Lordvarun23/qwen2.5-vl-3b-mllmu-ft
Full fine-tune vs Qwen/Qwen2.5-VL-3B-Instruct. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 2/2, Norms 74/138, MLP 274/304, Attention 376/380
4 derivatives checked
- Lordvarun23/qwen2.5-vl-3b-mllmu-graddiff-forget5: Partial fine-tune
- Lordvarun23/qwen2.5-vl-3b-mllmu-klmin-forget5: Partial fine-tune
- Lordvarun23/qwen2.5-vl-3b-mllmu-npo-forget5: Partial fine-tune
- Lordvarun23/qwen2.5-vl-3b-mllmu-manu-forget5: Partial fine-tune