mrcuddle/Tiny-DarkLlama-Chat
Same weights, converted vs TinyLlama/TinyLlama-1.1B-Chat-v1.0. Same weights stored in another dtype (BF16 → F32).
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/45, MLP 0/66, Attention 0/88
3 derivatives checked
- mradermacher/Tiny-DarkLlama-Chat-GGUF: Same weights, converted
- mrcuddle/Tiny-DarkLlama: Identical copy
- mrcuddle/Tiny-DarkLlama-SPIN-Implementation: Full fine-tune
41 other repos hold byte-for-byte identical weight files.