Issue with Tabby Completions Choices - Text Field Empty

[vannchic]

I am experiencing an issue where the text field in Tabby completions choices is always empty. Additionally, the system shows "Unreachable: Failed to connect to the completion model." I am using OpenAI's model, and I can confirm that I have normal access to OpenAI services. Can anyone help me troubleshoot this issue?

version V0.23.0

~/.tabby/config.toml

# Chat model
[model.chat.http]
kind = "openai/chat"
model_name = "gpt-4o"  # Please make sure to use a chat model, such as gpt-4o
api_endpoint = "http://api.openai.com/v1"   # DO NOT append the `/chat/completions` suffix
api_key = "skxxx"

# Completion model
[model.completion.http]
kind = "openai/completion"
model_name = "gpt-4o"   # Please make sure to use a completion model, such as gpt-3.5-turbo-instruct
api_endpoint = "http://api.openai.com/v1"   # DO NOT append the `/chat/completions` suffix
api_key = "skxxx"

# Embedding model
[model.embedding.http]
kind = "openai/embedding"
model_name = "text-embedding-3-large"   # Please make sure to use a embedding model, such as text-embedding-3-small
api_endpoint = "http://api.openai.com/v1"   # DO NOT append the `/chat/completions` suffix
api_key = "skxxx"

[vannchic]

I tried changing the OpenAI model to gpt-3.5-turbo-instruct. After this change, the system indicates that it is connected. However, when I actually use it, the completions still return empty results. Any insights on how to resolve this would be appreciated.

[zwpaper]

Hi @vannchic, I've noticed you used http in your api_endpoint = "http://api.openai.com/v1".

This seems to be a typo, it should be https, making the correct endpoint api_endpoint = "https://api.openai.com/v1".

Could you please try that again? Just so you know, all three URLs are incorrect.

[vannchic]

Hi @vannchic, I've noticed you used in your .http``api_endpoint = "http://api.openai.com/v1"

This seems to be a typo, it should be , making the correct endpoint .https``api_endpoint = "https://api.openai.com/v1"

Could you please try that again? Just so you know, all three URLs are incorrect.

Thank you for pointing that out. I double-checked to ensure that my httpsapi_endpoint and secret are correct, and I can confirm that the connection to OpenAI is functioning properly, as shown in the attached screenshot. The chat feature is working as expected, but the completions endpoint is returning an empty response.

I appreciate your assistance and am open to any further suggestions you might have.

[wsxiaoys]

openai no longer provides official support for fim models (they only provide support for chat models nowadays) - please try a different model vendors for code completion.