qingy2024/GRMR-V3-Q1.7B

Full fine-tune vs unsloth/Qwen3-1.7B-Base. Fine-tune: 100% of the weights changed, across 28 of 28 layers.

Tensors changed: Norms 99/113, MLP 84/84, Attention 112/112

1 derivatives checked

• qingy2024/GRMR-V3-Q1.7B-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse