HOW_TO_UPLOAD_TO_HF_SPACES.md

# How to Upload app.py to HuggingFace Spaces

## ✅ Good News: app.py is Already Fixed!

Your local `app.py` file is correctly configured with:
- ✅ `USE_HF_API = "True"` (line 143)
- ✅ `USE_LMSTUDIO = "False"` (line 144)
- ✅ `LLM_BACKEND = "hf_api"` (line 145)
- ✅ `LLM_TIMEOUT = "180"` (line 147)

**Location**: `/home/john/TranscriptorEnhanced/app.py`
**File size**: 44KB
**Last modified**: Oct 30, 18:04

---

## 🚀 How to Upload to Your HuggingFace Space

### **Method 1: Via HuggingFace Web Interface (Easiest)**

1. **Open your Space in browser**
   - Go to: https://huggingface.co/spaces/YOUR_USERNAME/YOUR_SPACE

2. **Click "Files" tab**

3. **Click on "app.py"** to open it

4. **Click the "Edit" button** (pencil icon)

5. **Delete ALL content** in the editor (Ctrl+A, Delete)

6. **Open your local file**:
   ```bash
   # On your local machine, open:
   /home/john/TranscriptorEnhanced/app.py
   ```

7. **Copy ALL content** from your local file (Ctrl+A, Ctrl+C)

8. **Paste into HF editor** (Ctrl+V)

9. **Click "Commit changes to main"**

10. **Wait 2-3 minutes** for Space to rebuild

---

### **Method 2: Via Git (If You Have Git Access)**

If you cloned your Space repository:

```bash
# Navigate to your Space repo
cd /path/to/your-space-repo

# Copy the fixed file
cp /home/john/TranscriptorEnhanced/app.py .

# Commit and push
git add app.py
git commit -m "Fix: Force HF API mode to resolve timeout errors"
git push
```

---

### **Method 3: Direct File Upload**

1. **Go to your Space**: https://huggingface.co/spaces/YOUR_USERNAME/YOUR_SPACE

2. **Click "Files" tab**

3. **Click "Upload files"** button

4. **Select your local file**:
   ```
   /home/john/TranscriptorEnhanced/app.py
   ```

5. **Choose**: "Overwrite existing file"

6. **Click "Commit"**

---

## ✅ Verification After Upload

### **Step 1: Check the Logs**

After your Space restarts, click the **"Logs"** tab and look for:

```
🚀 Forcing HF API mode for HuggingFace Spaces deployment...
✅ HuggingFace token detected
✅ Configuration loaded for HuggingFace Spaces
🚀 TranscriptorAI Enterprise - LLM Backend: hf_api
🔧 USE_HF_API: True
🔧 USE_LMSTUDIO: False
🔧 LLM_TIMEOUT: 180s
```

**Good signs**:
- ✅ "Forcing HF API mode"
- ✅ "HuggingFace token detected"
- ✅ "LLM Backend: hf_api"
- ✅ "USE_HF_API: True"

**Bad signs** (means old file still there):
- ❌ "LLM Backend: local"
- ❌ "USE_HF_API: False"
- ❌ "Loading local model: microsoft/Phi-3"

### **Step 2: Test Processing**

Upload a test transcript and check logs for:

```
INFO: Calling HF API: microsoft/Phi-3-mini-4k-instruct
```

**Should NOT see**:
```
INFO: Generating with local model
ERROR: LLM generation timed out
```

### **Step 3: Check Quality Score**

After processing completes:
- ✅ Quality Score should be **0.70-1.00** (not 0.00)
- ✅ Processing time: **30-60 minutes** for 10 files (not hours)
- ✅ No timeout errors

---

## 🔍 Troubleshooting

### Issue: "Still seeing timeout errors"

**Check 1**: Verify file was uploaded
- Go to Files tab in your Space
- Click app.py
- Search for line 143
- Should say: `os.environ["USE_HF_API"] = "True"`
- If it says `setdefault` instead, the file wasn't uploaded correctly

**Check 2**: Verify token is set
- Go to Settings tab
- Look for HUGGINGFACE_TOKEN in secrets
- If not there, add it from https://huggingface.co/settings/tokens

**Check 3**: Force rebuild
- Settings tab → "Factory reboot"
- This clears all caches and rebuilds from scratch

### Issue: "Logs show 'USE_HF_API: False'"

**Cause**: Old file still being used

**Fix**:
1. Delete app.py from your Space
2. Upload the fixed version again
3. Factory reboot

### Issue: "HuggingFace token detected" not showing

**Cause**: Token not set in Space secrets

**Fix**:
1. Go to: https://huggingface.co/settings/tokens
2. Create new token (type: Read)
3. Go to Space → Settings → Repository secrets
4. Add: Name=HUGGINGFACE_TOKEN, Value=(your token)
5. Factory reboot

---

## 📋 Quick Checklist

Before upload:
- [x] Local app.py has `USE_HF_API = "True"` on line 143 ✓ (already confirmed)
- [ ] HUGGINGFACE_TOKEN is set in Space secrets
- [ ] Ready to upload file

After upload:
- [ ] File uploaded successfully
- [ ] Space rebuilt (takes 2-3 minutes)
- [ ] Logs show "Forcing HF API mode"
- [ ] Logs show "USE_HF_API: True"
- [ ] Logs show "LLM Backend: hf_api"
- [ ] Test transcript processes without timeout
- [ ] Quality Score > 0.00

---

## 🎯 Expected Timeline

1. **Upload file**: 1 minute
2. **Space rebuild**: 2-3 minutes
3. **First transcript test**: 5-10 minutes (for a typical file)
4. **Total**: ~15 minutes to confirm it's working

---

## 📞 If Still Not Working

After uploading and waiting for rebuild, if you still see timeout errors:

1. **Copy the startup logs** (first 50 lines)
2. **Copy the error logs** (when processing fails)
3. **Check these specific lines**:
   - Line showing "LLM Backend: ???"
   - Line showing "USE_HF_API: ???"
   - Line showing "Calling HF API" or "Generating with local model"

This will help diagnose if the file uploaded correctly or if there's another issue.

---

## ✅ File is Ready - Just Upload It!

Your local file at `/home/john/TranscriptorEnhanced/app.py` is **100% correct**.

**All you need to do is**:
1. Copy it to your HuggingFace Space (Method 1, 2, or 3 above)
2. Wait for rebuild
3. Test!

The timeout issue will be **completely resolved** once this file is on your Space. 🚀