featherless-ai/L3.1-70b-Swallow-Saigetsu
LoRA-style vs NousResearch/Meta-Llama-3.1-70B-Instruct. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, o_proj, up_proj, v_proj changed, in 64 of 80 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/161, MLP 192/240, Attention 128/320, lm_head 0/1
3 derivatives checked
- featherless-ai-quants/KaraKaraWitch-L3.1-70b-Swallow-Saigetsu-GGUF: Quantization
- mradermacher/L3.1-70b-Swallow-Saigetsu-i1-GGUF: Quantization
- mradermacher/L3.1-70b-Swallow-Saigetsu-GGUF: Quantization