gmakstutis/ragflow
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix: avoid mixing different embedding models in document parsing (#8260)

Browse source 
### What problem does this PR solve?

Fix mixing different embedding models in document parsing.

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

---------

Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>
This commit is contained in:
Yongteng Lei 2025-06-16 13:40:12 +08:00 • committed by GitHub
parent dabbc852c8
commit 0fa1a1469e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 5 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
api/db/services/llm_service.py
View file

@ -169,7 +169,7 @@ class TenantLLMService(CommonService):
return 0 return 0
llm_map = { llm_map = {
LLMType.EMBEDDING.value: tenant.embd_id, LLMType.EMBEDDING.value: tenant.embd_id if not llm_name else llm_name,
LLMType.SPEECH2TEXT.value: tenant.asr_id, LLMType.SPEECH2TEXT.value: tenant.asr_id,
LLMType.IMAGE2TEXT.value: tenant.img2txt_id, LLMType.IMAGE2TEXT.value: tenant.img2txt_id,
LLMType.CHAT.value: tenant.llm_id if not llm_name else llm_name, LLMType.CHAT.value: tenant.llm_id if not llm_name else llm_name,
@ -235,7 +235,8 @@ class LLMBundle:
generation = self.trace.generation(name="encode", model=self.llm_name, input={"texts": texts}) generation = self.trace.generation(name="encode", model=self.llm_name, input={"texts": texts})
embeddings, used_tokens = self.mdl.encode(texts) embeddings, used_tokens = self.mdl.encode(texts)
if not TenantLLMService.increase_usage(self.tenant_id, self.llm_type, used_tokens): llm_name = getattr(self, "llm_name", None)
if not TenantLLMService.increase_usage(self.tenant_id, self.llm_type, used_tokens, llm_name):
logging.error("LLMBundle.encode can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens)) logging.error("LLMBundle.encode can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens))
if self.langfuse: if self.langfuse:
@ -248,7 +249,8 @@ class LLMBundle:
generation = self.trace.generation(name="encode_queries", model=self.llm_name, input={"query": query}) generation = self.trace.generation(name="encode_queries", model=self.llm_name, input={"query": query})
emd, used_tokens = self.mdl.encode_queries(query) emd, used_tokens = self.mdl.encode_queries(query)
if not TenantLLMService.increase_usage(self.tenant_id, self.llm_type, used_tokens): llm_name = getattr(self, "llm_name", None)
if not TenantLLMService.increase_usage(self.tenant_id, self.llm_type, used_tokens, llm_name):
logging.error("LLMBundle.encode_queries can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens)) logging.error("LLMBundle.encode_queries can't update token usage for {}/EMBEDDING used_tokens: {}".format(self.tenant_id, used_tokens))
if self.langfuse: if self.langfuse: