trollek/LittleInstructionMaker-4B-v0.2
Full fine-tune vs trollek/LittleInstructionMaker-4B-v0.1. Fine-tune: 97% of the weights changed, across 24 of 24 layers.
Tensors changed: lm_head 0/1, Embeddings 1/1, Norms 31/49, MLP 72/72, Attention 96/96
4 derivatives checked
- cgus/LittleInstructionMaker-4B-v0.2-iMat-GGUF: Quantization
- mradermacher/LittleInstructionMaker-4B-v0.2-GGUF: Same weights, converted
- trollek/Tyr-4B-v0.1: Full fine-tune
- trollek/ThoughtStream-4B-v0.2: Full fine-tune