allenai/Molmo2-4B

Different architecture vs Qwen/Qwen3-4B-Instruct-2507. Different architecture: only 0 of 398/706 tensors match by name and shape.

4 derivatives checked

• allenai/Molmo2-ER: Partial fine-tune
• tollea1234/Molmo2-4B-FP8: Quantization
• reubk/Molmo2-4B-GGUF: Not comparable yet
• mu2solutions/Molmo2-4B-GGUF: Not comparable yet

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

On the Hub · Downloads on Model Pulse