fxmeng/DSV2-Lite-DSA-Sparse-1250
Full fine-tune vs deepseek-ai/DeepSeek-V2-Lite. Fine-tune: 100% of the weights changed, across 27 of 27 layers.
Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 81/82, MLP 2673/2673, Attention 108/108
Full fine-tune vs deepseek-ai/DeepSeek-V2-Lite. Fine-tune: 100% of the weights changed, across 27 of 27 layers.
Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 81/82, MLP 2673/2673, Attention 108/108