YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
MOSS VoiceGenerator — GRPO LoRA (Rank 8)
Overview
- Base model: OpenMOSS-Team/MOSS-VoiceGenerator (Qwen3-1.7B backbone)
- Method: GRPO (Group Relative Policy Optimization)
- LoRA rank: 8, alpha: 16
- Training: 500 steps, batch=8, G=4, lr=5e-5
- Rewards: Speaker similarity (ECAPA-TDNN, w=0.6) + CLAP emotion (w=0.4) + WER penalty
Results (500 steps)
| Metric | Step 1-100 | Step 401-500 | Improvement |
|---|---|---|---|
| Speaker Sim | 0.378 | 0.450 | +19% |
| CLAP | 0.183 | 0.231 | +26% |
| WER | 0.225 | 0.135 | -40% |
Eval (1 sample per condition)
| Model | Mean Sim | Success Rate |
|---|---|---|
| Baseline | 0.264 | 56/75 |
| GRPO r8 | 0.355 | 75/75 |
Usage
from transformers import AutoModel
from peft import PeftModel
model = AutoModel.from_pretrained("OpenMOSS-Team/MOSS-VoiceGenerator", trust_remote_code=True)
model.language_model = PeftModel.from_pretrained(model.language_model, "laion/voicenet-1.7B-wip", subfolder="grpo-r8-500steps")
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support