takuma104/Qwen3-TTS-Tokenizer-12Hz-48kHz
Partial fine-tune vs Qwen/Qwen3-TTS-Tokenizer-12Hz. Partial change: 31% of the weights changed (other), in 5 of 31 layers.
Tensors changed: other 96/255, Norms 0/53, MLP 0/56, Attention 0/80, Embeddings 0/48
Partial fine-tune vs Qwen/Qwen3-TTS-Tokenizer-12Hz. Partial change: 31% of the weights changed (other), in 5 of 31 layers.
Tensors changed: other 96/255, Norms 0/53, MLP 0/56, Attention 0/80, Embeddings 0/48