Jongbin-kr/exaone-3.5-7.8b-verireason-full-ft-reasoning-co-ratio1.0-epoch3
Partial fine-tune vs LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct. Partial change: 89% of the weights changed (attention, mlp, norms), in 32 of 32 layers.
Tensors changed: Attention 128/128, Norms 24/65, MLP 96/96, Embeddings 0/1, lm_head 0/1