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

[V1] Avoid redundant input processing in n>1 case (#14985)

Browse Source 
Signed-off-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
Nick Hill
2025-03-20 22:24:10 -07:00
committed by GitHub
parent 7297941b38
commit da6ea29f7a
13 changed files with 85 additions and 145 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/lora/test_tokenizer_group.py
View File

@@ -24,12 +24,10 @@ async def test_tokenizer_group_lora(sql_lora_files, tokenizer_group_type):
)
lora_request = LoRARequest("1", 1, sql_lora_files)
assert reference_tokenizer.encode("prompt") == tokenizer_group.encode(
request_id="request_id", prompt="prompt", lora_request=lora_request)
prompt="prompt", lora_request=lora_request)
assert reference_tokenizer.encode(
"prompt") == await tokenizer_group.encode_async(
request_id="request_id",
prompt="prompt",
lora_request=lora_request)
prompt="prompt", lora_request=lora_request)
assert isinstance(tokenizer_group.get_lora_tokenizer(None),
PreTrainedTokenizerBase)
assert tokenizer_group.get_lora_tokenizer(