Finetuned on RoboTwin multi-task dataset, RDT2-FM, but vlm was not frozen.

Full parameter finetune.

batchsize=16, 30000steps

Downloads last month
-
Safetensors
Model size
8B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for zaleni/RDT2-RoboTwin-Full

Finetuned
(2)
this model

Dataset used to train zaleni/RDT2-RoboTwin-Full