lunahr/Celune-1.7B-Upbeat
Partial fine-tune vs Qwen/Qwen3-TTS-12Hz-1.7B-Base. Partial change: 82% of the weights changed (attention, embeddings, lm_head, mlp, norms, other), in 28 of 28 layers.
Tensors changed: lm_head 14/15, Embeddings 9/17, Norms 73/134, MLP 103/103, Attention 132/132, other 3/3
1 derivatives checked
- mradermacher/Celune-1.7B-Upbeat-GGUF: Same weights, converted