biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Frontend] Chat-based Embeddings API (#9759)

Browse Source
This commit is contained in:
Cyrus Leung
2024-11-01 16:13:35 +08:00
committed by GitHub
parent d3aa2a8b2f
commit 06386a64dd
21 changed files with 846 additions and 408 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
tests/entrypoints/openai/test_tokenization.py
View File

@@ -1,4 +1,3 @@
import openai # use the official client for correctness check
import pytest
import pytest_asyncio
import requests
@@ -55,9 +54,11 @@ async def client(server):
[(MODEL_NAME, MODEL_NAME), ("zephyr-lora2", "zephyr-lora2")],
indirect=["tokenizer_name"],
)
async def test_tokenize_completions(client: openai.AsyncOpenAI,
model_name: str, tokenizer_name: str):
base_url = str(client.base_url)[:-3].strip("/")
async def test_tokenize_completions(
server: RemoteOpenAIServer,
model_name: str,
tokenizer_name: str,
):
tokenizer = get_tokenizer(tokenizer_name=tokenizer_name,
tokenizer_mode="fast")
@@ -65,7 +66,7 @@ async def test_tokenize_completions(client: openai.AsyncOpenAI,
prompt = "vllm1 This is a test prompt."
tokens = tokenizer.encode(prompt, add_special_tokens=add_special)
response = requests.post(base_url + "/tokenize",
response = requests.post(server.url_for("tokenize"),
json={
"add_special_tokens": add_special,
"model": model_name,
@@ -86,9 +87,11 @@ async def test_tokenize_completions(client: openai.AsyncOpenAI,
[(MODEL_NAME, MODEL_NAME), ("zephyr-lora2", "zephyr-lora2")],
indirect=["tokenizer_name"],
)
async def test_tokenize_chat(client: openai.AsyncOpenAI, model_name: str,
tokenizer_name: str):
base_url = str(client.base_url)[:-3].strip("/")
async def test_tokenize_chat(
server: RemoteOpenAIServer,
model_name: str,
tokenizer_name: str,
):
tokenizer = get_tokenizer(tokenizer_name=tokenizer_name,
tokenizer_mode="fast")
@@ -121,7 +124,7 @@ async def test_tokenize_chat(client: openai.AsyncOpenAI, model_name: str,
tokens = tokenizer.encode(prompt,
add_special_tokens=add_special)
response = requests.post(base_url + "/tokenize",
response = requests.post(server.url_for("tokenize"),
json={
"add_generation_prompt":
add_generation,
@@ -146,17 +149,18 @@ async def test_tokenize_chat(client: openai.AsyncOpenAI, model_name: str,
[(MODEL_NAME, MODEL_NAME), ("zephyr-lora2", "zephyr-lora2")],
indirect=["tokenizer_name"],
)
async def test_detokenize(client: openai.AsyncOpenAI, model_name: str,
tokenizer_name: str):
base_url = str(client.base_url)[:-3].strip("/")
async def test_detokenize(
server: RemoteOpenAIServer,
model_name: str,
tokenizer_name: str,
):
tokenizer = get_tokenizer(tokenizer_name=tokenizer_name,
tokenizer_mode="fast")
prompt = "This is a test prompt. vllm1"
tokens = tokenizer.encode(prompt, add_special_tokens=False)
print(f"CALLING {base_url} FOR {model_name}")
response = requests.post(base_url + "/detokenize",
response = requests.post(server.url_for("detokenize"),
json={
"model": model_name,
"tokens": tokens