prithivMLmods/KAIROS-MM-Qwen2.5-VL-7B-RL
Full fine-tune vs prithivMLmods/KAIROS-MM-Qwen2.5-VL-7B-SFT. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: lm_head 1/1, Embeddings 2/2, Norms 62/122, MLP 277/280, Attention 324/324
3 derivatives checked
- mradermacher/KAIROS-MM-Qwen2.5-VL-7B-RL-i1-GGUF: Quantization (fewer layers)
- prithivMLmods/KAIROS-MM-Qwen2.5-VL-7B-RL-AIO-GGUF: Same weights, converted (fewer layers)
- mradermacher/KAIROS-MM-Qwen2.5-VL-7B-RL-GGUF: Same weights, converted (fewer layers)