Zhipu API issues

[3414391918]

Does this project currently not support the model interface of the Zhipu AI platform

[xuhe2]

No in the code, like this function, you can see that all the api_base need the suffix v1, but Zhipu AI use v4 as suffix, I want to make a PR for Zhipu AI platform

[xuhe2]

I change the url in project, but it only works part of the time

    # Initialize embedding model
    if llm_service_type == "openai_chat":
        # Initialize OpenAI client for embeddings
        openai_client = OpenAI(
            api_key=embeddings_api_key or "dummy_key",
            base_url=f"{embeddings_api_base}/v1"
        )
    elif llm_service_type == "zhipu_chat":
        # Initialize Zhipu client for embeddings
        openai_client = OpenAI(
            api_key=embeddings_api_key or "dummy_key",
            base_url=f"{embeddings_api_base}/v4"
        )

I make some code like this, and the code above can work now, the log show that

15:25:52,743 graphrag.index.create_pipeline_config INFO Using LLM Config {
    "api_key": "*****",
    "type": "openai_chat",
    "model": "glm-4",
    "max_tokens": 8192,
    "temperature": 0.0,
    "top_p": 1.0,
    "n": 1,
    "request_timeout": 180.0,
    "api_base": "https://open.bigmodel.cn/api/paas/v4",
    "api_version": null,
    "organization": null,
    "proxy": null,
    "cognitive_services_endpoint": null,
    "deployment_name": null,
    "model_supports_json": true,
    "tokens_per_minute": 0,
    "requests_per_minute": 0,
    "max_retries": 3,
    "max_retry_wait": 10.0,
    "sleep_on_rate_limit_recommendation": true,
    "concurrent_requests": 25
}

the url has been changed in this place, but it also has

2024-08-03 15:25:54,254 - api - INFO - None
2024-08-03 15:25:54,267 - api - INFO - ⠋ GraphRAG Indexer
2024-08-03 15:25:54,267 - api - INFO - ├── Loading Input (InputFileType.text) - 1 files loaded (0 filtered) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00 0:00:00
2024-08-03 15:25:54,267 - api - INFO - ├── create_base_text_units
2024-08-03 15:25:54,267 - api - INFO - ├── create_base_extracted_entities
2024-08-03 15:25:54,267 - api - INFO - ├── create_summarized_entities
2024-08-03 15:25:54,268 - api - INFO - └── create_base_entity_graph❌ Errors occurred during the pipeline run, see logs for more details.
2024-08-03 15:25:54,517 - watchfiles.main - INFO - 2 changes detected
2024-08-03 15:25:55,119 - api - ERROR - Indexing failed

the details for the error is

15:25:53,762 httpx INFO HTTP Request: POST https://open.bigmodel.cn/api/paas/v1/chat/completions "HTTP/1.1 404 Not Found"
15:25:53,762 httpcore.http11 DEBUG receive_response_body.started request=<Request [b'POST']>
15:25:53,763 httpcore.http11 DEBUG receive_response_body.complete
15:25:53,763 httpcore.http11 DEBUG response_closed.started
15:25:53,763 httpcore.http11 DEBUG response_closed.complete
15:25:53,763 openai._base_client DEBUG HTTP Request: POST https://open.bigmodel.cn/api/paas/v1/chat/completions "404 Not Found"
15:25:53,763 openai._base_client DEBUG Encountered httpx.HTTPStatusError
Traceback (most recent call last):
  File "/home/xuhe/miniconda3/envs/graphrag-local/lib/python3.11/site-packages/openai/_base_client.py", line 1599, in _request
    response.raise_for_status()
  File "/home/xuhe/miniconda3/envs/graphrag-local/lib/python3.11/site-packages/httpx/_models.py", line 761, in raise_for_status
    raise HTTPStatusError(message, request=request, response=self)
httpx.HTTPStatusError: Client error '404 Not Found' for url 'https://open.bigmodel.cn/api/paas/v1/chat/completions'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/404

I guess this is because of openai package(but the Zhipu docs show it can use openai SDK), I don't know how to solve it :(