AxionML/Inkling-Small-NVFP4
Partial fine-tune vs thinkingmachines/Inkling-Small. Partial change: 0% of the weights changed (mlp), in 40 of 42 layers.
Tensors changed: MLP 80/282, Attention 0/450, Norms 0/214, other 0/13, Embeddings 0/11
Partial fine-tune vs thinkingmachines/Inkling-Small. Partial change: 0% of the weights changed (mlp), in 40 of 42 layers.
Tensors changed: MLP 80/282, Attention 0/450, Norms 0/214, other 0/13, Embeddings 0/11