codefuse-ai/F2LLM-v2-160M
Full fine-tune vs codefuse-ai/F2LLM-v2-0.6B-Preview-Pruned-160M. Fine-tune: 100% of the weights changed, across 9 of 9 layers.
Tensors changed: Embeddings 1/1, Norms 37/37, MLP 27/27, Attention 36/36
2 derivatives checked
- mradermacher/F2LLM-v2-160M-GGUF: Same weights, converted
- cstr/f2llm-v2-160m-GGUF: Not comparable yet