Spaces:
Sleeping
Sleeping
File size: 5,089 Bytes
5a061ee d4abd8e 5a061ee |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 |
# π IMPORTANT: HuggingFace API Migration Notice
## Critical Update - November 2025
HuggingFace is migrating from their old Inference API endpoint to a new Inference Providers API.
**β οΈ UPDATE:** The new endpoint format is still being tested. **For now, we recommend using OpenAI or Anthropic for production use.**
---
## β οΈ What Changed
**Old Endpoint (DEPRECATED - Returns 404):**
```
https://api-inference.huggingface.co/models/{model}
```
**New Endpoint (REQUIRED as of Nov 1, 2025):**
```
https://router.huggingface.co/hf-inference/models/{model}
```
---
## β
This Version is Updated
**Good news:** If you're using this version of ConversAI, the update is already applied!
The code has been updated to use the new Inference Providers API automatically.
---
## π How to Verify
**Run the test script:**
```bash
python test_hf_backend.py
```
**Look for this line:**
```
β Using new Inference Providers API endpoint
```
**Or check manually:**
```bash
python check_env.py
```
---
## π¦ What You Need to Do
### If You Deployed BEFORE November 2025:
**Option 1: Update the Code (Recommended)**
1. Pull the latest version of `llm_backend.py`
2. Restart your application
3. Test with `python test_hf_backend.py`
**Option 2: Manual Update**
Edit `llm_backend.py` around line 80:
```python
# Find this line:
return f"https://api-inference.huggingface.co/models/{self.model}"
# Replace with:
return f"https://router.huggingface.co/hf-inference/models/{self.model}"
```
### If You're Deploying NEW:
**No action needed!** The latest code already uses the new endpoint.
---
## π Symptoms of Old Endpoint
If you're still using the old endpoint, you'll see:
```
β Error generating survey: Survey generation failed:
LLM generation failed: 404 Client Error: Not Found for url:
https://api-inference.huggingface.co/models/mistralai/Mixtral-8x7B-Instruct-v0.1
```
**Solution:** Update to the new endpoint as described above.
---
## π More Information
**Official HuggingFace Announcement:**
- Migration deadline: November 1, 2025
- Old endpoint returns 404 errors after deadline
- New API docs: https://huggingface.co/docs/inference-providers
**Benefits of New API:**
- Access to more models
- Better performance and reliability
- Unified API across all inference services
- Future-proof for new features
---
## β FAQ
**Q: Do I need a new API token?**
A: No, your existing HF_TOKEN or HUGGINGFACE_API_KEY works with the new endpoint.
**Q: Will this affect my deployed HuggingFace Space?**
A: Yes, if it was deployed before this update. Simply update the code and restart.
**Q: Do I need to change any configuration?**
A: No, just update the code. All configuration remains the same.
**Q: What if I use OpenAI or Anthropic?**
A: This change only affects HuggingFace users. Other providers are unaffected.
**Q: Will my old surveys/data still work?**
A: Yes, this is only an API endpoint change. All data formats remain compatible.
**Q: How do I know if I'm affected?**
A: If you see 404 errors when using HuggingFace provider, you need to update.
---
## β
Verification Checklist
After updating, verify everything works:
- [ ] Updated `llm_backend.py` to use new endpoint
- [ ] Ran `python test_hf_backend.py` - shows new endpoint
- [ ] Tested survey generation - works without 404 errors
- [ ] Tested translation - works correctly
- [ ] Tested analysis - works correctly
- [ ] No 404 errors in logs
- [ ] Application runs normally
---
## π Need Help?
**If you're still getting 404 errors after updating:**
1. **Verify the update:**
```bash
grep "router.huggingface.co" llm_backend.py
```
Should return a match. If not, update wasn't applied.
2. **Check you're using HuggingFace:**
```bash
python check_env.py
```
Look for which provider is being used.
3. **Test the connection:**
```bash
python test_hf_backend.py
```
Should show "β Using new Inference Providers API endpoint"
4. **Clear any caches:**
- Restart your application
- On HF Spaces: Factory reboot the Space
5. **Still stuck?**
- Check TROUBLESHOOTING.md
- Verify your HF_TOKEN is still valid
- Try with a different model
---
## π Migration Timeline
| Date | Event |
|------|-------|
| January 2025 | Old endpoint deprecated, warnings sent |
| November 1, 2025 | Old endpoint returns 404 errors |
| **Today** | **This version updated to new endpoint** |
---
## π Bottom Line
**If you're using this version:** β
You're all set! The update is already applied.
**If you deployed before:** π Update your code to the latest version.
**If you see 404 errors:** β οΈ Follow the update instructions above.
---
**Updated:** November 2025
**Status:** Migration Complete β
**Action Required:** Update existing deployments to latest code
For more details, see [CHANGELOG.md](CHANGELOG.md)
|