Tinman-Lab/Tinman-gemma4-companion-merged
Partial fine-tune vs OBLITERATUS/gemma-4-E4B-it-OBLITERATED. Partial change: 56% of the weights changed (attention, embeddings, mlp, other), in 42 of 42 layers.
Tensors changed: Embeddings 3/6, Norms 0/484, other 110/261, MLP 222/606, Attention 257/719
2 derivatives checked
- Tinman-Lab/Tinman-gemma4-companion-gguf: Same weights, converted (fewer layers)
- Bedhoy/Tinman-gemma4-companion-gguf: Same weights, converted (fewer layers)
2 other repos hold byte-for-byte identical weight files.