LaaLM/LaaLM-exp-v1
Full fine-tune vs Qwen/Qwen2.5-3B-Instruct. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 55/73, MLP 108/108, Attention 245/252
3 derivatives checked
- mradermacher/LaaLM-exp-v1-i1-GGUF: Quantization
- LaaLM/LaaLM-exp-v1-GGUF: Same weights, converted
- mradermacher/LaaLM-exp-v1-GGUF: Same weights, converted