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

5
tests/plugins_tests/test_io_processor_plugins.py
View File

@@ -93,7 +93,7 @@ def test_prithvi_mae_plugin_offline(vllm_runner, model_name: str):
out_data_format="b64_json",
)
pooling_params = PoolingParams(task="encode", softmax=False)
pooling_params = PoolingParams(activation=False)
with vllm_runner(
model_name,
@@ -108,8 +108,7 @@ def test_prithvi_mae_plugin_offline(vllm_runner, model_name: str):
io_processor_plugin="prithvi_to_tiff",
) as llm_runner:
pooler_output = llm_runner.get_llm().encode(
img_prompt,
pooling_params=pooling_params,
img_prompt, pooling_params=pooling_params, pooling_task="token_classify"
)
output = pooler_output[0].outputs