shiningstar1128/llm-trainer-v08

QWEN3Vox voice-design TTS for the Vocence subnet.

Hugging Face repo id

model_name in vocence_config.yaml, on-chain commit, and chute wrapper VOCENCE_REPO must all be:

shiningstar1128/llm-trainer-v08

Validator /speak contract

Validators POST JSON (natural-language prompts after source extraction):

{
  "text": "<transcription to synthesize>",
  "instruction": "An adult female with an American accent, speaking at a normal pace in a mid-range pitch, sounding neutral and formal throughout"
}

• text — spoken content (verbatim; length capped by limits.max_text_chars).
• instruction — single-sentence voice description covering gender, age, accent, emotion, tone, speed, and pitch (not the old gender: x | pitch: y form).

The canonical wrapper forwards both fields unchanged to Miner.generate_wav(instruction, text).

How this engine uses them (verbatim):

• instruction is written directly to the processor system_prompt (no trait parsing or paraphrasing).
• text is passed directly as the synthesis script (plain transcript or existing Speaker N: lines; no auto-wrapping).
• A fixed default aux audio shard supplies acoustic prefill only; it is not chosen from instruction keywords.

Repo layout

File	Purpose
miner.py	QWEN3Vox engine (Miner class)
vocence_config.yaml	model_name + runtime/generation limits
chute_config.yml	Chutes image, GPU (64 GB VRAM), pip deps
config.json	Model config
model-*.safetensors	Sharded weights (≥ 50 MiB total required)
aux_lm_residual_projection.safetensors	Aux conditioning shard
tokenizer.json, tokenizer_config.json, vocab.json, merges.txt	Tokenizer (must ship in-repo)

Deploy

1. Upload this folder to shiningstar1128/llm-trainer-v08 and pin a 40-char commit SHA.
2. Build/deploy: deploy/deploy_chutes_08.py (VOCENCE_REPO must match this repo).
3. Commit: uv run vocence miner commit --model-name shiningstar1128/llm-trainer-v08 --model-revision <sha> --chute-id <uuid>.

See miner_sample/MINER_GUIDE.md for anti-cheat rules (from_pretrained(model_name) only, no prompt enrichment).