AI/ragflow
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/infiniflow/ragflow.git synced 2025-07-13 03:41:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix: huggingface rerank model issue. (#6385)

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

#6348

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)
This commit is contained in:
Kevin Hu 2025-03-21 12:43:32 +08:00 committed by GitHub
parent bc58ecbfd7
commit d83911b632
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
rag/llm/rerank_model.py
View File

@ -537,7 +537,7 @@ class HuggingfaceRerank(DefaultRerank):
return np.array(scores) return np.array(scores)
def __init__(self, key, model_name="BAAI/bge-reranker-v2-m3", base_url="http://127.0.0.1"): def __init__(self, key, model_name="BAAI/bge-reranker-v2-m3", base_url="http://127.0.0.1"):
self.model_name = model_name self.model_name = model_name.split("___")[0]
self.base_url = base_url self.base_url = base_url
def similarity(self, query: str, texts: list) -> tuple[np.ndarray, int]: def similarity(self, query: str, texts: list) -> tuple[np.ndarray, int]:

4
rag/svr/task_executor.py
View File

@ -57,7 +57,7 @@ from rag.app import laws, paper, presentation, manual, qa, table, book, resume,
from rag.nlp import search, rag_tokenizer from rag.nlp import search, rag_tokenizer
from rag.raptor import RecursiveAbstractiveProcessing4TreeOrganizedRetrieval as Raptor from rag.raptor import RecursiveAbstractiveProcessing4TreeOrganizedRetrieval as Raptor
from rag.settings import DOC_MAXIMUM_SIZE, SVR_CONSUMER_GROUP_NAME, get_svr_queue_name, get_svr_queue_names, print_rag_settings, TAG_FLD, PAGERANK_FLD from rag.settings import DOC_MAXIMUM_SIZE, SVR_CONSUMER_GROUP_NAME, get_svr_queue_name, get_svr_queue_names, print_rag_settings, TAG_FLD, PAGERANK_FLD
from rag.utils import num_tokens_from_string from rag.utils import num_tokens_from_string, truncate
from rag.utils.redis_conn import REDIS_CONN from rag.utils.redis_conn import REDIS_CONN
from rag.utils.storage_factory import STORAGE_IMPL from rag.utils.storage_factory import STORAGE_IMPL
from graphrag.utils import chat_limiter from graphrag.utils import chat_limiter
@ -404,7 +404,7 @@ async def embedding(docs, mdl, parser_config=None, callback=None):
cnts_ = np.array([]) cnts_ = np.array([])
for i in range(0, len(cnts), batch_size): for i in range(0, len(cnts), batch_size):
vts, c = await trio.to_thread.run_sync(lambda: mdl.encode(cnts[i: i + batch_size])) vts, c = await trio.to_thread.run_sync(lambda: mdl.encode([truncate(c, mdl.max_length-10) for c in cnts[i: i + batch_size]]))
if len(cnts_) == 0: if len(cnts_) == 0:
cnts_ = vts cnts_ = vts
else: else: