maple-research-lab/RemeDi-RL

Full fine-tune vs GSAI-ML/LLaDA-8B-Instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Attention 128/128, Norms 54/65, Embeddings 1/1, other 65/65, MLP 32/32

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

On the Hub · Downloads on Model Pulse