Use our OpenAI chat completion endpoint
Atla provides a drop-in replacement for OpenAI’s /v1/chat/completions
API. You can use it with OpenAI-compatible SDKs (e.g. openai
, LangChain, LlamaIndex) by simply pointing to our base URL and API key. Please refer to the models page for the list of supported models.
If you don’t have an Atla API key, you can get one here.
Mapping between the Atla evaluation endpoint and fields in this prompt template
If you are familiar with the Atla evaluation endpoint, here is a quick note on how the fields on the endpoint map to the fields in the prompt template.
Atla Evaluation endpoint | Prompt template |
---|---|
model_input | Instruction |
model_output | Response |
evaluation_criteria | Score Rubrics |
critique | Reasoning |
score | Result |
Your prompt should include these fields and should follow this template closely.
If you are using atla-selene-mini
we strongly advise you to use the prompt templates linked in this github
page.
Use our OpenAI chat completion endpoint
Atla provides a drop-in replacement for OpenAI’s /v1/chat/completions
API. You can use it with OpenAI-compatible SDKs (e.g. openai
, LangChain, LlamaIndex) by simply pointing to our base URL and API key. Please refer to the models page for the list of supported models.
If you don’t have an Atla API key, you can get one here.
Mapping between the Atla evaluation endpoint and fields in this prompt template
If you are familiar with the Atla evaluation endpoint, here is a quick note on how the fields on the endpoint map to the fields in the prompt template.
Atla Evaluation endpoint | Prompt template |
---|---|
model_input | Instruction |
model_output | Response |
evaluation_criteria | Score Rubrics |
critique | Reasoning |
score | Result |
Your prompt should include these fields and should follow this template closely.
If you are using atla-selene-mini
we strongly advise you to use the prompt templates linked in this github
page.