Making use of Proxy for the prompt optimization

[noamKayzer]

Hey.

I'm trying to use the promptim optimization with azure client.
Its seems like configuring for doing so isn't do much (it keeps ask for the regular OPENAPI_KEY)
Is the use of an OpenAI (or Claude) proxy implemented yet, or is there a problem with my configuration?

part of my config file:
"optimizer": {
"model": {
"model": "gpt4o",
"model_provider": "azure_openai",
"max_tokens_to_sample": 8192,
"configurable_fields": "any",
"api_key": "db64482d5b7d-------------1",
"deployment_name": "gpt-4o",
"api_version": "2024-08-01-preview",
"base_url": "https://----------------.openai.azure.com"
}
}
Thanks

[hinthornw]

Could you test again on verson 0.0.7? Thank you! 🙏

[Squidward1012]

Could you test again on verson 0.0.7? Thank you! 🙏

{
  "name": "Test",
  "dataset": "tweet-optim",
  "description": "test task",
  "evaluators": "./task.py:evaluators",
  "evaluator_descriptions": {
    "my_example_criterion": "CHANGEME: This is a description of what the example criterion is testing. It is provided to the metaprompt to improve how it responds to different results."
  },
  "optimizer": {
    "model": {
      "model": "qwen2.5-72b-instruct",
      "base_url": "https://dashscope.aliyuncs.com/compatible-mode/v1",
      "model_provider": "openai"
    }
  },
  "initial_prompt": {
    "identifier": "tweet-generator-example-with-nothing"
  },
  "$schema": "https://raw.githubusercontent.com/hinthornw/promptimizer/refs/heads/main/config-schema.json"
}

File "C:\Users\23939.conda\envs\3.12\Lib\site-packages\openai_base_client.py", line 1634, in _request
raise self._make_status_error_from_response(err.response) from None
openai.NotFoundError: Error code: 404 - {'error': {'message': 'The model gpt-4o-mini does not exist or you do not
have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}, 'request_id':
'3125e679-3b43-9990-bc94-6a5321d11279'}
Getting baseline scores... ---------------------------------------- 100% 0:00:00
Error running target function: Error code: 404 - {'error': {'message': 'The model gpt-4o-mini does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}, 'request_id': 'de583ec4-f4b7-9997-8407-672e56e2f288'}

[noamKayzer]

Could you test again on verson 0.0.7? Thank you! 🙏

Its seems like nothing change after upgrading, it keeps looking for the API-KEY for the regular openai client.

raise OpenAIError(
openai.OpenAIError: The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable

[EtienneFerrandi]

{
  "name": "agent_smade",
  "dataset": "examples_Q/A",
  "description": "prompt optimization",
  "evaluators": "./task.py:evaluators",
  "evaluator_descriptions": {
    "accuracy": "Accuracy evaluator based on semantic similarity."
  },
  "optimizer": {
    "model": {
      "model": "Qwen/Qwen3-30B-A3B",
      "base_url": "https://api.studio.nebius.com/v1/",
      "model_provider": "openai"
    }
  },
  "initial_prompt": {
    "identifier": "prompt_agent_smade"
  },
  "$schema": "https://raw.githubusercontent.com/hinthornw/promptimizer/refs/heads/main/config-schema.json"
}

TypeError: "Could not resolve authentication method. Expected either api_key or auth_token to be set. Or for one of the `X-Api-Key` or `Authorization` headers to be explicitly omitted"
Error running target function: "Could not resolve authentication method. Expected either api_key or auth_token to be set. Or for one of the `X-Api-Key` or `Authorization` headers to be explicitly omitted"