remyxai/SpaceThinker-Qwen2.5VL-3B
LoRA-style vs UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of q_proj, v_proj changed, in 36 of 36 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, Norms 0/138, MLP 0/304, Attention 72/380
2 derivatives checked
- mradermacher/SpaceThinker-Qwen2.5VL-3B-i1-GGUF: Quantization (fewer layers)
- mradermacher/SpaceThinker-Qwen2.5VL-3B-GGUF: Same weights, converted (fewer layers)