trl-lib/qwen1.5-1.8b-sft
Full fine-tune vs Qwen/Qwen1.5-1.8B. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Norms 10/49, MLP 72/72, Attention 168/168
1 derivatives checked
- tensorblock/qwen1.5-1.8b-sft-GGUF: Quantization