allenai/MolmoPoint-8B

Different architecture vs Qwen/Qwen3-8B. Different architecture: only 0 of 399/719 tensors match by name and shape.

2 derivatives checked

• allenai/MolmoPoint-GUI-8B: Full fine-tune
• erdi28/MolmoPoint-8B-bnb-nf4-vision-bf16: Quantization

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse