C-Tianyu/NanoJev
Partial fine-tune vs Qwen/Qwen3-0.6B. Partial change: 74% of the weights changed (attention, mlp, norms), in 28 of 28 layers.
Tensors changed: Embeddings 0/1, Norms 104/113, MLP 84/84, Attention 112/112
2 derivatives checked
- zerodegress/NanoJev-mlx-4bit: Quantization
- zerodegress/NanoJev-bf16: Partial fine-tune