Pinkstackorg/Fijik1-3b-instruct
Full fine-tune (fewer layers) vs Qwen/Qwen2.5-1.5B-Instruct. Qwen/Qwen2.5-1.5B-Instruct matches part of Pinkstackorg/Fijik1-3b-instruct (337 of its 674 tensors; the rest of Pinkstackorg/Fijik1-3b-instruct has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Norms 14/57, MLP 84/84, Attention 133/196
2 derivatives checked
- mradermacher/Fijik-3b-Instruct-i1-GGUF: Quantization
- mradermacher/Fijik-3b-Instruct-GGUF: Same weights, converted