Kinyarwanda Orpheus v0.6 (Backup β€” NOT WORKING)

Status: NO CLEAR VOICE β€” This is a backup checkpoint only. Do not use for inference.

Details

  • Base: vysakh25/Vysakh-Orpheus-3B-Kinyarwanda-v0.5 (merged)
  • Data: AfriVoice 40K + Common Voice 5K (45K clips)
  • Training: 21,000 steps (37%), loss 3.92, epoch 14.9
  • Result: Partial speech on 1/3 test prompts (t2 only), very quiet, no consistent voice

Version Chain

v0.1 β†’ v0.2 β†’ v0.3 (first speech) β†’ v0.4 (silent, batch too big) β†’ v0.5 (best speech) β†’ v0.6 (this, partial) β†’ v0.7 (RBA data, training...)

Downloads last month
50
Safetensors
Model size
3B params
Tensor type
BF16
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support