xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 20:45:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Properly propagate trust_remote_code settings (#10047)

Browse Source 
Signed-off-by: Zifei Tong <zifeitong@gmail.com>
This commit is contained in:
zifeitong 2024-11-05 16:34:40 -08:00 committed by GitHub
parent ca9844b340
commit 43300bd98a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 16 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/model_executor/models/chatglm.py
View File

@ -54,8 +54,9 @@ def mm_input_mapper_for_glmv(
data: MultiModalData[object], data: MultiModalData[object],
) -> Dict: ) -> Dict:
model_config = ctx.model_config model_config = ctx.model_config
tokenizer = cached_get_tokenizer(model_config.tokenizer, tokenizer = cached_get_tokenizer(
trust_remote_code=True) model_config.tokenizer,
trust_remote_code=model_config.trust_remote_code)
if tokenizer is None: if tokenizer is None:
raise RuntimeError("No HuggingFace processor is available " raise RuntimeError("No HuggingFace processor is available "
"to process the image object") "to process the image object")

18
vllm/model_executor/models/molmo.py
View File

@ -844,8 +844,9 @@ def get_max_tokens(max_crops: int, crop_patches: int, left_margin: int,
def get_max_molmo_image_tokens(ctx: InputContext) -> int: def get_max_molmo_image_tokens(ctx: InputContext) -> int:
processor = cached_get_processor(ctx.model_config.model, processor = cached_get_processor(
trust_remote_code=True, ctx.model_config.model,
trust_remote_code=ctx.model_config.trust_remote_code,
revision=ctx.model_config.code_revision) revision=ctx.model_config.code_revision)
image_processor = processor.image_processor image_processor = processor.image_processor
max_llm_image_tokens = get_max_tokens( max_llm_image_tokens = get_max_tokens(
@ -870,8 +871,9 @@ def image_input_mapper_for_molmo(
def dummy_data_for_molmo(ctx: InputContext, seq_len: int, def dummy_data_for_molmo(ctx: InputContext, seq_len: int,
mm_counts: Mapping[str, int]): mm_counts: Mapping[str, int]):
processor = cached_get_processor(ctx.model_config.model, processor = cached_get_processor(
trust_remote_code=True, ctx.model_config.model,
trust_remote_code=ctx.model_config.trust_remote_code,
revision=ctx.model_config.code_revision) revision=ctx.model_config.code_revision)
image_processor = processor.image_processor image_processor = processor.image_processor
@ -935,11 +937,11 @@ def input_processor_for_molmo(ctx: InputContext, inputs: DecoderOnlyInputs):
multi_modal_data = inputs.get("multi_modal_data") multi_modal_data = inputs.get("multi_modal_data")
image = None if multi_modal_data is None else multi_modal_data.get("image") image = None if multi_modal_data is None else multi_modal_data.get("image")
processor = cached_get_processor(ctx.model_config.model,
trust_remote_code=True,
revision=ctx.model_config.code_revision)
model_config = ctx.model_config model_config = ctx.model_config
processor = cached_get_processor(
ctx.model_config.model,
trust_remote_code=model_config.trust_remote_code,
revision=ctx.model_config.code_revision)
tokenizer = cached_get_tokenizer( tokenizer = cached_get_tokenizer(
model_config.tokenizer, model_config.tokenizer,
trust_remote_code=model_config.trust_remote_code) trust_remote_code=model_config.trust_remote_code)