jaspionjader/Kosmos-EVAA-gamma-v14-8B
Partial fine-tune vs jaspionjader/Kosmos-EVAA-gamma-light-8B. Partial change: 90% of the weights changed (attention, embeddings, lm_head, mlp, norms), in 32 of 32 layers.
Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 65/65, MLP 85/96, Attention 113/128
2 derivatives checked
- mradermacher/Kosmos-EVAA-gamma-v14-8B-GGUF: Same weights, converted
- jaspionjader/Kosmos-EVAA-gamma-v17-8B: Full fine-tune