sglang 0.2.7__py3-none-any.whl → 0.2.9__py3-none-any.whl

sglang/test/simple_eval_mmlu.py

@@ -0,0 +1,120 @@
+# Adapted from https://github.com/openai/simple-evals/
+
+"""
+Measuring Massive Multitask Language Understanding
+Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, Jacob Steinhardt
+https://arxiv.org/abs/2009.03300
+"""
+
+import random
+import re
+
+import pandas
+
+from sglang.test import simple_eval_common as common
+from sglang.test.simple_eval_common import (
+    ANSWER_PATTERN_MULTICHOICE,
+    HTML_JINJA,
+    Eval,
+    EvalResult,
+    SamplerBase,
+    SingleEvalResult,
+    format_multichoice_question,
+)
+
+subject2category = {
+    "abstract_algebra": "stem",
+    "anatomy": "other",
+    "astronomy": "stem",
+    "business_ethics": "other",
+    "clinical_knowledge": "other",
+    "college_biology": "stem",
+    "college_chemistry": "stem",
+    "college_computer_science": "stem",
+    "college_mathematics": "stem",
+    "college_medicine": "other",
+    "college_physics": "stem",
+    "computer_security": "stem",
+    "conceptual_physics": "stem",
+    "econometrics": "social_sciences",
+    "electrical_engineering": "stem",
+    "elementary_mathematics": "stem",
+    "formal_logic": "humanities",
+    "global_facts": "other",
+    "high_school_biology": "stem",
+    "high_school_chemistry": "stem",
+    "high_school_computer_science": "stem",
+    "high_school_european_history": "humanities",
+    "high_school_geography": "social_sciences",
+    "high_school_government_and_politics": "social_sciences",
+    "high_school_macroeconomics": "social_sciences",
+    "high_school_mathematics": "stem",
+    "high_school_microeconomics": "social_sciences",
+    "high_school_physics": "stem",
+    "high_school_psychology": "social_sciences",
+    "high_school_statistics": "stem",
+    "high_school_us_history": "humanities",
+    "high_school_world_history": "humanities",
+    "human_aging": "other",
+    "human_sexuality": "social_sciences",
+    "international_law": "humanities",
+    "jurisprudence": "humanities",
+    "logical_fallacies": "humanities",
+    "machine_learning": "stem",
+    "management": "other",
+    "marketing": "other",
+    "medical_genetics": "other",
+    "miscellaneous": "other",
+    "moral_disputes": "humanities",
+    "moral_scenarios": "humanities",
+    "nutrition": "other",
+    "philosophy": "humanities",
+    "prehistory": "humanities",
+    "professional_accounting": "other",
+    "professional_law": "humanities",
+    "professional_medicine": "other",
+    "professional_psychology": "social_sciences",
+    "public_relations": "social_sciences",
+    "security_studies": "social_sciences",
+    "sociology": "social_sciences",
+    "us_foreign_policy": "social_sciences",
+    "virology": "other",
+    "world_religions": "humanities",
+}
+
+
+class MMLUEval(Eval):
+    def __init__(self, filename: str, num_examples: int | None, num_threads: int):
+        df = pandas.read_csv(filename)
+        examples = [row.to_dict() for _, row in df.iterrows()]
+        if num_examples:
+            examples = random.Random(0).sample(examples, num_examples)
+        self.examples = examples
+        self.num_threads = num_threads
+
+    def __call__(self, sampler: SamplerBase) -> EvalResult:
+        def fn(row: dict):
+            prompt_messages = [
+                sampler._pack_message(
+                    content=format_multichoice_question(row), role="user"
+                )
+            ]
+            response_text = sampler(prompt_messages)
+            match = re.search(ANSWER_PATTERN_MULTICHOICE, response_text)
+            extracted_answer = match.group(1) if match else None
+            score = 1.0 if extracted_answer == row["Answer"] else 0.0
+            html = common.jinja_env.from_string(HTML_JINJA).render(
+                prompt_messages=prompt_messages,
+                next_message=dict(content=response_text, role="assistant"),
+                score=score,
+                correct_answer=row["Answer"],
+                extracted_answer=extracted_answer,
+            )
+            convo = prompt_messages + [dict(content=response_text, role="assistant")]
+            category = subject2category.get(row["Subject"], "other")
+            return SingleEvalResult(
+                html=html, score=score, metrics={category: score}, convo=convo
+            )
+
+        results = common.map_with_progress(fn, self.examples, self.num_threads)
+        return common.aggregate_results(results)

sglang/test/test_programs.py

@@ -105,23 +105,21 @@ def test_decode_json_regex():
     def decode_json(s):
         from sglang.lang.ir import REGEX_FLOAT, REGEX_INT, REGEX_STRING
 
-        s += "Generate a JSON object to describe the basic information of a city.\n"
+        s += "Generate a JSON object to describe the basic city information of Paris.\n"
 
         with s.var_scope("json_output"):
             s += "{\n"
             s += '  "name": ' + sgl.gen(regex=REGEX_STRING + ",") + "\n"
             s += '  "population": ' + sgl.gen(regex=REGEX_INT + ",") + "\n"
             s += '  "area": ' + sgl.gen(regex=REGEX_INT + ",") + "\n"
-            s += '  "latitude": ' + sgl.gen(regex=REGEX_FLOAT + ",") + "\n"
-            s += '  "country": ' + sgl.gen(regex=REGEX_STRING + ",") + "\n"
-            s += '  "timezone": ' + sgl.gen(regex=REGEX_STRING) + "\n"
+            s += '  "latitude": ' + sgl.gen(regex=REGEX_FLOAT) + "\n"
             s += "}"
 
-    ret = decode_json.run()
+    ret = decode_json.run(temperature=0.0)
     try:
         js_obj = json.loads(ret["json_output"])
     except json.decoder.JSONDecodeError:
-        print(ret["json_output"])
+        print("JSONDecodeError", ret["json_output"])
         raise
     assert isinstance(js_obj["name"], str)
     assert isinstance(js_obj["population"], int)
@@ -130,7 +128,7 @@ def test_decode_json_regex():
 def test_decode_json():
     @sgl.function
     def decode_json(s):
-        s += "Generate a JSON object to describe the basic information of a city.\n"
+        s += "Generate a JSON object to describe the basic city information of Paris.\n"
 
         with s.var_scope("json_output"):
             s += "{\n"
@@ -141,8 +139,12 @@ def test_decode_json():
             s += '  "timezone": ' + sgl.gen(dtype=str) + "\n"
             s += "}"
 
-    ret = decode_json.run()
-    js_obj = json.loads(ret["json_output"])
+    ret = decode_json.run(max_new_tokens=64)
+    try:
+        js_obj = json.loads(ret["json_output"])
+    except json.decoder.JSONDecodeError:
+        print("JSONDecodeError", ret["json_output"])
+        raise
     assert isinstance(js_obj["name"], str)
     assert isinstance(js_obj["population"], int)
 
@@ -261,6 +263,7 @@ def test_parallel_decoding():
         s += "\nIn summary," + sgl.gen("summary", max_tokens=512)
 
     ret = parallel_decoding.run(topic="writing a good blog post", temperature=0.3)
+    assert isinstance(ret["summary"], str)
 
 
 def test_parallel_encoding(check_answer=True):

sglang/test/test_utils.py

@@ -1,6 +1,8 @@
 """Common utilities for testing and benchmarking"""
 
 import asyncio
+import subprocess
+import time
 from functools import partial
 
 import numpy as np
@@ -11,6 +13,8 @@ from sglang.lang.backend.openai import OpenAI
 from sglang.lang.backend.runtime_endpoint import RuntimeEndpoint
 from sglang.utils import get_exception_traceback
 
+MODEL_NAME_FOR_TEST = "meta-llama/Meta-Llama-3.1-8B-Instruct"
+
 
 def call_generate_lightllm(prompt, temperature, max_tokens, stop=None, url=None):
     assert url is not None
@@ -379,3 +383,31 @@ def get_call_select(args):
             raise
 
     return func
+
+
+def popen_launch_server(model, port, timeout, *args):
+    command = [
+        "python3",
+        "-m",
+        "sglang.launch_server",
+        "--model-path",
+        model,
+        "--host",
+        "localhost",
+        "--port",
+        str(port),
+        *args,
+    ]
+    process = subprocess.Popen(command, stdout=None, stderr=None)
+    base_url = f"http://localhost:{port}/v1"
+
+    start_time = time.time()
+    while time.time() - start_time < timeout:
+        try:
+            response = requests.get(f"{base_url}/models")
+            if response.status_code == 200:
+                return process
+        except requests.RequestException:
+            pass
+        time.sleep(10)
+    raise TimeoutError("Server failed to start within the timeout period.")

sglang/version.py

@@ -1 +1 @@
-__version__ = "0.2.7"
+__version__ = "0.2.9"

{sglang-0.2.7.dist-info → sglang-0.2.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sglang
-Version: 0.2.7
+Version: 0.2.9
 Summary: SGLang is yet another fast serving framework for large language models and vision language models.
 License: Apache License
                                    Version 2.0, January 2004
@@ -299,8 +299,8 @@ pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.3/
 
 ### Method 2: From source
 ```
-# Use the stable release branch
-git clone -b release https://github.com/sgl-project/sglang.git
+# Use the stable v0.2.9 branch
+git clone -b v0.2.9 https://github.com/sgl-project/sglang.git
 cd sglang
 
 pip install --upgrade pip
@@ -312,7 +312,7 @@ pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.3/
 
 ### Method 3: Using docker
 The docker images are available on Docker Hub as [lmsysorg/sglang](https://hub.docker.com/r/lmsysorg/sglang/tags), built from [Dockerfile](docker).
-Repalce `<secret>` below with your huggingface hub [token](https://huggingface.co/docs/hub/en/security-tokens).
+Replace `<secret>` below with your huggingface hub [token](https://huggingface.co/docs/hub/en/security-tokens).
 
 ```bash
 docker run --gpus all \

{sglang-0.2.7.dist-info → sglang-0.2.9.dist-info}/RECORD

@@ -1,18 +1,18 @@
 sglang/__init__.py,sha256=ECjvAWlxIwKtUIXGchfkoCIbF-iqLjH-Q0o8xHTlVNY,1352
 sglang/api.py,sha256=s_P8BvGDCQ0PiqOapr2TLFge1NA7QmKqUx6bFQ8Q5GQ,5676
 sglang/bench_latency.py,sha256=JPatRvstM3nXb-ViVgtR-TaRrFHpcHzqoDG7BQmRYK8,10539
-sglang/bench_serving.py,sha256=6DK6Ps8y6-Eb9QlbGBRlhPRTseDqVIRoDreO5GDHZ64,34846
+sglang/bench_serving.py,sha256=M0YQT6xElpkx-FtmyUe6lhX1DZfVLGh54qd6qfFYquc,34801
 sglang/check_env.py,sha256=Eeb_20VetnlEFYSRcHFlNqt85lYUQN60NEtkoX7ahPA,4121
 sglang/global_config.py,sha256=CyhGL7PE-KlMcg7IHWykzImU1y4NQlpeIlh9lHA77uo,1749
 sglang/launch_server.py,sha256=Gg8CwNlTCCfg1dF65ZT9ePLxOT9LKtY79GhIPG6PCrU,358
 sglang/launch_server_llavavid.py,sha256=40uaazMsavKuk6YXFa5v37kdUpFGuealgJJeph1g8gU,1025
 sglang/utils.py,sha256=r0Z7hY_bFFk-b6WeQJir9br-hCW2-p7n5E7Et2WziaQ,8776
-sglang/version.py,sha256=XHypfHSPdgXFKmOdoewn7czU670gt8InhHhzlP5j_aA,22
+sglang/version.py,sha256=F8OVhAhMXSkvvXYgZtbPn2SG1AQC3joK4yu-FrHt81Y,22
 sglang/lang/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sglang/lang/chat_template.py,sha256=psIlhaDo70twgLrx5Lgln03metLEA3-FZuixeI0Y7Ao,13309
 sglang/lang/compiler.py,sha256=UiXUmPR9wBAPtnORrLcyQX8Uh0ZL0nKeV8ZgBozAJPw,7531
-sglang/lang/interpreter.py,sha256=dt_NAAMv2oSYxwSMjhMr2pIGTe5_d12cSR91SUWvpCQ,30298
-sglang/lang/ir.py,sha256=THa6hwnuTVXVYxnovNQP_o7A9v5O8uXE4eLXH9vDRLA,16648
+sglang/lang/interpreter.py,sha256=_MbvYB0vweCgALklpM2DlofiCXuITCmX_fl8rPPcp5U,30340
+sglang/lang/ir.py,sha256=0r-mhA4aO-uuS97Dvkw99ERTcJXfzuV6jJQMmuCwHEg,16615
 sglang/lang/tracer.py,sha256=borJmlSJOhg1RUndGRnilnR60eEZz2Y9aU7BpftsOxU,8287
 sglang/lang/backend/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sglang/lang/backend/anthropic.py,sha256=EXRX7xJgA5KZszX7toSLVnKzFQ5EO0Loj-YjHFtxSxg,2081
@@ -26,18 +26,18 @@ sglang/srt/hf_transformers_utils.py,sha256=Fg-3panb6lsqOhHmAYA0ivkXyBjdnvY5mqvil
 sglang/srt/mm_utils.py,sha256=n7_GmbOM_0IWVXovpM34rKIBw0Py9yb_NXSQw27u4OA,9454
 sglang/srt/model_config.py,sha256=DO7m84WiT3dzPWmyKz_UXDAHEdqEjq8Lq5wCjzjYMME,6023
 sglang/srt/sampling_params.py,sha256=uZFDlTUPnNR5_3IDH-INDeN-tm6LlRkC2KT-B3njxJs,3687
-sglang/srt/server.py,sha256=2qgluP7_6-e36PDK_mr-rLK9us3_9KvXLG3255h-tS4,16022
-sglang/srt/server_args.py,sha256=0cV-r5QTV_9Arl3hVf9mc20BlOhYhWSkICU0T3dS180,15412
-sglang/srt/utils.py,sha256=uIatocIFzqi6fWSscz2MjF3jUcIRBJlqLgYeicM_W9s,22950
+sglang/srt/server.py,sha256=cDHUmLqj7MjF-3L9WcfA-4z9dRl55cwF5ygXuncMl-Q,15852
+sglang/srt/server_args.py,sha256=wdRlxR-509RfNYuMQoxUAefMwoc5eme6sYwEMyRBHmk,16034
+sglang/srt/utils.py,sha256=5wgGe6kI59JAmf8kxLsItulJ4xQaOJHHYaWWd6_WWmo,23384
 sglang/srt/constrained/__init__.py,sha256=NLpZGj9RIx83ejDrM_pfaRtqGgaPq_ggJszPQENUJ2E,2037
-sglang/srt/constrained/base_cache.py,sha256=Aeu2HMPhXMPNQNEwPJ19sECN0PYPZKjisrZiCcocHiw,1970
-sglang/srt/constrained/fsm_cache.py,sha256=Q7wfGx7VOghErqcC_0kK4aI-lBEO9TxoFPyUiBxEGVE,2626
-sglang/srt/constrained/jump_forward.py,sha256=SYKj5Pd3d7oym5fAI8zUzj3zKk-lV30m_ksAy0ubgO8,6180
+sglang/srt/constrained/base_tool_cache.py,sha256=1_m-AivPtWRwUgGiEZBafCrSFUGahK4UM4vgAd8TkMg,2004
+sglang/srt/constrained/fsm_cache.py,sha256=GoPBr_9ZdJizF2PKbYoQw2I4ckfrUYwCeMZxB9sY3TM,2639
+sglang/srt/constrained/jump_forward.py,sha256=IgZ8D0woy5FLIQvXkE8wZRYejDsfVkjU0sqUlkiv_f4,6193
 sglang/srt/layers/context_flashattention_nopad.py,sha256=r_TpHuYAVgq1pN81PiWe1bebtY-p9MBndBaoIE2VXrk,5180
 sglang/srt/layers/extend_attention.py,sha256=zuNnAdL_wF6BX0Mwn1dgDJvh3YJjYwqa5Fbzp8muOVc,12573
 sglang/srt/layers/fused_moe.py,sha256=KmyXwau2OOZpQimGIQrHptzGNs1trIud5AKEEKXdzPU,20823
 sglang/srt/layers/linear.py,sha256=3Se2FRXyqXcd-uvNx2b7s-jolsUTEVeYBMYHmV82wPw,34518
-sglang/srt/layers/logits_processor.py,sha256=JE0NYlQniy9wmPeIKs3QbYbpaXqAoNtVdEPkV_qt59I,11076
+sglang/srt/layers/logits_processor.py,sha256=5Cg3h5b4H0EUeOJRst3IOMWL5dniP63A5s15BRkAMmk,11091
 sglang/srt/layers/radix_attention.py,sha256=tdA-kdd9LQY1wbw3iYuy-9cikVJYmy3EctwAlUfN-Uo,6945
 sglang/srt/layers/token_attention.py,sha256=ylUqUnozJCCohxTGAiiP3sxgUrcXfEVic8-qgcHYDj4,7968
 sglang/srt/layers/quantization/__init__.py,sha256=JMlgE-FWS759lfQ9Uc6mGFqBbTFLlvKeVEFpZLATe14,2536
@@ -47,14 +47,16 @@ sglang/srt/managers/controller_single.py,sha256=CdQ9_XPZdcWF5jArDmVR8K-WZ9_8Gpgk
 sglang/srt/managers/detokenizer_manager.py,sha256=GXWdW4n2N-otL3zcgdr0t1PcEe2EmQJA8AElntiNV1o,5606
 sglang/srt/managers/io_struct.py,sha256=Rz7Ur9Yw6prDGdy6XjsSiUmVBccS6cef-G_9TW7HA_4,7105
 sglang/srt/managers/policy_scheduler.py,sha256=ajSB-gCC6VJkXvnKU8FYU3Kgcigozp2pMTwF84Wp14o,3138
-sglang/srt/managers/schedule_batch.py,sha256=tbos5i4KSfk1K8VH5HCNm2pQGlJMKVAE_mZ8haVMelc,36620
-sglang/srt/managers/tokenizer_manager.py,sha256=tEct3shjjw_7ickj_cmt9IxoBHfgbryQHI7DZS0m4TA,20511
-sglang/srt/managers/tp_worker.py,sha256=91gbWi7hSuyTC3Qvo7EXKmHM6GKWTK0Nqpda001jOw0,34349
+sglang/srt/managers/schedule_batch.py,sha256=LIoVCPNivh0u1dOrrWRgFD6a4ywq3nrG_4dNgCK0kIw,37697
+sglang/srt/managers/tokenizer_manager.py,sha256=rtZ44aiZOMHLHkXDhMgj0HDR3gExpeGjWfoCD0PfG_o,20574
+sglang/srt/managers/tp_worker.py,sha256=JPLneFwcPlmPXZX1QxZHWgcdau8FC8wNuVqfCqsgOkU,35234
+sglang/srt/mem_cache/base_cache.py,sha256=czyN8IumXcMQskYOZDV3DzjfD4kdR-qwLVxceDqnOmE,788
+sglang/srt/mem_cache/chunk_cache.py,sha256=u1mkGoTI7_31H0i0mhKT7S57StYSsdmsSPqyGubE7lY,1560
 sglang/srt/mem_cache/flush_cache.py,sha256=pTLKPRB17U6vl5RFJJvuJ4jCL2SyomgkUBNlkDpGRqo,978
 sglang/srt/mem_cache/memory_pool.py,sha256=wkhjyYLbAZrl2FB5i4ODkxgMufBuDpe4N0kbXhu6ZO0,4509
-sglang/srt/mem_cache/radix_cache.py,sha256=Xk0c8nwyPHEUsobVJQrr7edwyzUMk9MBYTQBprN8a0Y,8775
+sglang/srt/mem_cache/radix_cache.py,sha256=pa5RD4xNKPSuvL55BnC4mimoca5oJRXr4Rg91-sbTcs,8881
 sglang/srt/model_executor/cuda_graph_runner.py,sha256=OdmO6R7nHWrRJCtZOxYkt0KNdGoX7Md4knsypwPYjaQ,9365
-sglang/srt/model_executor/model_runner.py,sha256=WyPsO73MD3ziKAk76j4HemePYZluXjs9WGYeajUgfQA,15507
+sglang/srt/model_executor/model_runner.py,sha256=fo3fbnNaHkcHz2UDkyvFjU7sGvdClhmhdelQh0n9PgA,16079
 sglang/srt/model_loader/model_loader.py,sha256=QmZUhHh1nmWrfYlunfnxMcTsIvip1l6aMIlrXoCED4I,10697
 sglang/srt/model_loader/utils.py,sha256=0AoWXX9uV5rKRYXJ4HduSnvdeerytI4ONCLCH6X4XFQ,10675
 sglang/srt/models/chatglm.py,sha256=vYWooqyPmcSFZNjxj_g5I_FgHJlDytbEiz6vyv3JBNM,13856
@@ -67,7 +69,7 @@ sglang/srt/models/gemma2.py,sha256=kTjZcsptgtYaO8BL_NlygjVSMSloq2Mc4Rf3FKvEhbs,1
 sglang/srt/models/gpt_bigcode.py,sha256=U7GmHKywSu12D-EwvuWv3RwHkx6bPawaRIjlFIpQkfs,10194
 sglang/srt/models/grok.py,sha256=NfZdsRVErDIUWFqjhtNf2pqC9G4cRdYHBFpgDq1IZ2A,27855
 sglang/srt/models/internlm2.py,sha256=Ld2GUxZeqqqJ2vd4QiX2s1y2AceJLA1nVnUYY88GMQk,12219
-sglang/srt/models/llama2.py,sha256=zhoCUh_3dNC7FOzDnaoHcHF3-y7vTVYDZzHKqIsUJgs,14764
+sglang/srt/models/llama2.py,sha256=zfOk3OK1_B6s6yuXsZFmNCf07RsfytVD72GunLBt8Cc,14282
 sglang/srt/models/llama_classification.py,sha256=4r_orFZqBR3U_yC4bus1K3Z3-ADscYGSzgA82_VDN0g,4926
 sglang/srt/models/llava.py,sha256=BJphgyQGdo7uTpJcKGEfWwdpH9GTMDnyiznLSSgmvm8,18476
 sglang/srt/models/llavavid.py,sha256=-7vaVqaIfukCvMkNakEPblpwjIHC6ezrAvmpE5RzlUY,13602
@@ -80,14 +82,16 @@ sglang/srt/models/qwen2.py,sha256=mXlVd6UTCXY3VdgodFpQnlaY-NYLIbA-SknxdA9R13w,12
 sglang/srt/models/qwen2_moe.py,sha256=YYdJEezic7GyW-_bXlNIaqBa0C4IHQpz_vuRBLxms4k,18141
 sglang/srt/models/stablelm.py,sha256=b3d-ZwLQoLjZ6CupnkIq7d-z9tzGSxAyIcgSmZiZxZw,11362
 sglang/srt/models/yivl.py,sha256=p4s_D_m4H2exP4b91Y-CTkq8T-eIG3DJsFy9pB0e7TM,4932
-sglang/srt/openai_api/adapter.py,sha256=Jn8Awi93zkb3Wq5gqK698kOhmqYdtxZlRePciA50Ud4,30213
-sglang/srt/openai_api/protocol.py,sha256=_mBNdxb_4ZRIeP0wmW8tMTc2x7zu4foVxBDCuCWkaiw,7822
-sglang/test/test_conversation.py,sha256=gF_AyOxQgpPQBPnA57-kq-M0p_zFu-rBDMFgAq655Rw,1596
-sglang/test/test_openai_protocol.py,sha256=DVx3r6hrb8oRqbo5AYIleldxbqMBTtb-gtORM6t_Y1c,1661
-sglang/test/test_programs.py,sha256=s4WGpTmYP4Yx5g8JYZpbkeF9RN5iUnlKdi8FGAZovTc,13756
-sglang/test/test_utils.py,sha256=kD_fQe3WroZ9Kc3NBRKPiZOFJ_JD2uEE9XIvPp6AD9Y,11048
-sglang-0.2.7.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-sglang-0.2.7.dist-info/METADATA,sha256=NU4S55-t6q87AKPkgbDORvX_Om0XbAJ9K67_p30JnQ0,33216
-sglang-0.2.7.dist-info/WHEEL,sha256=R0nc6qTxuoLk7ShA2_Y-UWkN8ZdfDBG2B6Eqpz2WXbs,91
-sglang-0.2.7.dist-info/top_level.txt,sha256=yxhh3pYQkcnA7v3Bg889C2jZhvtJdEincysO7PEB09M,7
-sglang-0.2.7.dist-info/RECORD,,
+sglang/srt/openai_api/adapter.py,sha256=h6TIU0Fu3jU361pye4J12vcDug7UJJRPiBAY_HfFUuE,32599
+sglang/srt/openai_api/protocol.py,sha256=JXLnnQ63I-bJv93ICPfP0cBpyomQA5IYE_mkUg5X4Es,8177
+sglang/test/run_eval.py,sha256=WvMLSi70G9fhruP8cPLOfDJ9XEKL7yNn2pylx-7tNsQ,3054
+sglang/test/simple_eval_common.py,sha256=Qh1-iEXJCKfJmgpAzNSp28fcP1TUJzt3s9i1FjvemHY,12340
+sglang/test/simple_eval_humaneval.py,sha256=IW0ZC6D4SXu06IJiMoAY9DK9SMsTOlDPAwu4cfbJco0,5826
+sglang/test/simple_eval_mmlu.py,sha256=KqSSdSu2qfoKQ870ttxev1NJ7c90xv2mvKOQsSODtAw,4326
+sglang/test/test_programs.py,sha256=e9_ifoIvuI1Ctkbkz3wfdZLBBSRikby8ywcodBIkf9M,13826
+sglang/test/test_utils.py,sha256=PndOL1zdseMrpHTHGmgsHHepxqYBn__eNLrlsSXLy6k,11905
+sglang-0.2.9.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+sglang-0.2.9.dist-info/METADATA,sha256=8vhH67MeR6EdJepUSvmqKSneJTQ8l_9LD9L6FfzyrHk,33214
+sglang-0.2.9.dist-info/WHEEL,sha256=R0nc6qTxuoLk7ShA2_Y-UWkN8ZdfDBG2B6Eqpz2WXbs,91
+sglang-0.2.9.dist-info/top_level.txt,sha256=yxhh3pYQkcnA7v3Bg889C2jZhvtJdEincysO7PEB09M,7
+sglang-0.2.9.dist-info/RECORD,,

sglang/test/test_conversation.py

@@ -1,46 +0,0 @@
-from sglang.srt.conversation import generate_chat_conv
-from sglang.srt.managers.openai_api.protocol import (
-    ChatCompletionMessageContentImagePart,
-    ChatCompletionMessageContentImageURL,
-    ChatCompletionMessageContentTextPart,
-    ChatCompletionMessageGenericParam,
-    ChatCompletionMessageUserParam,
-    ChatCompletionRequest,
-)
-
-
-def test_chat_completion_to_conv_image():
-    """Test that we can convert a chat image request to a convo"""
-    request = ChatCompletionRequest(
-        model="default",
-        messages=[
-            ChatCompletionMessageGenericParam(
-                role="system", content="You are a helpful AI assistant"
-            ),
-            ChatCompletionMessageUserParam(
-                role="user",
-                content=[
-                    ChatCompletionMessageContentTextPart(
-                        type="text", text="Describe this image"
-                    ),
-                    ChatCompletionMessageContentImagePart(
-                        type="image_url",
-                        image_url=ChatCompletionMessageContentImageURL(
-                            url="https://someurl.com"
-                        ),
-                    ),
-                ],
-            ),
-        ],
-    )
-    conv = generate_chat_conv(request, "vicuna_v1.1")
-    assert conv.messages == [
-        ["USER", "Describe this image<image>"],
-        ["ASSISTANT", None],
-    ]
-    assert conv.system_message == "You are a helpful AI assistant"
-    assert conv.image_data == ["https://someurl.com"]
-
-
-if __name__ == "__main__":
-    test_chat_completion_to_conv_image()

sglang/test/test_openai_protocol.py

@@ -1,51 +0,0 @@
-from sglang.srt.managers.openai_api.protocol import (
-    ChatCompletionMessageContentImagePart,
-    ChatCompletionMessageContentImageURL,
-    ChatCompletionMessageContentTextPart,
-    ChatCompletionMessageGenericParam,
-    ChatCompletionMessageUserParam,
-    ChatCompletionRequest,
-)
-
-
-def test_chat_completion_request_image():
-    """Test that Chat Completion Requests with images can be converted."""
-
-    image_request = {
-        "model": "default",
-        "messages": [
-            {"role": "system", "content": "You are a helpful AI assistant"},
-            {
-                "role": "user",
-                "content": [
-                    {"type": "text", "text": "Describe this image"},
-                    {"type": "image_url", "image_url": {"url": "https://someurl.com"}},
-                ],
-            },
-        ],
-        "temperature": 0,
-        "max_tokens": 64,
-    }
-    request = ChatCompletionRequest(**image_request)
-    assert len(request.messages) == 2
-    assert request.messages[0] == ChatCompletionMessageGenericParam(
-        role="system", content="You are a helpful AI assistant"
-    )
-    assert request.messages[1] == ChatCompletionMessageUserParam(
-        role="user",
-        content=[
-            ChatCompletionMessageContentTextPart(
-                type="text", text="Describe this image"
-            ),
-            ChatCompletionMessageContentImagePart(
-                type="image_url",
-                image_url=ChatCompletionMessageContentImageURL(
-                    url="https://someurl.com"
-                ),
-            ),
-        ],
-    )
-
-
-if __name__ == "__main__":
-    test_chat_completion_request_image()