Downloads · 30 days
15
12% of all-time downloads
vibevoice-1.5b-bnb-4bit
tantk/vibevoice-1.5b-bnb-4bit
vibevoice-1.5b-bnb-4bit is a text-to-speech model from tantk. Use it when you need text read aloud. It is set up for transformers. The card lists the license as mit.
4-bit NF4 quantization of microsoft/VibeVoice-1.5B.
All-time downloads
130
Public
Parameters
1.6B
3.5 GB on disk
Likes
1
Public
Repo makeup
Click a slice to open those files.
.safetensors3.5 GB · 100%
Parameter types
How the weights are stored.
U81.4B · 85%
Base models
Model card
From the Hugging Face model README
VibeVoice-1.5B - NF4 Quantized
4-bit NF4 quantization of microsoft/VibeVoice-1.5B.
Strategy
Backbone extraction approach:
- Downloaded raw safetensors (bypassed from_pretrained)
- Separated Qwen2.5-1.5B backbone from audio heads
- Quantized backbone as standard Qwen2ForCausalLM with NF4 + double quant
- Packaged quantized backbone + BF16 audio heads
| Component | Method | Size |
|---|---|---|
| LLM backbone (Qwen2.5-1.5B) | NF4 + double quant | ~0.8-1.0 GB |
| Audio heads (tokenizers, diffusion, connectors) | BF16 | ~1.8 GB |
Source
Quantized from microsoft/VibeVoice-1.5B (MIT license).
- Task
- Text-to-Speech
- Library
- transformers
- Type
- qwen2
- License
- mit
- Created
- Apr 2, 2026
- Updated
- Apr 2, 2026