transformers - 💡(How to fix) Fix Unauthorized error for non gated models also [6 comments, 3 participants]

I am continuously getting the same unauthorized error for every model i am trying to use. This is the error coming for every model. STILL ERRORING? Check this: 404 Client Error: Not Found for url: https://router.huggingface.co/hf-inference/models/microsoft/Phi-3-mini-4k-instruct/v1/chat/completions (Request ID: Root=1-69a3dacb-11cd4511013222d66bfd0a28;6ec1e299-5834-4924-8898-82dab3e982e0)

Problem Summary

Unauthorized error for model access

Root Cause Analysis

Token access not properly configured

Fix Plan

Step 1: Verify Token Configuration

Check if the token access is correctly set up in the environment variables or configuration files.

Step 2: Update Token Type

Update the token type to match the required type for the models being used.

Step 3: Validate Model URL

Verify that the model URL is correct and accessible.

Step 4: Update Code to Use Token

Update the code to use the token for authentication.

Example code snippet in Python:

import requests

# Set token access
token = "YOUR_TOKEN_ACCESS"
headers = {"Authorization": f"Bearer {token}"}

# Set model URL
model_url = "https://router.huggingface.co/hf-inference/models/microsoft/Phi-3-mini-4k-instruct/v1/chat/completions"

# Send request with token
response = requests.post(model_url, headers=headers)

# Check response status code
if response.status_code == 200:
    print("Request successful")
else:
    print("Request failed with status code:", response.status_code)

Verification

Verify that the fix works by checking the response status code and the model output.

Extra Tips

• Make sure to replace YOUR_TOKEN_ACCESS with the actual token access.
• Verify that the model URL is correct and accessible.
• Check the token expiration and refresh it if necessary.