[object Object] as ERROR

[j2l]

Describe the bug

Here's the error in terminal (not browser console):

 INFO   stream-text  Sending llm call to HuggingFace with model Qwen/Qwen2.5-Coder-32B-Instruct
 ERROR   api.chat  [object Object]
 DEBUG   api.chat  usage {"promptTokens":null,"completionTokens":null,"totalTokens":null}

I read here that others could used Qwen2.5-Coder-32B-Instruct, would you please share how you do it?
Thanks!

Link to the Bolt URL that caused the error

http://localhost:5173/chat/1?rewindTo=d3cy3j0zj9b

Steps to reproduce

1. select HuggingFace
2. select Qwen2.5-Coder-32B-Instruct
3. enter prompt
4. hit enter

Expected behavior

code 😄

Screen Recording / Screenshot

No response

Platform

• OS: Ubuntu 22.04
• Browser: Chrome
• Version: 134.0.6998.35 (Official Build) (64-bit)

Provider Used

HF

Model Used

Qwen2.5-Coder-32B-Instruct

Additional context

Running bolt.diy with npm run dev

[leex279]

Take a look here, maybe it helps: https://thinktank.ottomator.ai/t/how-to-use-hugging-faces/6231/2

[j2l]

@leex279 this was not a question
ERROR api.chat [object Object] is not helpful as an error 😸
My API key is full write so it's not an issue.

Redoing it, I have another message in the browser:
There was an error processing your request: Custom error: Input validation error: inputs tokens + max_new_tokens must be <= 16000. Given: 146038 inputs tokens and 8000 max_new_tokens

Still, this error should also be in terminal.

[leex279]

@thecodacus can you please take a look