zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-06-29 15:31:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: route visual agent calls to image model (#15906)

Browse Source 
### What problem does this PR solve?
Ensure agent components with image inputs route to `image2text` models
instead of staying on the chat path, so visual requests use the CV
wrapper when supported.

### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
This commit is contained in:
buua436
2026-06-10 19:09:18 +08:00
committed by GitHub
parent 0d3e410826
commit 2980981da2
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
agent/component/llm.py
View File

@@ -249,7 +249,12 @@ class LLM(ComponentBase):
self.imgs = self._uniq_images(self.imgs + extracted_imgs)
model_types = get_model_type_by_name(self._canvas.get_tenant_id(), self._param.llm_id)
model_type = LLMType.CHAT.value if LLMType.CHAT.value in model_types else model_types[0]
if self.imgs and LLMType.IMAGE2TEXT.value in model_types:
model_type = LLMType.IMAGE2TEXT.value
elif LLMType.CHAT.value in model_types:
model_type = LLMType.CHAT.value
else:
model_type = model_types[0]
model_config = get_model_config_from_provider_instance(self._canvas.get_tenant_id(), model_type, self._param.llm_id)
if self.imgs:
self.chat_mdl = LLMBundle(self._canvas.get_tenant_id(), model_config, max_retries=self._param.max_retries,