gmakstutis/openrag
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fixed models service to try api key with first available model

Browse source
This commit is contained in:
Lucas Oliveira 2025-11-14 17:54:13 -03:00
parent 6edddbebca
commit a525b07f6f
1 changed files with 33 additions and 28 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

61
src/services/models_service.py
View file

@ -333,34 +333,6 @@ class ModelsService:
if project_id:
headers["Project-ID"] = project_id
# Validate credentials with a minimal completion request
async with httpx.AsyncClient() as client:
validation_url = f"{watson_endpoint}/ml/v1/text/generation"
validation_params = {"version": "2024-09-16"}
validation_payload = {
"input": "test",
"model_id": "ibm/granite-3-2b-instruct",
"project_id": project_id,
"parameters": {
"max_new_tokens": 1,
},
}
validation_response = await client.post(
validation_url,
headers=headers,
params=validation_params,
json=validation_payload,
timeout=10.0,
)
if validation_response.status_code != 200:
raise Exception(
f"Invalid credentials or endpoint: {validation_response.status_code} - {validation_response.text}"
)
logger.info("IBM Watson credentials validated successfully")
# Fetch foundation models using the correct endpoint
models_url = f"{watson_endpoint}/ml/v1/foundation_model_specs"
@ -424,6 +396,39 @@ class ModelsService:
}
)
# Validate credentials with the first available LLM model
if language_models:
first_llm_model = language_models[0]["value"]
async with httpx.AsyncClient() as client:
validation_url = f"{watson_endpoint}/ml/v1/text/generation"
validation_params = {"version": "2024-09-16"}
validation_payload = {
"input": "test",
"model_id": first_llm_model,
"project_id": project_id,
"parameters": {
"max_new_tokens": 1,
},
}
validation_response = await client.post(
validation_url,
headers=headers,
params=validation_params,
json=validation_payload,
timeout=10.0,
)
if validation_response.status_code != 200:
raise Exception(
f"Invalid credentials or endpoint: {validation_response.status_code} - {validation_response.text}"
)
logger.info(f"IBM Watson credentials validated successfully using model: {first_llm_model}")
else:
logger.warning("No language models available to validate credentials")
if not language_models and not embedding_models:
raise Exception("No IBM models retrieved from API")