Downloads · 30 days
0
moshi-phase1b-rerun-2026-07-06
japanese-speech-models/moshi-phase1b-rerun-2026-07-06
moshi-phase1b-rerun-2026-07-06 is a machine learning model from japanese-speech-models. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for moshi. The card lists the license as cc-by-nc-4.0.
Fine-tuned from nu-dialogue/j-moshi-ext with 100 synthetic dialogs (Irodori-TTS Full FT amitaro voice, generated 2026-07-04) for voice-transfer validation in the Path B research programme.
Access
Public
Updated Jul 6, 2026
Parameters
7.7B
15.4 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors15.4 GB · 100%
At a glance
- Library
- moshi
- License
- cc-by-nc-4.0
- Access
- Public
- Created
- Jul 6, 2026
- Updated
- Jul 6, 2026
- SHA
- 46cb4abd
Base models
Model card
From the Hugging Face model README
Phase 1b re-run: J-Moshi-ext + Path B synthetic dialog FT
Fine-tuned from nu-dialogue/j-moshi-ext with 100 synthetic dialogs
(Irodori-TTS Full FT amitaro voice, generated 2026-07-04) for
voice-transfer validation in the Path B research programme.
Training config
- Base model: nu-dialogue/j-moshi-ext (with
--extend_modules_for_user_stream→ dep_q=16 during train) - Data: 100 synthetic dialogs (612 turns, 31.7 min stereo 24kHz)
- Params: all 7.5B trainable (
PARAMS_TO_FT=all) - Steps: 65 (5 epoch × 13 iter, effective batch 8)
- LR: 3e-5 (tempformer + depformer)
- DeepSpeed: ZeRO-3 + bf16 + CPU offload optimizer
- GPU: A100 SXM4 40GB × 2 (Slovenia vast.ai, $1.334/hr)
- Wall clock: ~1 h train + ~7 min consolidate
- Final loss: 1.27 (text 0.48 + audio 0.79)
- Cleaned: dep_q 16 → 8 via
tools.clean_moshi --remove_modules_for_user_stream
Files
model.safetensors(15 GB, bf16, dep_q=8) — inference-readymoshi_lm_kwargs.json— moshi LM config
- Library
- moshi
- License
- cc-by-nc-4.0
- Languages
- ja
- Created
- Jul 6, 2026
- Updated
- Jul 6, 2026