Letemoin/L3-Deluxe-Smurge_l3-12B
Partial fine-tune (fewer layers) vs ehristoforu/llama-3-12b-instruct. ehristoforu/llama-3-12b-instruct matches part of Letemoin/L3-Deluxe-Smurge_l3-12B (453 of its 948 tensors; the rest of Letemoin/L3-Deluxe-Smurge_l3-12B has no counterpart). On the shared part: Partial change: 6% of the weights changed (attention, lm_head, mlp, norms), in 1 of 50 layers.
Tensors changed: Embeddings 0/1, Norms 2/101, MLP 3/150, Attention 4/200, lm_head 1/1
2 derivatives checked
- mradermacher/L3-Deluxe-Smurge_l3-12B-GGUF: Quantization
- mradermacher/L3-Deluxe-Smurge_l3-12B-i1-GGUF: Quantization