HumeAI/tada-1b
Full fine-tune (fewer layers) vs meta-llama/Llama-3.2-1B. meta-llama/Llama-3.2-1B matches part of HumeAI/tada-1b (146 of its 394 tensors; the rest of HumeAI/tada-1b has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 16 of 16 layers.
Tensors changed: Embeddings 1/1, Norms 33/33, MLP 48/48, Attention 64/64
1 derivatives checked
- cstr/tada-tts-1b-GGUF: Not comparable yet
1 other repos hold byte-for-byte identical weight files.