HKUSTAudio/Llasa-1B

Full fine-tune vs meta-llama/Llama-3.2-1B-Instruct. Fine-tune: 100% of the weights changed, across 16 of 16 layers.

Tensors changed: Norms 33/33, MLP 48/48, Attention 64/64

2 derivatives checked

• unsloth/Llasa-1B: Identical copy
• SebastianBodza/Kartoffel-1B-v0.3: Full fine-tune

3 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse