Umranz/Hydra-Turbo
Partial fine-tune vs datalab-to/lift. Partial change: 89% of the weights changed (attention, embeddings, lm_head, mlp, norms), in 32 of 32 layers.
Tensors changed: lm_head 1/1, Embeddings 3/4, Norms 114/215, Attention 280/332, MLP 181/208
1 derivatives checked
- Umranz/Hydra-Turbo-GGUF: Quantization