declare-lab/Emma-X
Full fine-tune vs openvla/openvla-7b. Fine-tune: 99% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 7/7, Norms 261/275, MLP 289/304, Attention 324/343, other 50/52, lm_head 1/1
Full fine-tune vs openvla/openvla-7b. Fine-tune: 99% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 7/7, Norms 261/275, MLP 289/304, Attention 324/343, other 50/52, lm_head 1/1