arcee-ai/Trinity-Nano-Base
Full fine-tune vs arcee-ai/Trinity-Nano-Base-Pre-Anneal. Fine-tune: 100% of the weights changed, across 56 of 56 layers.
Tensors changed: Embeddings 1/1, Norms 337/337, MLP 2381/2381, Attention 280/280, lm_head 1/1
5 derivatives checked
- arcee-ai/Trinity-Nano-Preview: Full fine-tune
- Mungert/Trinity-Nano-Preview-GGUF: Full fine-tune
- mradermacher/Trinity-Nano-Base-i1-GGUF: Quantization
- mradermacher/Trinity-Nano-Base-GGUF: Same weights, converted
- bouncy72/trinity-nano-base-rom: Adapter
2 other repos hold byte-for-byte identical weight files.