchestrator.cumulate_output_tokens(outputs)
+                        for j in range(max(len(x) for x in outputs)):
+                            tmp_outputs = torch.tensor(
+                                [x[j] for x in outputs],
+                                dtype=torch.int32,
+                                device=orchestrator.device,
+                            )
+                            orchestrator.cumulate_output_tokens(tmp_outputs)
 
                     if penalizer.is_required():
                         self.assertTrue(penalizer.is_prepared())

sglang/test/test_utils.py

@@ -2,6 +2,7 @@
 
 import argparse
 import asyncio
+import copy
 import os
 import random
 import subprocess
@@ -438,18 +439,22 @@ def popen_launch_server(
         process = subprocess.Popen(command, stdout=None, stderr=None, env=env)
 
     start_time = time.time()
-    while time.time() - start_time < timeout:
-        try:
-            headers = {
-                "Content-Type": "application/json; charset=utf-8",
-                "Authorization": f"Bearer {api_key}",
-            }
-            response = requests.get(f"{base_url}/health_generate", headers=headers)
-            if response.status_code == 200:
-                return process
-        except requests.RequestException:
-            pass
-        time.sleep(10)
+    with requests.Session() as session:
+        while time.time() - start_time < timeout:
+            try:
+                headers = {
+                    "Content-Type": "application/json; charset=utf-8",
+                    "Authorization": f"Bearer {api_key}",
+                }
+                response = session.get(
+                    f"{base_url}/health_generate",
+                    headers=headers,
+                )
+                if response.status_code == 200:
+                    return process
+            except requests.RequestException:
+                pass
+            time.sleep(10)
     raise TimeoutError("Server failed to start within the timeout period.")
 
 
@@ -529,6 +534,7 @@ def run_bench_serving(
     random_input_len=4096,
     random_output_len=2048,
     disable_stream=False,
+    need_warmup=False,
 ):
     # Launch the server
     base_url = DEFAULT_URL_FOR_TEST
@@ -562,9 +568,14 @@ def run_bench_serving(
         disable_stream=disable_stream,
         disable_ignore_eos=False,
         extra_request_body=None,
+        profile=None,
     )
 
     try:
+        if need_warmup:
+            warmup_args = copy.deepcopy(args)
+            warmup_args.num_prompts = 16
+            run_benchmark(warmup_args)
         res = run_benchmark(args)
     finally:
         kill_child_process(process.pid, include_self=True)
@@ -573,11 +584,11 @@ def run_bench_serving(
     return res
 
 
-def run_bench_latency(model, other_args):
+def run_bench_one_batch(model, other_args):
     command = [
         "python3",
         "-m",
-        "sglang.bench_latency",
+        "sglang.bench_one_batch",
         "--model-path",
         model,
         "--batch-size",
@@ -664,7 +675,7 @@ def run_and_check_memory_leak(
     workload_func,
     disable_radix_cache,
     enable_mixed_chunk,
-    enable_overlap,
+    disable_overlap,
     chunked_prefill_size,
 ):
     other_args = ["--chunked-prefill-size", str(chunked_prefill_size)]
@@ -672,8 +683,8 @@ def run_and_check_memory_leak(
         other_args += ["--disable-radix-cache"]
     if enable_mixed_chunk:
         other_args += ["--enable-mixed-chunk"]
-    if enable_overlap:
-        other_args += ["--enable-overlap-schedule"]
+    if disable_overlap:
+        other_args += ["--disable-overlap-schedule"]
 
     model = DEFAULT_MODEL_NAME_FOR_TEST
     port = random.randint(4000, 5000)
@@ -725,7 +736,7 @@ def run_and_check_memory_leak(
 def run_mmlu_test(
     disable_radix_cache=False,
     enable_mixed_chunk=False,
-    enable_overlap=False,
+    disable_overlap=False,
     chunked_prefill_size=32,
 ):
     def workload_func(base_url, model):
@@ -748,7 +759,7 @@ def run_mmlu_test(
         workload_func,
         disable_radix_cache,
         enable_mixed_chunk,
-        enable_overlap,
+        disable_overlap,
         chunked_prefill_size,
     )
 

sglang/version.py

@@ -1 +1 @@
-__version__ = "0.3.5.post2"
+__version__ = "0.3.6"

{sglang-0.3.5.post2.dist-info → sglang-0.3.6.dist-info}/METADATA

@@ -1,8 +1,8 @@
 Metadata-Version: 2.1
 Name: sglang
-Version: 0.3.5.post2
+Version: 0.3.6
 Summary: SGLang is yet another fast serving framework for large language models and vision language models.
-License: Apache License
+License:                                  Apache License
                                    Version 2.0, January 2004
                                 http://www.apache.org/licenses/
         
@@ -215,74 +215,74 @@ Requires-Dist: requests
 Requires-Dist: tqdm
 Requires-Dist: numpy
 Requires-Dist: IPython
-Provides-Extra: all
-Requires-Dist: sglang[srt]; extra == "all"
-Requires-Dist: sglang[openai]; extra == "all"
-Requires-Dist: sglang[anthropic]; extra == "all"
-Requires-Dist: sglang[litellm]; extra == "all"
-Provides-Extra: all_hip
-Requires-Dist: sglang[srt_hip]; extra == "all-hip"
-Requires-Dist: sglang[openai]; extra == "all-hip"
-Requires-Dist: sglang[anthropic]; extra == "all-hip"
-Requires-Dist: sglang[litellm]; extra == "all-hip"
-Provides-Extra: all_xpu
-Requires-Dist: sglang[srt_xpu]; extra == "all-xpu"
-Requires-Dist: sglang[openai]; extra == "all-xpu"
-Requires-Dist: sglang[anthropic]; extra == "all-xpu"
-Requires-Dist: sglang[litellm]; extra == "all-xpu"
-Provides-Extra: anthropic
-Requires-Dist: anthropic>=0.20.0; extra == "anthropic"
-Provides-Extra: dev
-Requires-Dist: sglang[all]; extra == "dev"
-Requires-Dist: sglang[test]; extra == "dev"
-Provides-Extra: dev_hip
-Requires-Dist: sglang[all_hip]; extra == "dev-hip"
-Requires-Dist: sglang[test]; extra == "dev-hip"
-Provides-Extra: dev_xpu
-Requires-Dist: sglang[all_xpu]; extra == "dev-xpu"
-Requires-Dist: sglang[test]; extra == "dev-xpu"
-Provides-Extra: litellm
-Requires-Dist: litellm>=1.0.0; extra == "litellm"
-Provides-Extra: openai
-Requires-Dist: openai>=1.0; extra == "openai"
-Requires-Dist: tiktoken; extra == "openai"
-Provides-Extra: runtime_common
+Provides-Extra: runtime-common
 Requires-Dist: aiohttp; extra == "runtime-common"
 Requires-Dist: decord; extra == "runtime-common"
 Requires-Dist: fastapi; extra == "runtime-common"
-Requires-Dist: hf-transfer; extra == "runtime-common"
-Requires-Dist: huggingface-hub; extra == "runtime-common"
+Requires-Dist: hf_transfer; extra == "runtime-common"
+Requires-Dist: huggingface_hub; extra == "runtime-common"
 Requires-Dist: interegular; extra == "runtime-common"
 Requires-Dist: orjson; extra == "runtime-common"
+Requires-Dist: outlines<0.1.0,>=0.0.44; extra == "runtime-common"
 Requires-Dist: packaging; extra == "runtime-common"
 Requires-Dist: pillow; extra == "runtime-common"
 Requires-Dist: prometheus-client>=0.20.0; extra == "runtime-common"
 Requires-Dist: psutil; extra == "runtime-common"
 Requires-Dist: pydantic; extra == "runtime-common"
 Requires-Dist: python-multipart; extra == "runtime-common"
+Requires-Dist: pyzmq>=25.1.2; extra == "runtime-common"
 Requires-Dist: torchao; extra == "runtime-common"
 Requires-Dist: uvicorn; extra == "runtime-common"
 Requires-Dist: uvloop; extra == "runtime-common"
-Requires-Dist: pyzmq>=25.1.2; extra == "runtime-common"
-Requires-Dist: outlines<0.1.0,>=0.0.44; extra == "runtime-common"
 Requires-Dist: modelscope; extra == "runtime-common"
 Provides-Extra: srt
 Requires-Dist: sglang[runtime_common]; extra == "srt"
 Requires-Dist: torch; extra == "srt"
-Requires-Dist: vllm==0.6.3.post1; extra == "srt"
-Provides-Extra: srt_hip
+Requires-Dist: vllm>=0.6.3.post1; extra == "srt"
+Provides-Extra: srt-hip
 Requires-Dist: sglang[runtime_common]; extra == "srt-hip"
 Requires-Dist: torch; extra == "srt-hip"
 Requires-Dist: vllm==0.6.3.dev13; extra == "srt-hip"
-Provides-Extra: srt_xpu
+Provides-Extra: srt-xpu
 Requires-Dist: sglang[runtime_common]; extra == "srt-xpu"
+Provides-Extra: openai
+Requires-Dist: openai>=1.0; extra == "openai"
+Requires-Dist: tiktoken; extra == "openai"
+Provides-Extra: anthropic
+Requires-Dist: anthropic>=0.20.0; extra == "anthropic"
+Provides-Extra: litellm
+Requires-Dist: litellm>=1.0.0; extra == "litellm"
 Provides-Extra: test
 Requires-Dist: jsonlines; extra == "test"
 Requires-Dist: matplotlib; extra == "test"
 Requires-Dist: pandas; extra == "test"
-Requires-Dist: sentence-transformers; extra == "test"
+Requires-Dist: sentence_transformers; extra == "test"
 Requires-Dist: accelerate; extra == "test"
 Requires-Dist: peft; extra == "test"
+Provides-Extra: all
+Requires-Dist: sglang[srt]; extra == "all"
+Requires-Dist: sglang[openai]; extra == "all"
+Requires-Dist: sglang[anthropic]; extra == "all"
+Requires-Dist: sglang[litellm]; extra == "all"
+Provides-Extra: all-hip
+Requires-Dist: sglang[srt_hip]; extra == "all-hip"
+Requires-Dist: sglang[openai]; extra == "all-hip"
+Requires-Dist: sglang[anthropic]; extra == "all-hip"
+Requires-Dist: sglang[litellm]; extra == "all-hip"
+Provides-Extra: all-xpu
+Requires-Dist: sglang[srt_xpu]; extra == "all-xpu"
+Requires-Dist: sglang[openai]; extra == "all-xpu"
+Requires-Dist: sglang[anthropic]; extra == "all-xpu"
+Requires-Dist: sglang[litellm]; extra == "all-xpu"
+Provides-Extra: dev
+Requires-Dist: sglang[all]; extra == "dev"
+Requires-Dist: sglang[test]; extra == "dev"
+Provides-Extra: dev-hip
+Requires-Dist: sglang[all_hip]; extra == "dev-hip"
+Requires-Dist: sglang[test]; extra == "dev-hip"
+Provides-Extra: dev-xpu
+Requires-Dist: sglang[all_xpu]; extra == "dev-xpu"
+Requires-Dist: sglang[test]; extra == "dev-xpu"
 
 <div align="center"  id="sglangtop">
 <img src="https://raw.githubusercontent.com/sgl-project/sglang/main/assets/logo.png" alt="logo" width="400" margin="10px"></img>
@@ -323,7 +323,7 @@ The core features include:
 
 - **Fast Backend Runtime**: Provides efficient serving with RadixAttention for prefix caching, jump-forward constrained decoding, continuous batching, token attention (paged attention), tensor parallelism, FlashInfer kernels, chunked prefill, and quantization (INT4/FP8/AWQ/GPTQ).
 - **Flexible Frontend Language**: Offers an intuitive interface for programming LLM applications, including chained generation calls, advanced prompting, control flow, multi-modal inputs, parallelism, and external interactions.
-- **Extensive Model Support**: Supports a wide range of generative models (Llama, Gemma, Mistral, QWen, DeepSeek, LLaVA, etc.), embedding models (e5-mistral, gte) and reward models (Skywork), with easy extensibility for integrating new models.
+- **Extensive Model Support**: Supports a wide range of generative models (Llama, Gemma, Mistral, QWen, DeepSeek, LLaVA, etc.), embedding models (e5-mistral, gte, mcdse) and reward models (Skywork), with easy extensibility for integrating new models.
 - **Active Community**: SGLang is open-source and backed by an active community with industry adoption.
 
 ## Getting Started

{sglang-0.3.5.post2.dist-info → sglang-0.3.6.dist-info}/RECORD

@@ -1,15 +1,16 @@
 sglang/__init__.py,sha256=b_pqO9bR2fjK9En_tigfzKTiQzE8b_hUizY0DAKVk1M,1616
 sglang/api.py,sha256=3I9YUJNOeCqwKymZec2JR_agjTyKIx4XoT6IGdZ4_Cs,6953
-sglang/bench_latency.py,sha256=SSqZjcCNO88ExpT94qBZ5CmuA5o0T8wMTBnxLsNMqik,18259
-sglang/bench_offline_throughput.py,sha256=xBr7gI_ZbrpXXD72Nzu1F228oNyz1jggcblZCeUWJgw,9975
-sglang/bench_server_latency.py,sha256=N1MODIzcMk74yOWmY19d36aih3ewtHOemLxoieKtdhw,5866
-sglang/bench_serving.py,sha256=ytef89P9bqKRaMGXAqq69SmLTlNXWyHyhEraISLKYME,47975
-sglang/check_env.py,sha256=rGRABCgt-0SfUrow4px28b2P59aMn8eVTnN5eZc_a8s,5397
+sglang/bench_latency.py,sha256=oZjSAzX7dUiSu-zdz0dkyUPo-qAX_lsXFH1gf03akgI,76
+sglang/bench_offline_throughput.py,sha256=z6uA6Gxa_nFZa0cOXi7MJDuX82xcqk5WfqBMavd8a-s,10929
+sglang/bench_one_batch.py,sha256=Ww5Qd1ATaY8zw0mDEGoTYjwxMtxPKmpaHrIdjvS9iVE,15706
+sglang/bench_one_batch_server.py,sha256=nzeF_bcaXanQuYLBxAvd3OO4fwbKproMcahXdHIVR6w,5920
+sglang/bench_serving.py,sha256=hn5mihMey8Cik2nvwV30DUQ8C4Goxyt6BWm4YtyjIrI,50511
+sglang/check_env.py,sha256=nR2m0a9WbQmkimJihUx-Lqi7XjN0jyWTCO2vYyA7R2M,5356
 sglang/global_config.py,sha256=fnT0U9vlHdGaQFKN9tYTnUF4-eVW4HYQURd5zvPtrg0,1286
 sglang/launch_server.py,sha256=_XIqBcXArYtHTqilOFkYWKZBYXGCMHAxbYOST08LGj0,415
 sglang/launch_server_llavavid.py,sha256=tGc17S1vUfLwbi1GB26oOdXxTWr7gjlqpTrPnrMRNO8,1007
 sglang/utils.py,sha256=eCvD3fZCALr-MuyZxJL7HAeeqqpxAxf4LJrf7OiCbco,11547
-sglang/version.py,sha256=NlX-QUNR7ogIH-GcgzllsyHox7ItJoycFEUM_EYuhW4,28
+sglang/version.py,sha256=W_9dCm49nLvZulVAvvsafxLJjVBSKDBHz9K7szFZllo,22
 sglang/lang/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sglang/lang/chat_template.py,sha256=jprS3-In2FTUoedKwZg-HYvDwU8RTIYntOlf2zoN2sU,14814
 sglang/lang/choices.py,sha256=-W1DVw9N9ZliVpvmWrzIXG4cswAah8eMQrHWzkS3D8o,6234
@@ -27,38 +28,40 @@ sglang/lang/backend/vertexai.py,sha256=O-iBLD-y3vq80UxnrAoJri7bxpgd-_eakZ88Cf8bE
 sglang/srt/conversation.py,sha256=erz6wEXMcSmBlskuUhX2c-MT0EMyqyFpTem9PgastEE,21107
 sglang/srt/hf_transformers_utils.py,sha256=QbYVTnz0UdaXESPMAaq1OMzzznn95J_l08eXJuB68aU,6618
 sglang/srt/mm_utils.py,sha256=ml68nWUJhs_FS2FU1oB9UPHKZmF7P2DQHl1ddywn4ao,12272
-sglang/srt/server.py,sha256=JUYAE8MDGYou_HbmuR10QFZfg319fGt9VamskvBkpFo,28776
-sglang/srt/server_args.py,sha256=V8sx2oY0yphHC_uATwv4UTiLUFnvMQl85o6y5AyaoXM,30086
-sglang/srt/utils.py,sha256=jGSlxbvI50xEybdupDQNHpsCaF1U_5buADrD149766g,27013
+sglang/srt/model_parallel.py,sha256=QR-Alqo0sElDXPJ79N1PhUHHKiEHPQn3dyXduMP-SHQ,3664
+sglang/srt/server.py,sha256=caZPEoP3zdbEnQJnGzOEqvSdzSjsVUX8opSc-SplH2A,29709
+sglang/srt/server_args.py,sha256=1VhWGvMOtr7ozW2BJV8KInPyptzfh2UiBN4jqdDJYS8,30714
+sglang/srt/utils.py,sha256=5YIElk7hP1Zr7ff-jFXBUfM-acurnh5HR1ofC18FOTU,27540
 sglang/srt/configs/__init__.py,sha256=_usVIXHQjft4PAJ1Y-yGQOn2QNOv501GYMlQwpGXbns,208
 sglang/srt/configs/exaone.py,sha256=Duxd4yQoKy8GWEzZD_kCY_OzmN_67CTJL_Kgn0eXk3g,10731
 sglang/srt/configs/model_config.py,sha256=mBXeDfFUijQnxd38gVGJ6QxgsiitDklfHvbjYBJFKQY,9470
 sglang/srt/configs/qwen2vl.py,sha256=AYHuFgJ0bwhWYkD7S6fvP7yJejJnuhy4xp5Q2W-O6ps,4424
 sglang/srt/constrained/__init__.py,sha256=LHj0-NxDQ7S_N3Pc1gJ-FmIJVN_PTP9ytitWOICSMHk,691
 sglang/srt/constrained/base_grammar_backend.py,sha256=OPuBSd_F_fRwjVj6YFWBQuGeikj7UQtkTvc-JgEYt4I,2259
-sglang/srt/constrained/outlines_backend.py,sha256=J03QQiT9pkdXyoYGw3Rj6taEyWlIr4VCBvxQ3aMiB8A,5786
+sglang/srt/constrained/outlines_backend.py,sha256=i4dhg3hP406YHzEyP8x2FQmLlGEn8Uby51KNLAcdhak,6353
 sglang/srt/constrained/outlines_jump_forward.py,sha256=1fnYxlrc24xjcW3Wx59Hyg0L9hiHIVgMVUsld3UDfW4,6102
-sglang/srt/constrained/xgrammar_backend.py,sha256=wMWqkLN5KhnJXL6GBqbcrhxvAAMx60nG88KIBU1bFSc,4505
-sglang/srt/layers/activation.py,sha256=7VEkCrx2dvl629Lz0fkJcJfVoZA-ykEdkpTzKEc_drQ,5225
-sglang/srt/layers/layernorm.py,sha256=HCj8Y_X6MNNdtQU2sWKgyjIqVERxl9dqrmjbBbyJjpE,3796
+sglang/srt/constrained/xgrammar_backend.py,sha256=r11pWwtctbaBJGdjhQbaD_SN8n9qw902CUDh1I3ZPqo,4738
+sglang/srt/layers/activation.py,sha256=Yi2xdh7jmHUlRgERQFmStz9JwWvzT-kDmZbuf8yqy2I,5375
+sglang/srt/layers/custom_op_util.py,sha256=sE0dTU00Mkzu7RiWS0h1OvPzFey_m-StbkeR6grpY7o,827
+sglang/srt/layers/layernorm.py,sha256=1ceN6DLenmmKdxiif2uecplSUhc58qfd6s-6KWmXS9A,3943
 sglang/srt/layers/linear.py,sha256=EOdlpAf6srqxzvPpxcv10KFJKedNc22CGP1qEvpRbDg,46131
-sglang/srt/layers/logits_processor.py,sha256=1l-hJoeZUfrPPmCWcyscl0ThgKWpprUELiL1mVDfbPE,12556
+sglang/srt/layers/logits_processor.py,sha256=FFW8gVvEFxhUqDFaUPRYf3I5wA9HKsSa2IbDk7TjZZU,12575
 sglang/srt/layers/pooler.py,sha256=rj2lygvleBnyLCBZ8I11HGMgpfIDsT0l3PIkshJwdu4,1606
 sglang/srt/layers/radix_attention.py,sha256=i07VRXPDHj-zJ1TSrXEqCxumQwYSHwAvc8DoIg-Irtg,1964
 sglang/srt/layers/rotary_embedding.py,sha256=gfRKBB8FmsQKiDH0Crh_KRIGRUuvEgazH1p_n9D_m7E,3889
-sglang/srt/layers/sampler.py,sha256=3zfth1Kz24X4sUq7Z_cjZwHgPVivI-rgPtIeUbsiiWU,4589
-sglang/srt/layers/torchao_utils.py,sha256=1nzZkSzbF4qCAMeBKAeeDpMl_mK8imiY2RL3xFEgvAw,3340
+sglang/srt/layers/sampler.py,sha256=zgNwgUx7fozkWsEJFRKDV9SipHBijfpU9pTroNst6Ho,4552
+sglang/srt/layers/torchao_utils.py,sha256=v0hyr4hLsM42QwOPCdKb-ftRTjVokBZbqvRj4O4C-Nw,3415
 sglang/srt/layers/vocab_parallel_embedding.py,sha256=RmaZbgXbFnGKX1eGYxlmiko-6JwaJX6seHupUSCtAm8,21583
 sglang/srt/layers/attention/__init__.py,sha256=EL1o6Q5vLgViN3pOr2A7F6K9FlNEpMdBypFAVMeq_HA,2445
 sglang/srt/layers/attention/double_sparsity_backend.py,sha256=BlX7uXteQpnoOnKsdBKh8h20zMVMEiibB5F_PkZSlNI,10706
-sglang/srt/layers/attention/flashinfer_backend.py,sha256=843CbZsRfzWp5FTusNXXL1o4N3jd0hoCNpsoUR6Qjxk,23306
-sglang/srt/layers/attention/triton_backend.py,sha256=DKUEzxQE8iBvJPNHmQwP1pyx2wXmSsLqzBhLjJznIUk,6482
+sglang/srt/layers/attention/flashinfer_backend.py,sha256=9V5xVyx4CnT_vN8MPBOfREePgYonwzGa_PesdZClVuI,24619
+sglang/srt/layers/attention/triton_backend.py,sha256=gjxed2cvc2-8QEHkzyTVv6ui7oYOp2b_vgIUQVD1XuM,6538
 sglang/srt/layers/attention/triton_ops/decode_attention.py,sha256=Xbp2cQFYddenlReAqThN_EV7TmbSj5K3Cv5QTR5Ueqo,18787
 sglang/srt/layers/attention/triton_ops/double_sparsity_attention.py,sha256=1pSXfY3EEaM7iRN_uElHnAfsrJMhTFbu9fj8Z0O2PbE,21480
 sglang/srt/layers/attention/triton_ops/extend_attention.py,sha256=JKiDqyndNiLF8qUrG_rcdiyZvczXthO6WuSYTqd3fAo,11359
 sglang/srt/layers/attention/triton_ops/prefill_attention.py,sha256=LnuWqGAba03e25adxS_lFgjTV6nBWsVBUGUvrl-8alQ,5993
 sglang/srt/layers/fused_moe/__init__.py,sha256=bWCrDdOy2ANEXTb8CHYO63O3Iu3eZnn0PJbgl0z5vvE,75
-sglang/srt/layers/fused_moe/fused_moe.py,sha256=N15tWTm2SGuesJxDIJAdV5FsDUpE-15sb_AIgr4swlw,23656
+sglang/srt/layers/fused_moe/fused_moe.py,sha256=bxRcjdALxeY3FDnKivGOoNr6Er1kh6CCPtlAp7pjz50,23844
 sglang/srt/layers/fused_moe/layer.py,sha256=tbHnUJs3uvdDsl3VnwtyGA31VtFouNTPD7h7fPSCYOc,23613
 sglang/srt/layers/fused_moe/patch.py,sha256=K5CNLnFVxRPd8_jlY4hW6bj7pAACeCFZQA8y5loqqM4,4029
 sglang/srt/layers/quantization/__init__.py,sha256=QilMNqgu3eOFUkEjXLSDa1NvoNdi_CAvC8a1hprOgN8,2979
@@ -66,16 +69,17 @@ sglang/srt/layers/quantization/base_config.py,sha256=daK9p0aijMszLUm1W4Pc33FK87M
 sglang/srt/lora/lora.py,sha256=meRL7oBUx8mxV_isc3Lp0EIsFQWC2PvaN-fE78BmMwg,14970
 sglang/srt/lora/lora_config.py,sha256=paVB7F7SIuxr_vodvKf8zzAlH2fdVYHhXxcXV62D0Vo,1411
 sglang/srt/lora/lora_manager.py,sha256=gzBwYXZEPYj56PkGTshTbWRfl_370wb6uTcRhDaLiF8,12801
-sglang/srt/managers/data_parallel_controller.py,sha256=_XB6Ianc8TiqwLTW-7DH6gGjVYBeBU_6WjjaDk0snIY,5686
-sglang/srt/managers/detokenizer_manager.py,sha256=erRgf8RijFrGnYjZawu9an1u2mFPRY3tnxzF9PbKc80,7295
+sglang/srt/managers/data_parallel_controller.py,sha256=7Y3YOYJDe2GUyBBHJXUxDdoz24fuaO-5IGM0TwKxzFw,7895
+sglang/srt/managers/detokenizer_manager.py,sha256=ovux4AwPPTQ-JpPof7ClSTiA1sphY7IkAxPocCa1ZIs,7349
 sglang/srt/managers/image_processor.py,sha256=Pk_dtXzljTkFt7Acsv1RyDzEqvCvjc7BMngxGhtkpDU,13817
-sglang/srt/managers/io_struct.py,sha256=O_oHnikwmOexNqH4HP6bwAI5d_jG_C96JGapkLg8B7c,12289
-sglang/srt/managers/schedule_batch.py,sha256=4BgocYdKFTDCrrBkSXCT75EALBx-3RYnoN3SgtdsHlU,39595
-sglang/srt/managers/schedule_policy.py,sha256=LH0rh1PiI5LK-dSd3dar8_po6FidiBUuj0Xcp_yNQAA,12295
-sglang/srt/managers/scheduler.py,sha256=ty1sJ9U6JxifIGF4uzZX6CANMJtbjNWPe2k8aRPS6aI,48133
-sglang/srt/managers/tokenizer_manager.py,sha256=n_XCsCOwLZWCLv1ZJLGjyKgrAWCAQDyEhjnkxOptSa8,24436
-sglang/srt/managers/tp_worker.py,sha256=S5oim5xrkg1j68hYq6LfC8T533JYmQX9Kabt6U8ZXn4,5726
-sglang/srt/managers/tp_worker_overlap_thread.py,sha256=j5J4yHyR7w2HgAbN7S__299ADvsoyap5HK63SWMNavQ,7546
+sglang/srt/managers/io_struct.py,sha256=tp7RckbDklXW8YW03xXTX3Nv0DpZGjviGPx_iljoQdI,12885
+sglang/srt/managers/schedule_batch.py,sha256=kJvzb75Jmlo1iJvw1IWmLvKnBRuaUxok3MNOv-t5w18,41928
+sglang/srt/managers/schedule_policy.py,sha256=zPk5Um5-E65p0cLZ_ZwCCk7DO8dE6pWJAX9_SyfPUvw,12432
+sglang/srt/managers/scheduler.py,sha256=djbeXw7cfZBEu0uBOsQ-Wz4RCyvSWJ8ulpgaO6cSFyU,54711
+sglang/srt/managers/session_controller.py,sha256=vf2nQrxIu_14PO5xqVBhcw3WdqbdmufBOcIwnFpuyrc,2308
+sglang/srt/managers/tokenizer_manager.py,sha256=v1iCmFPhkT5IzK_LMJ-O0UPcov7pwjT49StRflBBK7Y,25882
+sglang/srt/managers/tp_worker.py,sha256=P8QQ9kAqPi7RYXkXVjFIWaZW2F5ezxQtYTJA6gJleBE,6082
+sglang/srt/managers/tp_worker_overlap_thread.py,sha256=f-zsbb6FcDrxNhLoRp2jjqSJE-tyAzZo0HAKVnx1PUY,7527
 sglang/srt/mem_cache/base_prefix_cache.py,sha256=qEQwEkG4E5rab2ZoTqcesf5pR_J4nV2jBxIHsBJHtIM,924
 sglang/srt/mem_cache/chunk_cache.py,sha256=VcCpyrf5FOQ5xoKeOouCI5ZQLkZo_pgY1SPbDDkagGg,2492
 sglang/srt/mem_cache/flush_cache.py,sha256=GYcxmNXh4hsMpFfNOuCTpKilW7guZwTtAg_usVeM3J0,979
@@ -83,18 +87,18 @@ sglang/srt/mem_cache/memory_pool.py,sha256=41fjuj_sD0yfJq-sy-X99cc2djBa6w4dy2y47
 sglang/srt/mem_cache/radix_cache.py,sha256=DzLCO_gYQ7X_C2NJSEHzzMZhb5HzWjKF9wXJQsnzr8M,10427
 sglang/srt/metrics/collector.py,sha256=9kidVhr4ldbSntAYfzwJt_2CTUFnnej0OoQdxUUwUWA,6767
 sglang/srt/metrics/func_timer.py,sha256=xe9UT4bPP1mA4GRZLsCd708cmv1B00hMpUmF7hzAKB4,3344
-sglang/srt/model_executor/cuda_graph_runner.py,sha256=ZMkyfZpWgDXfBpJ4cenh1TxXtt1O2xqeiXhDkq6E5pU,12936
-sglang/srt/model_executor/forward_batch_info.py,sha256=61TVExbiXDQRvZ6oevNz9AIxG7e-KVddgj4I6MTivLg,9426
-sglang/srt/model_executor/model_runner.py,sha256=QdFjQRnxZU8r7-MP-NdsnFnPWMRfxa-zTUmKOYmM8HE,26879
+sglang/srt/model_executor/cuda_graph_runner.py,sha256=Rm4yt4RSbFf2Dee4gI5UrbJKWgGk4quomRlVJ90TaH4,14521
+sglang/srt/model_executor/forward_batch_info.py,sha256=4PGHIQM-ZckRosIFF987xhTlotEHkt9dTMKrZQUUKqU,12397
+sglang/srt/model_executor/model_runner.py,sha256=iUKjnn0oaa2KMJgeRm4rUYrDYhg35Eg7DlBnB8OUPSw,29116
 sglang/srt/models/baichuan.py,sha256=RyvPQvi7wy9VUGvLwG17XttcTp43yRj6c3zNRImBToA,15005
 sglang/srt/models/chatglm.py,sha256=9hCXTqGX8DMvSPSn6wlK0YNNRWGS4UiS4-xjFsO9hYU,13135
 sglang/srt/models/commandr.py,sha256=leoQNn4VRqa9SXos6DcrkHVG6-Xp-kjBn2PUgqc9bs8,14051
 sglang/srt/models/dbrx.py,sha256=IiVIk_rVd0RlvfIJGIThPOPkoYT3U649PrduThiKRzg,14545
 sglang/srt/models/deepseek.py,sha256=DjW2B21isWE6A2C8A3VGZ-G0k1DkhWHO3dZZjcOVG50,15828
-sglang/srt/models/deepseek_v2.py,sha256=z6532MRN1tBltFNteFJfimnaGpyNmK6g_sdNmTzsVmk,28230
+sglang/srt/models/deepseek_v2.py,sha256=irh-2TE5PpwjsCojxpdDQCmBTuF016BTNKD673Gf4dY,32171
 sglang/srt/models/exaone.py,sha256=YMyH4zxyCaCB432vCcom800efPI19_vIQ3OXLkLiXxk,12984
 sglang/srt/models/gemma.py,sha256=D_zjG312BeOPeplGzo5Z8tSMH9xL7wZ4KIgczZ9yJ0E,12193
-sglang/srt/models/gemma2.py,sha256=iE56CYzPn-QCis4kcU7Yi0jvJ04KeU2deuZH2DaS2lM,14768
+sglang/srt/models/gemma2.py,sha256=6B999ZZBMl5twr_DMK9lnSmxwZAvVavpFHaOat71ANg,14783
 sglang/srt/models/gemma2_reward.py,sha256=zN3QYoKfMLmZlHJGVyak_kdI867rzjodYDg1SWhdW_s,2461
 sglang/srt/models/gpt2.py,sha256=Th7_Dnkw82GFBOuMOTrHtA44JBPHRUtY3Qd73rQwzMc,9741
 sglang/srt/models/gpt_bigcode.py,sha256=f6vvxBFPhV6GIZrOEKjJPu41TyVYw5Knq4h9WDvyEeY,10040
@@ -105,39 +109,40 @@ sglang/srt/models/llama.py,sha256=mIKyEHySlaCSOAAHA3x1DSnFHvlOzar7CYs2sQYZfdg,16
 sglang/srt/models/llama_classification.py,sha256=WcHYFez7qloTCpXLy1A6-dBGHWp22ebv6yG68jFVBjc,3318
 sglang/srt/models/llama_embedding.py,sha256=2ex2jrz31osaAd9V8sJeN0qyxmk-L5NgOBkXL1puGhI,3166
 sglang/srt/models/llama_reward.py,sha256=d-j00wj-_8mh2s2HJicTilNn8GWpcmxQVfmAhEJ1n7k,4524
-sglang/srt/models/llava.py,sha256=ny3sK2sgYwrEhawSAc1tZeltcgukphSTdxsqyq-Epkc,24857
-sglang/srt/models/llavavid.py,sha256=ztS5He-NF4fmfujdoMnKljOG1fNfPvp-6bduT7B6EMU,12137
+sglang/srt/models/llava.py,sha256=URAPE0xB878s_pNacA4Z2t4lAxMuzzMjLZu5gf5MseA,24847
+sglang/srt/models/llavavid.py,sha256=bqFZ0qIBlOqp-mDsBFB-QGVSemYmN6wftUKcff3r3MM,12127
 sglang/srt/models/minicpm.py,sha256=hAzgBImQ1xDeRdaQt5hKcLl1h1T-1QFSerG2MOlLjt8,13722
 sglang/srt/models/minicpm3.py,sha256=O6092exfoq8iHLmyfpVCubyQEzcfp4SmqtZJs7x4A8s,25014
 sglang/srt/models/mistral.py,sha256=tiYoKjyYVzlQl52QUZ33odD2yCxj9dxcqln474VuZOw,744
 sglang/srt/models/mixtral.py,sha256=b6AqEgL4y9wQpKKAGnhgzBtUypYo3dca5UOpGNLEt_A,13949
 sglang/srt/models/mixtral_quant.py,sha256=2ND-aOHjcyeQMUvqLLqhXwOdlR_bEftMFk3hc3lnpvc,13969
 sglang/srt/models/mllama.py,sha256=pET1x8wY04yoS8HMCncKx0tFPqGp78K8rlA7Eq7XioE,37889
-sglang/srt/models/olmo.py,sha256=eWPmo5AAnBhNGdMwklh1of3JnRzAszgQp4opeiiYidI,11887
+sglang/srt/models/olmo.py,sha256=OPEZCpFrwy47IGiwLZFYxX7UXpE5PP3KdC7UKxRhngE,11884
 sglang/srt/models/olmoe.py,sha256=fEWr-RmW6l6fVA8jM9KX8bumUWLNQQG8VxGpajlkhUs,15242
+sglang/srt/models/phi3_small.py,sha256=fxqGU0xphJzTeuBW38SRRYpRb2rcsg53JxuObK0pZig,15141
 sglang/srt/models/qwen.py,sha256=vQoq8Bv8A2zc-LE1i-E97A8i4ydtfxb2yt2JG6Tp9PQ,9851
 sglang/srt/models/qwen2.py,sha256=Y1f_PxZMTkSLgENbKl96VfNGBfvcU4cljpVe1a3vzVg,12328
 sglang/srt/models/qwen2_moe.py,sha256=RRuHLN1fIYFS4du4pUPNzGL-Rt2wLrjlgDfXiczZQ5c,16975
-sglang/srt/models/qwen2_vl.py,sha256=jb0RYMo0ShPIt4NtPCEcFGciZKstM-gYwVKND_LK7Ls,26052
+sglang/srt/models/qwen2_vl.py,sha256=G3FNa_N2-CzB56LVrukwBtJazxMrDC_GPNjK6Wqxc4s,26415
 sglang/srt/models/stablelm.py,sha256=rIQOv9OS_Vb2nOT_AMx0yGG2onwmCbbxvXL_SPdZX7k,11256
-sglang/srt/models/torch_native_llama.py,sha256=d8gVNurlVVZ-tD3Uc_aHyGCVUUp1gR8awOH4fLRZHDE,19145
+sglang/srt/models/torch_native_llama.py,sha256=RTIO2qp1SitOwNZNVzMBz8i0Gbud3t1nxTCImTguVQg,19362
 sglang/srt/models/xverse.py,sha256=meyCCdrZRYNK70hnmydgwhHa1FTBhKekEdpG0_IGTWY,13564
 sglang/srt/models/xverse_moe.py,sha256=xlrhJBAlRzxhp5o0WQU_2V5Uvf8I9fwZLOZBh95o3to,15673
 sglang/srt/models/yivl.py,sha256=xcWqkuZ29FmBBJY6aKetwItWIPl-kfXK-QmgdLONles,4765
-sglang/srt/openai_api/adapter.py,sha256=xYBmBLZ_JxfMt_m8LtVe_OB70GV4S9zBOL8e5g_VRvs,53432
-sglang/srt/openai_api/protocol.py,sha256=Mou5JUMKJkxVxoj4n8R4_sgnYY3OcwniiAi2TEM3hfY,10070
-sglang/srt/sampling/sampling_batch_info.py,sha256=7uoHypbbp4o71DfPmF22R_LeyM_Q9BTxBFg8O4lkd9w,7648
-sglang/srt/sampling/sampling_params.py,sha256=zzWVm8DxcUDdPwV1MIh5q76mmLwtkun0E08T6U3ZyWA,5192
+sglang/srt/openai_api/adapter.py,sha256=10jD3QLOAlbxTUO4-PnhgoaiNtWxbadUfb9bWyqN6gw,53540
+sglang/srt/openai_api/protocol.py,sha256=dRundxpM2kutsz-03u2nPfd3jVA0zJKmPYGAEY93t8c,10078
+sglang/srt/sampling/sampling_batch_info.py,sha256=8bQ1UvsJooPEBq_t6BXSocDAcm8OqivSUYXm4mBtnUQ,8379
+sglang/srt/sampling/sampling_params.py,sha256=u9RL8yTXYSPD6OZPvGdKvD1hmmRDY2_dg6cs2CaJhbg,5192
 sglang/srt/sampling/penaltylib/__init__.py,sha256=5vQw0Y5DSzmsoFg1IdMIKLwFVhYZ5ArADHVBYbSmOec,513
-sglang/srt/sampling/penaltylib/orchestrator.py,sha256=kizcPnxtRawmDt6utRuhbk4yfNs5H5mx1DAlDVEZRv8,11328
-sglang/srt/sampling/penaltylib/penalizers/frequency_penalty.py,sha256=IvYioX53Vq_ji-0Zhcz_r5mUa3T3GaIydVS6K4FhWfE,2557
-sglang/srt/sampling/penaltylib/penalizers/min_new_tokens.py,sha256=MmfqRqJ-leSoY9iO5Hg_ILlX-M0M0tObYrxrb_quStg,3717
-sglang/srt/sampling/penaltylib/penalizers/presence_penalty.py,sha256=0PlANTrR959foTA3Nj5qBE7ndaOZgG-9X6LhzlmEUc8,2533
-sglang/srt/sampling/penaltylib/penalizers/repetition_penalty.py,sha256=v9jOgA0-I31WcrhIydiFbpy2ZJPLytFLGM98NRPd2sU,2820
+sglang/srt/sampling/penaltylib/orchestrator.py,sha256=J-DEemZcKm1--o37kf3qDOE8SZ_6H3d5oex49Mgq2ZU,10762
+sglang/srt/sampling/penaltylib/penalizers/frequency_penalty.py,sha256=1Zp2aL6dD60mwD1tCcSG0x5IYo0v4z9ce-q_YwbJ9f8,2490
+sglang/srt/sampling/penaltylib/penalizers/min_new_tokens.py,sha256=_Nxv0XgUPirZjw2SEJYp_Cd9ZcLwmt7h6JE6J4hhFq4,3629
+sglang/srt/sampling/penaltylib/penalizers/presence_penalty.py,sha256=5tOgCg7OvE9kSN9VMCpH1hwqo1YMxt9iS5PVpct9HpU,2468
+sglang/srt/sampling/penaltylib/penalizers/repetition_penalty.py,sha256=m22Rfn1RuB1HpImBDECsiJ2VooBYpsFADAwnk1EPzk0,2751
 sglang/test/few_shot_gsm8k.py,sha256=ll-gNbcv829IwSPXAZt4JIEIu8IR3APCLcX3BHOFVp8,3968
 sglang/test/few_shot_gsm8k_engine.py,sha256=QQbrwOX6-cJDD3RZC_e7zPnt6aSo8JdF8X_lRHSjdDM,3886
 sglang/test/run_eval.py,sha256=9yO0hXZOcn4abEOs96T-XPguDEklK16Ltco0pGF3zCg,4020
-sglang/test/runners.py,sha256=JxfsGEW9L3cz87fHYmWqb3Vnbk6K1csLLLftR3LogxU,14297
+sglang/test/runners.py,sha256=31tkr6ZZ4WksLXZglAil05E1JiO71kftlg9dBiHq_u0,15034
 sglang/test/simple_eval_common.py,sha256=joqrGysuLnJFtzDRIgFkMsRyKUSyjVPFWp0_PHAL3Ik,12378
 sglang/test/simple_eval_gpqa.py,sha256=8Xt9Bw05c7SZTYrCZgB68OZUqUbLo69ywiyx0bTvSUk,3220
 sglang/test/simple_eval_humaneval.py,sha256=zmV3xWYc2OrpiT9Dy55RTKZL5DEROD1cJ0NA_-cU5zI,5685
@@ -147,10 +152,10 @@ sglang/test/simple_eval_mmlu.py,sha256=FkwamjGMjueTixymkedF-YiPloSLiy4ftILFUrKZ9
 sglang/test/test_activation.py,sha256=jkdNRzJnbd5OgZliQaIXpxovlcky17UrweomcOcMxoE,1442
 sglang/test/test_layernorm.py,sha256=IacByD5d-stXjzBz8Ypamc7povlcedpKPbb_4JLgo3c,3720
 sglang/test/test_programs.py,sha256=1Z0umrsUu9pagzyGH5SrXl_qhKSyTfUv_kWC2mcn0qo,18208
-sglang/test/test_utils.py,sha256=XvIAMeLXr4D7uLxCUSLTKP5Upc1EJd0JX2egL897Jfo,23100
-sglang/test/srt/sampling/penaltylib/utils.py,sha256=q98pQDikkmvvvvAG-AXMYaYte1iHHW2TFhKGtAeGvdE,12802
-sglang-0.3.5.post2.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-sglang-0.3.5.post2.dist-info/METADATA,sha256=ajoktPOWOAmE37TcZw562A22FmxntBUWO4zLOShVKpQ,21568
-sglang-0.3.5.post2.dist-info/WHEEL,sha256=R06PA3UVYHThwHvxuRWMqaGcr-PuniXahwjmQRFMEkY,91
-sglang-0.3.5.post2.dist-info/top_level.txt,sha256=yxhh3pYQkcnA7v3Bg889C2jZhvtJdEincysO7PEB09M,7
-sglang-0.3.5.post2.dist-info/RECORD,,
+sglang/test/test_utils.py,sha256=lBwINKlekJx03zJbnjEcO_KIkCMcBnfFa22LNt5Mwy4,23462
+sglang/test/srt/sampling/penaltylib/utils.py,sha256=CjxHgywh0hx_87iynzQt_ztHu6zBVuE-YrZ-XPmW6U4,12906
+sglang-0.3.6.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+sglang-0.3.6.dist-info/METADATA,sha256=Xqs3Fv5BkPx7ROZyCxhEBfIJzESsYz4PzjihzkA-ZZ8,21602
+sglang-0.3.6.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+sglang-0.3.6.dist-info/top_level.txt,sha256=yxhh3pYQkcnA7v3Bg889C2jZhvtJdEincysO7PEB09M,7
+sglang-0.3.6.dist-info/RECORD,,