jacklishufan/LaViDa-O-v1.0
Full fine-tune (fewer layers) vs GSAI-ML/LLaDA-8B-Instruct. GSAI-ML/LLaDA-8B-Instruct matches part of jacklishufan/LaViDa-O-v1.0 (289 of its 1266 tensors; the rest of jacklishufan/LaViDa-O-v1.0 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Attention 128/128, Norms 65/65, other 64/64, MLP 32/32