[Bug]: infiniflow/ragflow:v0.14.0 BAICHUAN embedding model trigger exception

[neo-workship]

Is there an existing issue for the same bug?

• I have checked the existing issues.

RAGFlow workspace code commit ID

infiniflow/ragflow:v0.14.0 镜像

RAGFlow image version

v0.14.0 full

Other environment information

install environment： Docker Desktop
ragflow version： v0.14.0 full

Actual behavior

test uses 百川 Baichuan-Text-Embedding, ollama ：nomic-embed-text ， tongyi ：Text-Embedding-v3

using the Knowledge Graph method parse chinese. The problem is only with the Baichuan-text-embedding (the interface of Baichuan is compatible with the openai API, so the embedding model can be used when I use graphrag alone).

Expected behavior

log

2024-11-26 16:40:52 2024-11-26 16:40:52,640 WARNING  17 /ragflow/.venv/lib/python3.10/site-packages/networkx/readwrite/json_graph/node_link.py:142: FutureWarning: 
2024-11-26 16:40:52 The default value will be `edges="edges" in NetworkX 3.6.
2024-11-26 16:40:52 
2024-11-26 16:40:52 To make this warning go away, explicitly set the edges kwarg, e.g.:
2024-11-26 16:40:52 
2024-11-26 16:40:52   nx.node_link_data(G, edges="links") to preserve current behavior, or
2024-11-26 16:40:52   nx.node_link_data(G, edges="edges") for forward compatibility.
2024-11-26 16:40:52   warnings.warn(
2024-11-26 16:40:52 
2024-11-26 16:40:52 2024-11-26 16:40:52,973 INFO     17 HTTP Request: POST https://api.deepseek.com/v1/chat/completions "HTTP/1.1 200 OK"

............

2024-11-26 16:41:26 openai.BadRequestError: Error code: 400 - {'error': {'code': None, 'param': None, 'type': 'invalid_request_error', 'message': 'Input batch list exceeds the limit of 16.'}}
2024-11-26 16:41:26 2024-11-26 16:41:26,051 ERROR    17 handle_task got exception for task {"id": "431446bcabd111ef999b0242ac120006", "doc_id": "4147acacabd111ef8f5c0242ac120006", "from_page": 0, "to_page": 100000000, "retry_count": 0, "kb_id": "2e0a28feabd111ef8abc0242ac120006", "parser_id": "knowledge_graph", "parser_config": {"entity_types": ["organization", "person", "location", "event", "time"], "chunk_token_num": 8192, "delimiter": "\\n!?;\u3002\uff1b\uff01\uff1f", "layout_recognize": true}, "name": "hd.txt", "type": "doc", "location": "hd.txt", "size": 1826, "tenant_id": "01a5f070abce11efbf5a0242ac120006", "language": "Chinese", "embd_id": "Baichuan-Text-Embedding@BaiChuan", "img2txt_id": "qwen-vl-max@Tongyi-Qianwen", "asr_id": "paraformer-realtime-8k-v1@Tongyi-Qianwen", "llm_id": "deepseek-chat@DeepSeek", "update_time": 1732610073371}
2024-11-26 16:41:26 Traceback (most recent call last):
2024-11-26 16:41:26   File "/ragflow/rag/svr/task_executor.py", line 448, in handle_task
2024-11-26 16:41:26     do_handle_task(task)
2024-11-26 16:41:26   File "/ragflow/rag/svr/task_executor.py", line 402, in do_handle_task
2024-11-26 16:41:26     tk_count, vector_size = embedding(cks, embd_mdl, r["parser_config"], callback)
2024-11-26 16:41:26   File "/ragflow/rag/svr/task_executor.py", line 306, in embedding
2024-11-26 16:41:26     vts, c = mdl.encode(cnts[i: i + batch_size])
2024-11-26 16:41:26   File "/ragflow/api/db/services/llm_service.py", line 209, in encode
2024-11-26 16:41:26     emd, used_tokens = self.mdl.encode(texts, batch_size)
2024-11-26 16:41:26   File "/ragflow/rag/llm/embedding_model.py", line 106, in encode
2024-11-26 16:41:26     res = self.client.embeddings.create(input=texts,
2024-11-26 16:41:26   File "/ragflow/.venv/lib/python3.10/site-packages/openai/resources/embeddings.py", line 125, in create
2024-11-26 16:41:26     return self._post(
2024-11-26 16:41:26   File "/ragflow/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 1260, in post
2024-11-26 16:41:26     return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
2024-11-26 16:41:26   File "/ragflow/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 937, in request
2024-11-26 16:41:26     return self._request(
2024-11-26 16:41:26   File "/ragflow/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 1041, in _request
2024-11-26 16:41:26     raise self._make_status_error_from_response(err.response) from None
2024-11-26 16:41:26 openai.BadRequestError: Error code: 400 - {'error': {'code': None, 'param': None, 'type': 'invalid_request_error', 'message': 'Input batch list exceeds the limit of 16.'}}
2024-11-26 16:41:32 2024-11-26 16:41:32,581 INFO     17 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-11-26T16:41:32.580988", "boot_at": "2024-11-26T16:10:31.078836", "pending": 0, "lag": 0, "done": 1, "failed": 1, "current": null}
2024-11-26 16:41:37 2024-11-26 16:41:37,252 INFO     18 172.18.0.6 - - [26/Nov/2024 16:41:37] "GET /v1/document/list?kb_id=2e0a28feabd111ef8abc0242ac120006&keywords=&page_size=10&page=1 HTTP/1.1" 200 -
2024-11-26 16:42:02 2024-11-26 16:42:02,612 INFO     17 task_consumer_0 reported heartbeat: {"name": "task_consumer_0", "now": "2024-11-26T16:42:02.611833", "boot_at": "2024-11-26T16:10:31.078836", "pending": 0, "lag": 0, "done": 1, "failed": 1, "current": null}

Steps to reproduce

- select Knowledge Graph method
- select embedding model: Baichuan-Text-Embedding 、ollama nomic-embed-text 、tongyi text-embedding-v3
- only Baichuan-Text-Embedding Trigger error，it's api Compatible with oepnai api，but using graphrag alone, no problem

Additional information

No response