Language-conditioned Panda control in MuJoCo: frozen CLIP, a compact learned policy, audited data, videos, and results.
Samyak Jain
Samisaliveagain
AI & ML interests
Researcher in ML
Recent Activity
updated a collection 11 days ago
LIBERO updated a collection 11 days ago
LIBERO updated a collection 11 days ago
JEPAOrganizations
SO100
SO-100 / SO-101 SmolVLA nut-stacking: datasets and saved policy exports. Original provenance labels are retained.
-
Samisaliveagain/so100_smolvla_rc
SO-101 • Updated • 160 episodes • 80 -
Samisaliveagain/smolvla-so100-nut-stacking-batch16-step-unverified-base-init-export20261011
Robotics • 0.5B • Updated • 32 -
Samisaliveagain/smolvla-so100-nut-stacking-batch16-step-unverified-resumed-export20261011
Robotics • 0.5B • Updated • 28
LIBERO
Language-conditioned Panda control in MuJoCo: frozen CLIP, a compact learned policy, audited data, videos, and results.
JEPA
VLA-JEPA and action-conditioned world-model work: SO-101 demonstrations and the 30,000-step LoRA policy.
SO100
SO-100 / SO-101 SmolVLA nut-stacking: datasets and saved policy exports. Original provenance labels are retained.
-
Samisaliveagain/so100_smolvla_rc
SO-101 • Updated • 160 episodes • 80 -
Samisaliveagain/smolvla-so100-nut-stacking-batch16-step-unverified-base-init-export20261011
Robotics • 0.5B • Updated • 32 -
Samisaliveagain/smolvla-so100-nut-stacking-batch16-step-unverified-resumed-export20261011
Robotics • 0.5B • Updated • 28