phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-04 04:04:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chore: update the vLLM inference impl to use OpenAIMixin for openai-compat functions

Browse source 
inference recordings from Qwen3-0.6B and vLLM 0.8.3 -
```
docker run --gpus all -v ~/.cache/huggingface:/root/.cache/huggingface -p 8000:8000 --ipc=host \
    vllm/vllm-openai:latest \
    --model Qwen/Qwen3-0.6B --enable-auto-tool-choice --tool-call-parser hermes
```

test with -

```
./scripts/integration-tests.sh --stack-config server:ci-tests --setup vllm --subdirs inference
```
This commit is contained in:
Matthew Farrellee 2025-09-10 10:10:10 -04:00
parent c86e45496e
commit c2a9c65fff
33 changed files with 51813 additions and 203 deletions
Download patch file Download diff file Expand all files Collapse all files

197
llama_stack/providers/remote/inference/vllm/vllm.py
View file

@ -4,7 +4,7 @@
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import json
from collections.abc import AsyncGenerator, AsyncIterator
from collections.abc import AsyncGenerator
from typing import Any
import httpx
@ -38,13 +38,6 @@ from llama_stack.apis.inference import (
LogProbConfig,
Message,
ModelStore,
OpenAIChatCompletion,
OpenAICompletion,
OpenAIEmbeddingData,
OpenAIEmbeddingsResponse,
OpenAIEmbeddingUsage,
OpenAIMessageParam,
OpenAIResponseFormatParam,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -71,11 +64,11 @@ from llama_stack.providers.utils.inference.openai_compat import (
convert_message_to_openai_dict,
convert_tool_call,
get_sampling_options,
prepare_openai_completion_params,
process_chat_completion_stream_response,
process_completion_response,
process_completion_stream_response,
)
from llama_stack.providers.utils.inference.openai_mixin import OpenAIMixin
from llama_stack.providers.utils.inference.prompt_adapter import (
completion_request_to_prompt,
content_has_media,
@ -288,7 +281,7 @@ async def _process_vllm_chat_completion_stream_response(
yield c
class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
class VLLMInferenceAdapter(OpenAIMixin, Inference, ModelsProtocolPrivate):
# automatically set by the resolver when instantiating the provider
__provider_id__: str
model_store: ModelStore | None = None
@ -296,7 +289,6 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
def __init__(self, config: VLLMInferenceAdapterConfig) -> None:
self.register_helper = ModelRegistryHelper(build_hf_repo_model_entries())
self.config = config
self.client = None
async def initialize(self) -> None:
if not self.config.url:
@ -308,8 +300,6 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
return self.config.refresh_models
async def list_models(self) -> list[Model] | None:
self._lazy_initialize_client()
assert self.client is not None # mypy
models = []
async for m in self.client.models.list():
model_type = ModelType.llm # unclear how to determine embedding vs. llm models
@ -340,8 +330,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
HealthResponse: A dictionary containing the health status.
"""
try:
client = self._create_client() if self.client is None else self.client
_ = [m async for m in client.models.list()] # Ensure the client is initialized
_ = [m async for m in self.client.models.list()] # Ensure the client is initialized
return HealthResponse(status=HealthStatus.OK)
except Exception as e:
return HealthResponse(status=HealthStatus.ERROR, message=f"Health check failed: {str(e)}")
@ -351,19 +340,14 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
raise ValueError("Model store not set")
return await self.model_store.get_model(model_id)
def _lazy_initialize_client(self):
if self.client is not None:
return
def get_api_key(self):
return self.config.api_token
log.info(f"Initializing vLLM client with base_url={self.config.url}")
self.client = self._create_client()
def get_base_url(self):
return self.config.url
def _create_client(self):
return AsyncOpenAI(
base_url=self.config.url,
api_key=self.config.api_token,
http_client=httpx.AsyncClient(verify=self.config.tls_verify),
)
def get_extra_client_params(self):
return {"http_client": httpx.AsyncClient(verify=self.config.tls_verify)}
async def completion(
self,
@ -374,7 +358,6 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
stream: bool | None = False,
logprobs: LogProbConfig | None = None,
) -> CompletionResponse | AsyncGenerator[CompletionResponseStreamChunk, None]:
self._lazy_initialize_client()
if sampling_params is None:
sampling_params = SamplingParams()
model = await self._get_model(model_id)
@ -406,7 +389,6 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
logprobs: LogProbConfig | None = None,
tool_config: ToolConfig | None = None,
) -> ChatCompletionResponse | AsyncGenerator[ChatCompletionResponseStreamChunk, None]:
self._lazy_initialize_client()
if sampling_params is None:
sampling_params = SamplingParams()
model = await self._get_model(model_id)
@ -479,16 +461,12 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
yield chunk
async def register_model(self, model: Model) -> Model:
# register_model is called during Llama Stack initialization, hence we cannot init self.client if not initialized yet.
# self.client should only be created after the initialization is complete to avoid asyncio cross-context errors.
# Changing this may lead to unpredictable behavior.
client = self._create_client() if self.client is None else self.client
try:
model = await self.register_helper.register_model(model)
except ValueError:
pass # Ignore statically unknown model, will check live listing
try:
res = await client.models.list()
res = await self.client.models.list()
except APIConnectionError as e:
raise ValueError(
f"Failed to connect to vLLM at {self.config.url}. Please check if vLLM is running and accessible at that URL."
@ -543,8 +521,6 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
output_dimension: int | None = None,
task_type: EmbeddingTaskType | None = None,
) -> EmbeddingsResponse:
self._lazy_initialize_client()
assert self.client is not None
model = await self._get_model(model_id)
kwargs = {}
@ -560,154 +536,3 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
embeddings = [data.embedding for data in response.data]
return EmbeddingsResponse(embeddings=embeddings)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
self._lazy_initialize_client()
assert self.client is not None
model_obj = await self._get_model(model)
assert model_obj.model_type == ModelType.embedding
# Convert input to list if it's a string
input_list = [input] if isinstance(input, str) else input
# Call vLLM embeddings endpoint with encoding_format
response = await self.client.embeddings.create(
model=model_obj.provider_resource_id,
input=input_list,
dimensions=dimensions,
encoding_format=encoding_format,
)
# Convert response to OpenAI format
data = [
OpenAIEmbeddingData(
embedding=embedding_data.embedding,
index=i,
)
for i, embedding_data in enumerate(response.data)
]
# Not returning actual token usage since vLLM doesn't provide it
usage = OpenAIEmbeddingUsage(prompt_tokens=-1, total_tokens=-1)
return OpenAIEmbeddingsResponse(
data=data,
model=model_obj.provider_resource_id,
usage=usage,
)
async def openai_completion(
self,
model: str,
prompt: str | list[str] | list[int] | list[list[int]],
best_of: int | None = None,
echo: bool | None = None,
frequency_penalty: float | None = None,
logit_bias: dict[str, float] | None = None,
logprobs: bool | None = None,
max_tokens: int | None = None,
n: int | None = None,
presence_penalty: float | None = None,
seed: int | None = None,
stop: str | list[str] | None = None,
stream: bool | None = None,
stream_options: dict[str, Any] | None = None,
temperature: float | None = None,
top_p: float | None = None,
user: str | None = None,
guided_choice: list[str] | None = None,
prompt_logprobs: int | None = None,
suffix: str | None = None,
) -> OpenAICompletion:
self._lazy_initialize_client()
model_obj = await self._get_model(model)
extra_body: dict[str, Any] = {}
if prompt_logprobs is not None and prompt_logprobs >= 0:
extra_body["prompt_logprobs"] = prompt_logprobs
if guided_choice:
extra_body["guided_choice"] = guided_choice
params = await prepare_openai_completion_params(
model=model_obj.provider_resource_id,
prompt=prompt,
best_of=best_of,
echo=echo,
frequency_penalty=frequency_penalty,
logit_bias=logit_bias,
logprobs=logprobs,
max_tokens=max_tokens,
n=n,
presence_penalty=presence_penalty,
seed=seed,
stop=stop,
stream=stream,
stream_options=stream_options,
temperature=temperature,
top_p=top_p,
user=user,
extra_body=extra_body,
)
return await self.client.completions.create(**params) # type: ignore
async def openai_chat_completion(
self,
model: str,
messages: list[OpenAIMessageParam],
frequency_penalty: float | None = None,
function_call: str | dict[str, Any] | None = None,
functions: list[dict[str, Any]] | None = None,
logit_bias: dict[str, float] | None = None,
logprobs: bool | None = None,
max_completion_tokens: int | None = None,
max_tokens: int | None = None,
n: int | None = None,
parallel_tool_calls: bool | None = None,
presence_penalty: float | None = None,
response_format: OpenAIResponseFormatParam | None = None,
seed: int | None = None,
stop: str | list[str] | None = None,
stream: bool | None = None,
stream_options: dict[str, Any] | None = None,
temperature: float | None = None,
tool_choice: str | dict[str, Any] | None = None,
tools: list[dict[str, Any]] | None = None,
top_logprobs: int | None = None,
top_p: float | None = None,
user: str | None = None,
) -> OpenAIChatCompletion | AsyncIterator[OpenAIChatCompletionChunk]:
self._lazy_initialize_client()
model_obj = await self._get_model(model)
params = await prepare_openai_completion_params(
model=model_obj.provider_resource_id,
messages=messages,
frequency_penalty=frequency_penalty,
function_call=function_call,
functions=functions,
logit_bias=logit_bias,
logprobs=logprobs,
max_completion_tokens=max_completion_tokens,
max_tokens=max_tokens,
n=n,
parallel_tool_calls=parallel_tool_calls,
presence_penalty=presence_penalty,
response_format=response_format,
seed=seed,
stop=stop,
stream=stream,
stream_options=stream_options,
temperature=temperature,
tool_choice=tool_choice,
tools=tools,
top_logprobs=top_logprobs,
top_p=top_p,
user=user,
)
return await self.client.chat.completions.create(**params) # type: ignore

28
llama_stack/providers/utils/inference/openai_mixin.py
View file

@ -67,6 +67,17 @@ class OpenAIMixin(ABC):
"""
pass
def get_extra_client_params(self) -> dict[str, Any]:
"""
Get any extra parameters to pass to the AsyncOpenAI client.
Child classes can override this method to provide additional parameters
such as timeout settings, proxies, etc.
:return: A dictionary of extra parameters
"""
return {}
@property
def client(self) -> AsyncOpenAI:
"""
@ -78,6 +89,7 @@ class OpenAIMixin(ABC):
return AsyncOpenAI(
api_key=self.get_api_key(),
base_url=self.get_base_url(),
**self.get_extra_client_params(),
)
async def _get_provider_model_id(self, model: str) -> str:
@ -124,10 +136,15 @@ class OpenAIMixin(ABC):
"""
Direct OpenAI completion API call.
"""
if guided_choice is not None:
logger.warning("guided_choice is not supported by the OpenAI API. Ignoring.")
if prompt_logprobs is not None:
logger.warning("prompt_logprobs is not supported by the OpenAI API. Ignoring.")
# Handle parameters that are not supported by OpenAI API, but may be by the provider
# prompt_logprobs is supported by vLLM
# guided_choice is supported by vLLM
# TODO: test coverage
extra_body: dict[str, Any] = {}
if prompt_logprobs is not None and prompt_logprobs >= 0:
extra_body["prompt_logprobs"] = prompt_logprobs
if guided_choice:
extra_body["guided_choice"] = guided_choice
# TODO: fix openai_completion to return type compatible with OpenAI's API response
return await self.client.completions.create( # type: ignore[no-any-return]
@ -150,7 +167,8 @@ class OpenAIMixin(ABC):
top_p=top_p,
user=user,
suffix=suffix,
)
),
extra_body=extra_body,
)
async def openai_chat_completion(

3917
tests/integration/recordings/responses/004d45cd1d80.json Normal file
View file

File diff suppressed because it is too large Load diff

66
tests/integration/recordings/responses/032f8c5a1289.json Normal file
View file

@ -0,0 +1,66 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
"content": [
{
"type": "text",
"text": "Which planet has rings around it with a name starting with letter S?"
}
]
}
],
"stream": false,
"temperature": 0.0,
"max_tokens": 4096
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-307d8869f9a341439fef1e92274a89ec",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, so the question is asking which planet has rings around it and the name starts with the letter S. Let me think. I know that there are many planets in our solar system, right? The ones we usually mention are Mercury, Venus, Earth, Mars, Jupiter, Saturn, etc. Now, the key here is that the planet has rings, and the name starts with S.\n\nFirst, I remember that Saturn is the planet with rings. But wait, the name starts with S, so that's correct. But let me make sure. Are there any other planets with rings that start with S? Like maybe something else? Let me think. Oh, right! There's also Neptune, but Neptune is a gas giant, not a planet with rings. Wait, no, Neptune does have rings, but they're not as prominent as Saturn's. So Saturn is the one with rings and the name starting with S. \n\nWait, but maybe I should double-check. Let me recall. The planets in order are Mercury, Venus, Earth, Mars, Jupiter, Saturn, and then Neptune. So Saturn comes after Jupiter. So yes, Saturn has rings. And the name starts with S. So the answer should be Saturn.\n</think>\n\nThe planet with rings around it and a name starting with the letter S is **Saturn**.",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524152,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 274,
"prompt_tokens": 22,
"total_tokens": 296,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

105
tests/integration/recordings/responses/033340f58ebc.json Normal file
View file

@ -0,0 +1,105 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"tools": [
{
"type": "function",
"function": {
"name": "get_weather",
"description": "Get the current weather",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "The city and state (both required), e.g. San Francisco, CA."
}
},
"required": [
"location"
]
}
}
}
],
"messages": [
{
"role": "system",
"content": [
{
"type": "text",
"text": "Pretend you are a weather assistant."
}
]
},
{
"role": "user",
"content": [
{
"type": "text",
"text": "What's the weather like in San Francisco, CA?"
}
]
}
],
"stream": false,
"temperature": 0.0,
"max_tokens": 4096
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-3d55352696e445f598c3d881a1130454",
"choices": [
{
"finish_reason": "tool_calls",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, the user is asking about the weather in San Francisco, CA. I need to use the get_weather function. The function requires the location parameter, which in this case is \"San Francisco, CA\". I should make sure to format the arguments correctly as a JSON object. Let me check the required parameters again to confirm. The location is required, so I'll include that. No other parameters are needed. Alright, I'll structure the tool call accordingly.\n</think>\n\n",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [
{
"id": "chatcmpl-tool-98de38728cf6486d925102e0515d6d8f",
"function": {
"arguments": "{\"location\": \"San Francisco, CA\"}",
"name": "get_weather"
},
"type": "function"
}
],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524137,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 120,
"prompt_tokens": 185,
"total_tokens": 305,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

1021
tests/integration/recordings/responses/07eabb983d8e.json Normal file
View file

File diff suppressed because it is too large Load diff

2888
tests/integration/recordings/responses/0dfab8a5deb8.json Normal file
View file

File diff suppressed because it is too large Load diff

59
tests/integration/recordings/responses/27463384d1a3.json Normal file
View file

@ -0,0 +1,59 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
"content": "Hello, world!"
}
],
"stream": false
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-5434d48fcb354eb498e3909a87c4bf39",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, the user just said \"Hello, world!\" so I need to respond in a friendly way. Let me start by acknowledging their input. Maybe say \"Hello, world!\" and then offer help. Keep it simple and positive. No need for any complex messages. Just a straightforward reply.\n</think>\n\nHello, world! \ud83d\ude0a How can I assist you today?",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524166,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 77,
"prompt_tokens": 12,
"total_tokens": 89,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

9609
tests/integration/recordings/responses/299952f00fd8.json Normal file
View file

File diff suppressed because it is too large Load diff

3750
tests/integration/recordings/responses/42a07c178a75.json Normal file
View file

File diff suppressed because it is too large Load diff

129
tests/integration/recordings/responses/50053de3a8a8.json Normal file
View file

@ -0,0 +1,129 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"tools": [
{
"type": "function",
"function": {
"name": "get_object_namespace_list",
"description": "Get the list of objects in a namespace",
"parameters": {
"type": "object",
"properties": {
"kind": {
"type": "string",
"description": "the type of object"
},
"namespace": {
"type": "string",
"description": "the name of the namespace"
}
},
"required": [
"kind",
"namespace"
]
}
}
}
],
"messages": [
{
"role": "system",
"content": [
{
"type": "text",
"text": "You are a helpful assistant."
}
]
},
{
"role": "user",
"content": [
{
"type": "text",
"text": "What pods are in the namespace openshift-lightspeed?"
}
]
},
{
"role": "assistant",
"content": [
{
"type": "text",
"text": ""
}
],
"tool_calls": [
{
"id": "1",
"type": "function",
"function": {
"name": "get_object_namespace_list",
"arguments": "{\"kind\": \"pod\", \"namespace\": \"openshift-lightspeed\"}"
}
}
]
},
{
"role": "tool",
"content": [
{
"type": "text",
"text": "the objects are pod1, pod2, pod3"
}
]
}
],
"stream": false,
"temperature": 0.0,
"max_tokens": 4096
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-042380355dac4046b0510320198e5563",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, the user asked for the pods in the namespace openshift-lightspeed. I called the get_object_namespace_list function with kind \"pod\" and namespace \"openshift-lightspeed\". The response came back with three objects: pod1, pod2, pod3. Now I need to present this information clearly.\n\nI should confirm the answer by listing each pod. Maybe mention that the list includes those specific pods. Keep it straightforward and make sure the user knows the result. No need for extra details since the tool response already provided the list. Just relay the information back in a friendly manner.\n</think>\n\nThe pods in the namespace openshift-lightspeed are: pod1, pod2, and pod3.",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524154,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 148,
"prompt_tokens": 256,
"total_tokens": 404,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

70
tests/integration/recordings/responses/551066a5c7f4.json Normal file
View file

@ -0,0 +1,70 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "Hello, world!",
"stream": false,
"extra_body": {
"prompt_logprobs": 0
}
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-f7da1cba36ba427c8acd9ad2f0722fa4",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": null,
"text": " I'm a 3-year-old girl who loves to play with my cat.",
"stop_reason": null,
"prompt_logprobs": [
null,
{
"11": {
"logprob": -8.8668212890625,
"rank": 1031,
"decoded_token": ","
}
},
{
"1879": {
"logprob": -4.7821879386901855,
"rank": 10,
"decoded_token": "\u0120world"
}
},
{
"0": {
"logprob": -1.6575366258621216,
"rank": 2,
"decoded_token": "!"
}
}
]
}
],
"created": 1757524144,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 16,
"prompt_tokens": 4,
"total_tokens": 20,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

1021
tests/integration/recordings/responses/592d9733fb28.json Normal file
View file

File diff suppressed because it is too large Load diff

124
tests/integration/recordings/responses/62b67ba9b8ea.json Normal file
View file

@ -0,0 +1,124 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "system",
"content": [
{
"type": "text",
"text": "You are a helpful assistant. Michael Jordan was born in 1963. He played basketball for the Chicago Bulls for 15 seasons."
}
]
},
{
"role": "user",
"content": [
{
"type": "text",
"text": "Please give me information about Michael Jordan."
}
]
}
],
"extra_body": {
"guided_json": {
"$defs": {
"NBAStats": {
"properties": {
"year_for_draft": {
"title": "Year For Draft",
"type": "integer"
},
"num_seasons_in_nba": {
"title": "Num Seasons In Nba",
"type": "integer"
}
},
"required": [
"year_for_draft",
"num_seasons_in_nba"
],
"title": "NBAStats",
"type": "object"
}
},
"properties": {
"first_name": {
"title": "First Name",
"type": "string"
},
"last_name": {
"title": "Last Name",
"type": "string"
},
"year_of_birth": {
"title": "Year Of Birth",
"type": "integer"
},
"nba_stats": {
"$ref": "#/$defs/NBAStats"
}
},
"required": [
"first_name",
"last_name",
"year_of_birth",
"nba_stats"
],
"title": "AnswerFormat",
"type": "object"
}
},
"stream": false,
"temperature": 0.0,
"max_tokens": 4096
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-ba863ae481964bcca497a7feba2a2689",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "{\"first_name\": \"Michael\", \"last_name\": \"Jordan\", \"year_of_birth\": 1963, \"nba_stats\": {\"year_for_draft\": 1984, \"num_seasons_in_nba\": 15}}",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524142,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 61,
"prompt_tokens": 51,
"total_tokens": 112,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

46
tests/integration/recordings/responses/62b82c42a270.json Normal file
View file

@ -0,0 +1,46 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "<|begin_of_text|>Complete the sentence using one word: Roses are red, violets are ",
"stream": false,
"temperature": 0.0,
"max_tokens": 50
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-75aab3472ee24335962c490f4efdb884",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": null,
"text": " _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _",
"stop_reason": null,
"prompt_logprobs": null
}
],
"created": 1757524131,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 50,
"prompt_tokens": 23,
"total_tokens": 73,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

8685
tests/integration/recordings/responses/666d8719f984.json Normal file
View file

File diff suppressed because it is too large Load diff

2684
tests/integration/recordings/responses/67c8e604fcd3.json Normal file
View file

File diff suppressed because it is too large Load diff

50
tests/integration/recordings/responses/6825a67ceccd.json Normal file
View file

@ -0,0 +1,50 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "I am feeling really sad today.",
"stream": false,
"extra_body": {
"guided_choice": [
"joy",
"sadness"
]
}
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-b4feb568934e44e7a8b7ac8071dce285",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"text": "sadness",
"stop_reason": null,
"prompt_logprobs": null
}
],
"created": 1757524114,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 3,
"prompt_tokens": 7,
"total_tokens": 10,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

86
tests/integration/recordings/responses/6cf57d8dfcfc.json Normal file
View file

@ -0,0 +1,86 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "<|begin_of_text|>Complete the sentence: Micheael Jordan is born in ",
"logprobs": 1,
"stream": false,
"temperature": 0.0,
"max_tokens": 5
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-b114bec84b5a430ba94520ac9e7a6d8d",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": {
"text_offset": [
0,
1,
2,
3,
4
],
"token_logprobs": [
-0.03187580779194832,
-0.014971747994422913,
-0.3393988609313965,
-1.2200194597244263,
-0.762129545211792
],
"tokens": [
"1",
"9",
"5",
"8",
","
],
"top_logprobs": [
{
"1": -0.03187580779194832
},
{
"9": -0.014971747994422913
},
{
"5": -0.3393988609313965
},
{
"8": -1.2200194597244263
},
{
",": -0.762129545211792
}
]
},
"text": "1958,",
"stop_reason": null,
"prompt_logprobs": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 5,
"prompt_tokens": 18,
"total_tokens": 23,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

59
tests/integration/recordings/responses/7ef63231b9f8.json Normal file
View file

@ -0,0 +1,59 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
"content": "Which planet has rings around it with a name starting with letter S?"
}
],
"stream": false
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-ab6b31bfc8244a48bee3f9ed8bff6455",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, so the question is asking which planet has rings around it and the name starts with the letter S. Let me think. I know that there are many planets in our solar system, like Earth, Mars, Venus, Jupiter, etc. But the question specifically mentions rings and the name starting with S. \n\nFirst, I should recall the planets and their names. Let me list them in order of their distance from the Sun. From the Sun, we have Mercury, Venus, Earth, Mars, Jupiter, Saturn, Uranus, Neptune, and then Pluto. Wait, Pluto was once considered a planet but was reclassified. So maybe the question is about the planets that are now in order with the rings. \n\nNow, Saturn is the next one after Jupiter. Saturn is the planet that has rings around it. The name starts with S. So the answer would be Saturn. Let me double-check. Yes, Saturn is the planet with rings and the name starts with S. I don't think there's another planet with rings and a starting letter S. For example, maybe another planet in another system, but the question is about the solar system. So I think the answer is Saturn.\n</think>\n\nThe planet with rings around it and a name starting with the letter S is **Saturn**.",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524161,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 266,
"prompt_tokens": 22,
"total_tokens": 288,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

4055
tests/integration/recordings/responses/89b141855b81.json Normal file
View file

File diff suppressed because it is too large Load diff

524
tests/integration/recordings/responses/8ea4d7e002fa.json Normal file
View file

@ -0,0 +1,524 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "<|begin_of_text|>Return the exact same sentence and don't add additional words): Michael Jordan was born in the year of 1963",
"stream": true,
"temperature": 0.0,
"max_tokens": 50,
"stop": [
"1963"
]
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": [
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": ".",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": " <|",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "end",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "_of_t",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "e",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "xt|",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": ">\n",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "\n",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "*",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "*A",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": ".** Mich",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "ael Jor",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "dan ",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "was b",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "orn",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": " in ",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "the y",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "ear",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": " ",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "o",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": "f",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": null,
"text": " ",
"stop_reason": null
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-8ca2b3d09fbe49f99b86bc496a6c4bf8",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"text": "",
"stop_reason": "1963"
}
],
"created": 1757524133,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
}
],
"is_streaming": true
}
}

71
tests/integration/recordings/responses/934eabb2c7bd.json Normal file
View file

@ -0,0 +1,71 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "<|begin_of_text|>Michael Jordan was born in 1963. He played basketball for the Chicago Bulls. He retired in 2003.Please respond in JSON format with the schema: {\"properties\": {\"name\": {\"title\": \"Name\", \"type\": \"string\"}, \"year_born\": {\"title\": \"Year Born\", \"type\": \"string\"}, \"year_retired\": {\"title\": \"Year Retired\", \"type\": \"string\"}}, \"required\": [\"name\", \"year_born\", \"year_retired\"], \"title\": \"AnswerFormat\", \"type\": \"object\"}",
"extra_body": {
"guided_json": {
"properties": {
"name": {
"title": "Name",
"type": "string"
},
"year_born": {
"title": "Year Born",
"type": "string"
},
"year_retired": {
"title": "Year Retired",
"type": "string"
}
},
"required": [
"name",
"year_born",
"year_retired"
],
"title": "AnswerFormat",
"type": "object"
}
},
"stream": false,
"temperature": 0.0,
"max_tokens": 50
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-d0abbd0dc22c41a091a2a72ce913cbed",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"text": "{ \n \"name\": \"Michael Jordan\",\n \"year_born\": \"1963\",\n \"year_retired\": \"2003\"\n}",
"stop_reason": null,
"prompt_logprobs": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 36,
"prompt_tokens": 128,
"total_tokens": 164,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

197
tests/integration/recordings/responses/a1f9695ef183.json Normal file
View file

@ -0,0 +1,197 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "<|begin_of_text|>Complete the sentence: Micheael Jordan is born in ",
"logprobs": 1,
"stream": true,
"temperature": 0.0,
"max_tokens": 5
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": [
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-7968eb721ac34b348eb879f300052b1b",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": {
"text_offset": [
0
],
"token_logprobs": [
-0.031588248908519745
],
"tokens": [
"1"
],
"top_logprobs": [
{
"1": -0.031588248908519745
}
]
},
"text": "1",
"stop_reason": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-7968eb721ac34b348eb879f300052b1b",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": {
"text_offset": [
1
],
"token_logprobs": [
-0.01468715537339449
],
"tokens": [
"9"
],
"top_logprobs": [
{
"9": -0.01468715537339449
}
]
},
"text": "9",
"stop_reason": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-7968eb721ac34b348eb879f300052b1b",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": {
"text_offset": [
2
],
"token_logprobs": [
-0.30242931842803955
],
"tokens": [
"5"
],
"top_logprobs": [
{
"5": -0.30242931842803955
}
]
},
"text": "5",
"stop_reason": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-7968eb721ac34b348eb879f300052b1b",
"choices": [
{
"finish_reason": null,
"index": 0,
"logprobs": {
"text_offset": [
3
],
"token_logprobs": [
-1.2276304960250854
],
"tokens": [
"8"
],
"top_logprobs": [
{
"8": -1.2276304960250854
}
]
},
"text": "8",
"stop_reason": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
},
{
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-7968eb721ac34b348eb879f300052b1b",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": {
"text_offset": [
4
],
"token_logprobs": [
-0.7615041732788086
],
"tokens": [
","
],
"top_logprobs": [
{
",": -0.7615041732788086
}
]
},
"text": ",",
"stop_reason": null
}
],
"created": 1757524134,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": null
}
}
],
"is_streaming": true
}
}

5747
tests/integration/recordings/responses/a5163d8d2d01.json Normal file
View file

File diff suppressed because it is too large Load diff

3509
tests/integration/recordings/responses/aacf9abc51d4.json Normal file
View file

File diff suppressed because it is too large Load diff

66
tests/integration/recordings/responses/c42baf6a3700.json Normal file
View file

@ -0,0 +1,66 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
"content": [
{
"type": "text",
"text": "Which planet do humans live on?"
}
]
}
],
"stream": false,
"temperature": 0.0,
"max_tokens": 4096
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-f7cda312aefe4a4ca4c34e7342f23802",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, the user is asking which planet humans live on. Let me think. I know that humans are on Earth. But maybe they're confused because sometimes people think of other planets. I should clarify that Earth is the only planet where humans have been found. Also, maybe mention that Earth is a unique planet with all the necessary conditions for life. I should keep it simple and direct.\n</think>\n\nHumans live on **Earth**. This planet is unique in having the conditions necessary for life, such as liquid water, breathable air, and suitable temperatures.",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524135,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 114,
"prompt_tokens": 15,
"total_tokens": 129,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

59
tests/integration/recordings/responses/e08e01e5652a.json Normal file
View file

@ -0,0 +1,59 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/chat/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
"content": "Which planet do humans live on?"
}
],
"stream": false
},
"endpoint": "/v1/chat/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.chat.chat_completion.ChatCompletion",
"__data__": {
"id": "chatcmpl-8e11e56bbb9744e1ac92ae6af3fb5148",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"logprobs": null,
"message": {
"content": "<think>\nOkay, the user is asking which planet humans live on. Hmm, I know that humans are on Earth, but I should make sure. Wait, Earth is our home. But maybe they're confused because sometimes people think of other planets. Let me check. The answer is Earth. But maybe I should explain why it's correct. Oh, right, Earth is the planet where we live. No other planets have been confirmed to have humans. So the answer is Earth.\n</think>\n\nHumans live on **Earth**.",
"refusal": null,
"role": "assistant",
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning_content": null
},
"stop_reason": null
}
],
"created": 1757524156,
"model": "Qwen/Qwen3-0.6B",
"object": "chat.completion",
"service_tier": null,
"system_fingerprint": null,
"usage": {
"completion_tokens": 107,
"prompt_tokens": 15,
"total_tokens": 122,
"completion_tokens_details": null,
"prompt_tokens_details": null
},
"prompt_logprobs": null
}
},
"is_streaming": false
}
}

3041
tests/integration/recordings/responses/f518ea4fde7d.json Normal file
View file

File diff suppressed because it is too large Load diff

45
tests/integration/recordings/responses/fadf16af6d35.json Normal file
View file

@ -0,0 +1,45 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "Respond to this question and explain your answer. Complete the sentence using one word: Roses are red, violets are ",
"stream": false,
"extra_body": {}
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-52758d074c7541f99998bf277e553e78",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": null,
"text": " ____. \n\nThe sentence is: \"Roses are red, violets",
"stop_reason": null,
"prompt_logprobs": null
}
],
"created": 1757524112,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 16,
"prompt_tokens": 25,
"total_tokens": 41,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

85
tests/integration/recordings/responses/fded9afd616c.json Normal file
View file

@ -0,0 +1,85 @@
{
"request": {
"method": "POST",
"url": "http://localhost:8000/v1/v1/completions",
"headers": {},
"body": {
"model": "Qwen/Qwen3-0.6B",
"prompt": "Hello, world!",
"stream": false,
"extra_body": {
"prompt_logprobs": 1
}
},
"endpoint": "/v1/completions",
"model": "Qwen/Qwen3-0.6B"
},
"response": {
"body": {
"__type__": "openai.types.completion.Completion",
"__data__": {
"id": "cmpl-df4fc1ac600a43778b6ad1e3781a4129",
"choices": [
{
"finish_reason": "length",
"index": 0,
"logprobs": null,
"text": " I'm a 14-year-old girl with a passion for science and tech",
"stop_reason": null,
"prompt_logprobs": [
null,
{
"11": {
"logprob": -8.8668212890625,
"rank": 1031,
"decoded_token": ","
},
"21806": {
"logprob": -4.8355712890625,
"rank": 1,
"decoded_token": "\u0120Answer"
}
},
{
"1879": {
"logprob": -4.7821879386901855,
"rank": 10,
"decoded_token": "\u0120world"
},
"358": {
"logprob": -0.9071879386901855,
"rank": 1,
"decoded_token": "\u0120I"
}
},
{
"0": {
"logprob": -1.6575366258621216,
"rank": 2,
"decoded_token": "!"
},
"13": {
"logprob": -1.4700366258621216,
"rank": 1,
"decoded_token": "."
}
}
]
}
],
"created": 1757524113,
"model": "Qwen/Qwen3-0.6B",
"object": "text_completion",
"system_fingerprint": null,
"usage": {
"completion_tokens": 16,
"prompt_tokens": 4,
"total_tokens": 20,
"completion_tokens_details": null,
"prompt_tokens_details": null
}
}
},
"is_streaming": false
}
}

2
tests/integration/suites.py
View file

@ -78,7 +78,7 @@ SETUP_DEFINITIONS: dict[str, Setup] = {
"VLLM_URL": "http://localhost:8000/v1",
},
defaults={
"text_model": "vllm/meta-llama/Llama-3.2-1B-Instruct",
"text_model": "vllm/Qwen/Qwen3-0.6B",
"embedding_model": "sentence-transformers/all-MiniLM-L6-v2",
},
),

21
tests/unit/providers/inference/test_remote_vllm.py
View file

@ -11,7 +11,7 @@ import threading
import time
from http.server import BaseHTTPRequestHandler, HTTPServer
from typing import Any
from unittest.mock import AsyncMock, MagicMock, patch
from unittest.mock import AsyncMock, MagicMock, PropertyMock, patch
import pytest
from openai.types.chat.chat_completion_chunk import (
@ -150,10 +150,12 @@ async def test_tool_call_response(vllm_inference_adapter):
"""Verify that tool call arguments from a CompletionMessage are correctly converted
into the expected JSON format."""
# Patch the call to vllm so we can inspect the arguments sent were correct
with patch.object(
vllm_inference_adapter.client.chat.completions, "create", new_callable=AsyncMock
) as mock_nonstream_completion:
# Patch the client property to avoid instantiating a real AsyncOpenAI client
with patch.object(VLLMInferenceAdapter, "client", new_callable=PropertyMock) as mock_create_client:
mock_client = MagicMock()
mock_client.chat.completions.create = AsyncMock()
mock_create_client.return_value = mock_client
messages = [
SystemMessage(content="You are a helpful assistant"),
UserMessage(content="How many?"),
@ -179,7 +181,7 @@ async def test_tool_call_response(vllm_inference_adapter):
tool_config=ToolConfig(tool_choice=ToolChoice.auto),
)
assert mock_nonstream_completion.call_args.kwargs["messages"][2]["tool_calls"] == [
assert mock_client.chat.completions.create.call_args.kwargs["messages"][2]["tool_calls"] == [
{
"id": "foo",
"type": "function",
@ -641,9 +643,7 @@ async def test_health_status_success(vllm_inference_adapter):
This test verifies that the health method returns a HealthResponse with status OK, only
when the connection to the vLLM server is successful.
"""
# Set vllm_inference_adapter.client to None to ensure _create_client is called
vllm_inference_adapter.client = None
with patch.object(vllm_inference_adapter, "_create_client") as mock_create_client:
with patch.object(VLLMInferenceAdapter, "client", new_callable=PropertyMock) as mock_create_client:
# Create mock client and models
mock_client = MagicMock()
mock_models = MagicMock()
@ -674,8 +674,7 @@ async def test_health_status_failure(vllm_inference_adapter):
This test verifies that the health method returns a HealthResponse with status ERROR
and an appropriate error message when the connection to the vLLM server fails.
"""
vllm_inference_adapter.client = None
with patch.object(vllm_inference_adapter, "_create_client") as mock_create_client:
with patch.object(VLLMInferenceAdapter, "client", new_callable=PropertyMock) as mock_create_client:
# Create mock client and models
mock_client = MagicMock()
mock_models = MagicMock()