ASMR LLM director (LoRA)

LoRA adapter for Qwen3.5-0.8B trained to write Chinese ASMR roleplay lines with [tone:...] control tokens. 14,389 instruction rows from the WebASMR segment corpus (VibeVoice-ASR segments reconciled with Qwen3-ASR and 101 human labels).

Training: 10 epochs, LoRA r=16, explicit prompt masking, EOS appended.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support