BytedTsinghua-SIA/Open-MOPD-SmolLM3-3B-MixSFT
Full fine-tune vs HuggingFaceTB/SmolLM3-3B-Base. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 11/73, MLP 108/108, Attention 144/144
9 derivatives checked
- BytedTsinghua-SIA/Open-MOPD-SmolLM3-3B-RL-Math: Partial fine-tune
- BytedTsinghua-SIA/Open-MOPD-SmolLM3-3B-RL-Code: Partial fine-tune
- Huangsssz/SmolLM3-3B-MOPD-PG-SGD-lr2p5e-3-NoThinking-500: Partial fine-tune
- Huangsssz/SmolLM3-3B-MOPD-Top64-DT-NoThinking-500: Full fine-tune
- Huangsssz/SmolLM3-3B-MOPD-Top64-DR-NoThinking-500: Full fine-tune
- Huangsssz/SmolLM3-3B-MOPD-Top64-GT-NoThinking-500: Full fine-tune
- open-athena/MarinSkyRL-Open-MOPD-SmolLM3-3B-step-32: Full fine-tune
- BytedTsinghua-SIA/Open-MOPD-SmolLM3-3B-RL-IF: Partial fine-tune
- BytedTsinghua-SIA/Open-MOPD-SmolLM3-3B-Final: Partial fine-tune