[Bug] 关于使用embedding模型batch_size超过上限的建议 #669
Description
Search before asking
- I had searched in the issues and found no similar issues.
Operating system information
Linux
What happened
我使用的是阿里的text-embedding-v3嵌入模型，api限制了一次性输入的batch_size大小不超过10。因此当知识库检索出过多知识时，会出现如下报错：
2025-08-01 16:15:02 - ERROR - __main__ - Error: Error code: 400 - {'error': {'code': 'InvalidParameter', 'param': None, 'message': '<400> InternalError.Algo.InvalidParameter: Value error, batch size is invalid, it should not be larger than 10.: input.contents', 'type': 'InvalidParameter'}, 'id': '31e5f835-8113-9318-ac76-d049c267', 'request_id': '31e5f5-8113-9318-ac76-d0b3d947'}
How to reproduce
在这里我给出我的解决方案，
修改./KAG-master/kag/common/text_sim_by_vector.py这个文件的63行，添加如下代码：
if len(need_call_emb_text) > 0:
if len(need_call_emb_text) <= 10:
emb_res = self.vectorize_model.vectorize(need_call_emb_text)
else:
emb_res = []
for i in range((len(need_call_emb_text) + 9 )// 10):
emb_res_part = self.vectorize_model.vectorize(need_call_emb_text[i*10:(i+1)*10])
emb_res += emb_res_part
以上只是我的个人不成熟的见解。
Are you willing to submit PR?
- Yes I am willing to submit a PR!