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

[V0 Deprecation] Deprecate BlockSparse Attention & Phi3-Small (#21217)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon
2025-07-19 13:53:17 -07:00
committed by GitHub
parent 881e3cbe3b
commit 752c6ade2e
38 changed files with 65 additions and 2435 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/models/registry.py
View File

@@ -247,10 +247,6 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
"PersimmonForCausalLM": _HfExamplesInfo("adept/persimmon-8b-chat"),
"PhiForCausalLM": _HfExamplesInfo("microsoft/phi-2"),
"Phi3ForCausalLM": _HfExamplesInfo("microsoft/Phi-3-mini-4k-instruct"),
# Blocksparse attention not supported in V1 yet
"Phi3SmallForCausalLM": _HfExamplesInfo("microsoft/Phi-3-small-8k-instruct",
trust_remote_code=True,
v0_only=True),
"Phi4FlashForCausalLM": _HfExamplesInfo("microsoft/Phi-4-mini-flash-reasoning", # noqa: E501
trust_remote_code=True,
v0_only=True,