Ttimofeyka/bitnet-5B-v1
Partial fine-tune vs 1bitLLM/bitnet_b1_58-3B. Partial change: 37% of the weights changed (attention, mlp, norms), in 10 of 26 layers.
Tensors changed: Embeddings 0/1, Norms 20/53, MLP 30/78, Attention 40/104
Partial fine-tune vs 1bitLLM/bitnet_b1_58-3B. Partial change: 37% of the weights changed (attention, mlp, norms), in 10 of 26 layers.
Tensors changed: Embeddings 0/1, Norms 20/53, MLP 30/78, Attention 40/104