robotics-diffusion-transformer/RDT2-VQ
Full fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 2/2, Norms 122/122, MLP 280/280, Attention 324/324, lm_head 1/1
2 derivatives checked
- mradermacher/RDT2-VQ-i1-GGUF: Quantization (fewer layers)
- mradermacher/RDT2-VQ-GGUF: Same weights, converted (fewer layers)
1 other repos hold byte-for-byte identical weight files.