[Bug]: Paper chunk method does not support Zhipu GLM model for OCR

[TraceIvan]

Is there an existing issue for the same bug?

• I have checked the existing issues.

RAGFlow workspace code commit ID

9d717f0

RAGFlow image version

9d717f0

Other environment information

Actual behavior

When using the glm-4-flash and glm-4v-flash models, the following error is reported when parsing a PDF document using the paper chunk method:

Knowledge graph extraction error:**ERROR**: Completions.create() got an unexpected keyword argument 'logit_bias'

Expected behavior

No response

Steps to reproduce

Add glm-4v-flash model information in the file llm_factories.json:

{
     "llm_name": "glm-4v-flash",
     "tags": "LLM,CHAT,IMAGE2TEXT",
     "max_tokens": 8000,
     "model_type": "image2text"
},

Set the zhipu api key and system model in user setting. Then upload a PDF file in the knowledge base, select the paper chunk method, and select "glm-4v-flash" in the "Layout recognition & OCR" option. After starting the parsing, the above error will appear.

Additional information

No response