CuriousCat29/l3-stack-mocha-2
Full fine-tune (fewer layers) vs DS-Archive/L3.3-70B-Magnum-v4-SE. DS-Archive/L3.3-70B-Magnum-v4-SE matches part of CuriousCat29/l3-stack-mocha-2 (723 of its 1263 tensors; the rest of CuriousCat29/l3-stack-mocha-2 has no counterpart). On the shared part: Fine-tune: 99% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 120/161, MLP 240/240, Attention 320/320, lm_head 1/1
1 derivatives checked
- mradermacher/l3-stack-mocha-2-i1-GGUF: Quantization