fix multimodal input bug
This commit is contained in:
@@ -206,7 +206,7 @@ class OpenAI_ComputeNode(ComputeNode):
|
||||
if mode_name == "gpt-4-vision-preview":
|
||||
response_format = NOT_GIVEN
|
||||
llm_inner_functions = None
|
||||
if max_token_size > 4096:
|
||||
if max_token_size > 4096 or max_token_size < 50:
|
||||
result_token = 4096
|
||||
else:
|
||||
result_token = -1
|
||||
|
||||
Reference in New Issue
Block a user