qikp/kite-1.5-1m
LoRA-style vs hmellor/tiny-random-LlamaForCausalLM. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 2 of 2 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/5, MLP 6/6, Attention 8/8