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

[Model][2/N] Improve all pooling task | Support multi-vector retrieval (#25370)

Browse Source 
Signed-off-by: wang.yuqi <noooop@126.com>
This commit is contained in:
wang.yuqi
2025-10-15 19:14:41 +08:00
committed by GitHub
parent d4d1a6024f
commit f54f85129e
41 changed files with 786 additions and 399 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/conftest.py
View File

@@ -1011,8 +1011,12 @@ class VllmRunner:
req_outputs = self.llm.embed(inputs, *args, **kwargs)
return [req_output.outputs.embedding for req_output in req_outputs]
def encode(self, prompts: list[str]) -> list[list[float]]:
req_outputs = self.llm.encode(prompts)
def token_embed(self, prompts: list[str]) -> list[list[float]]:
req_outputs = self.llm.encode(prompts, pooling_task="token_embed")
return [req_output.outputs.data for req_output in req_outputs]
def token_classify(self, prompts: list[str]) -> list[list[float]]:
req_outputs = self.llm.encode(prompts, pooling_task="token_classify")
return [req_output.outputs.data for req_output in req_outputs]
def reward(self, prompts: list[str]) -> list[list[float]]: