scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-17 23:51:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 67

fix: skip token counting for multimodal inputs in process_response

Browse source
This commit is contained in:
Chesars 2026-03-21 23:29:45 -03:00
parent 9c4c34177e
commit bb247685da
2 changed files with 6 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
litellm/llms/vertex_ai/gemini_embeddings/batch_embed_content_transformation.py
View file

@ -312,8 +312,11 @@ def process_response(
model_response.data = openai_embeddings
model_response.model = model
input_text = get_formatted_prompt(data={"input": input}, call_type="embedding")
prompt_tokens = token_counter(model=model, text=input_text)
if _is_multimodal_input(input):
prompt_tokens = 0
else:
input_text = get_formatted_prompt(data={"input": input}, call_type="embedding")
prompt_tokens = token_counter(model=model, text=input_text)
model_response.usage = Usage(
prompt_tokens=prompt_tokens, total_tokens=prompt_tokens
)

2
tests/test_litellm/llms/vertex_ai/gemini_embeddings/test_batch_embed_content_transformation.py
View file

@ -192,4 +192,4 @@ class TestProcessResponse:
assert len(result.data) == 2
assert result.data[0]["index"] == 0
assert result.data[1]["index"] == 1
assert result.usage.prompt_tokens >= 0
assert result.usage.prompt_tokens == 0