prithivMLmods/Octantis-QwenR1-1.5B
Partial fine-tune vs prithivMLmods/Pisces-QwenR1-1.5B. Partial change: 87% of the weights changed (attention, embeddings, mlp, norms), in 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 30/57, MLP 84/84, Attention 194/196, lm_head 0/1
2 derivatives checked
- mradermacher/Octantis-QwenR1-1.5B-i1-GGUF: Quantization
- mradermacher/Octantis-QwenR1-1.5B-GGUF: Same weights, converted