idle-intelligence/pocket-tts-gguf

Full fine-tune vs kyutai/pocket-tts-without-voice-cloning. idle-intelligence/pocket-tts-gguf is a GGUF (Q8_0) of different weights. Fine-tune: 96% of the weights changed, across 6 of 6 layers.

Tensors changed: other 50/80, Embeddings 13/15, MLP 24/24, Norms 26/34, Attention 16/16

On the Hub · Downloads on Model Pulse