ahmedheakl/asm2asm-deepseek-1.3b-400k

Full fine-tune vs deepseek-ai/deepseek-coder-1.3b-instruct. Fine-tune: 95% of the weights changed, across 24 of 24 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 31/49, MLP 72/72, Attention 96/96

1 derivatives checked

• mradermacher/asm2asm-deepseek-1.3b-400k-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse