sail/Sailor2-20B
Full fine-tune vs Qwen/Qwen2.5-14B. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Norms 97/97, MLP 144/144, Attention 336/336
6 derivatives checked
- mradermacher/Sailor2-20B-GGUF: Quantization
- sail/Sailor2-20B-Chat: Full fine-tune
- sail/Sailor2-20B-Chat-1203: Full fine-tune
- sail/Sailor2-14B: Different architecture
- sail/Sailor2-20B-SFT: Full fine-tune
- sail/Sailor2-L-20B: Full fine-tune
1 other repos hold byte-for-byte identical weight files.