Update requirements.txt

requirements.txt

@@ -1,15 +1,5 @@
-gradio
 torch
-transformers
+git+https://github.com/huggingface/transformers
 git+https://github.com/hubertsiuzdak/snac
 soundfile
 accelerate
-spaces
-1. ✅ **Increased max_tokens default**: 500 → 800 (critical for audio generation)
-2. ✅ **Increased GPU duration**: 120s → 180s (more time for longer generation)
-3. ✅ **Added generation parameters**: `repetition_penalty=1.1`, `no_repeat_ngram_size=3`
-4. ✅ **Better debugging**: Shows token ranges and unique token IDs
-5. ✅ **Updated slider range**: min=500, max=1500 for max_tokens
-6. ✅ **All examples use 800 tokens**: Ensures consistent results
-7. ✅ **Better error messages**: Shows expected vs actual token ranges
-8. ✅ **Explicit generation params**: Clearer parameter passing to `model.generate()`