unsloth/gemma-3-12b-it
Full fine-tune vs google/gemma-3-12b-pt. Fine-tune: 97% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/4, Norms 290/400, MLP 144/252, Attention 192/408, other 1/1
13 derivatives checked
- mfielding92/Grok-3-gemma3-12B-distilled: Full fine-tune (fewer layers)
- mfielding92/Grok-3-reasoning-gemma3-12B-distilled-GGUF: Full fine-tune (fewer layers)
- mfielding92/gemma3-12B-claude-3.7-sonnet-reasoning-distilled: Full fine-tune (fewer layers)
- ThomasTheMaker/k-12b: LoRA-style
- MayIAskYou/Gemma3-12b-CXL3: LoRA-style
- Lambent/IsoC-Gemma-3-12B: Full fine-tune
- omarabb315/gemma_muharaf_12b_model: Full fine-tune
- bigmoveintelligent/chatter-12b-v1: LoRA-style
- ToastyPigeon/g3-12b-storyteller-v0.2-textonly: LoRA-style (fewer layers)
- jayanthanc/cali-3-12b-oss-120b: LoRA-style
- mergekit-community/mergekit-dare_ties-neobneo: Same weights, repacked
- Hoshino-Yumetsuki/PopiT-Gemma-3-12B-Medical-SFT-1218: LoRA-style
- mergekit-community/mergekit-slerp-xbklfzn: Same weights, repacked
31 other repos hold byte-for-byte identical weight files.