Ex0bit/MiniMax-SLURPY
Full fine-tune vs MiniMaxAI/MiniMax-M2.5. Fine-tune: 100% of the weights changed, across 62 of 62 layers.
Tensors changed: other 124/124, MLP 2129/2129, Attention 496/496, Embeddings 1/1, Norms 249/249, lm_head 1/1
Full fine-tune vs MiniMaxAI/MiniMax-M2.5. Fine-tune: 100% of the weights changed, across 62 of 62 layers.
Tensors changed: other 124/124, MLP 2129/2129, Attention 496/496, Embeddings 1/1, Norms 249/249, lm_head 1/1