Pinkstack/Superthoughts-lite-v1
LoRA-style vs HuggingFaceTB/SmolLM2-1.7B-Instruct. 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 24 of 24 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/49, MLP 72/72, Attention 96/96
3 derivatives checked
- mradermacher/Superthoughts-lite-v1-i1-GGUF: Quantization
- mradermacher/Superthoughts-lite-v1-GGUF: Same weights, converted
- Pinkstack/Superthoughts-lite-v1-GGUF: Same weights, converted