tinyvla / tinyvla_b200

10.3 GB

Ctrl+K

1 contributor

History: 60 commits

AlexWortega

LIBERO offline: TinyVLA v1 (2.5GB) ratio 0.092 vs pi0.5-libero-finetuned (13.8GB) ratio 0.057; GR00T not loadable

603f504 verified 5 days ago

ckpt_dagger_iter0
Upload tinyvla_b200/ckpt_dagger_iter0/README.md with huggingface_hub 5 days ago
ckpt_v4_duckft
microduck fine-tune ckpt (8k steps on duck v2 data), expert keys fixed 6 days ago
ckpt_v4_global20k
fix expert keys (_orig_mod stripped) 6 days ago
ckpt_v4_global60k
fix expert keys (_orig_mod stripped): prior file had a random flow head on load 6 days ago
configs
wds mixture: unitree+nav converters (gnm/vamos), wds source branch in train_fast, AMP autoselect (V100 fp16), embedding expand 16->32, push/fetch/validate tooling, DATA.md 11 days ago
eval_libero
LIBERO offline: TinyVLA v1 (2.5GB) ratio 0.092 vs pi0.5-libero-finetuned (13.8GB) ratio 0.057; GR00T not loadable 5 days ago
eval_zero_shot
duck FT eval (fixed expert): 13/21 better than zero-baseline 6 days ago
patches
tinyvla_b200: B200 streaming training on nvidia/physical-ai (745 samples/s e2e, 7x ViT fast path, hub episode streaming, full instructions) 12 days ago
renders_closedloop
Upload tinyvla_b200/renders_closedloop/cl_vla_stand.mp4 with huggingface_hub 6 days ago
renders_dagger_closedloop
Upload tinyvla_b200/renders_dagger_closedloop/cl_vla_stand.mp4 with huggingface_hub 5 days ago
renders_microduck_sim_20k
microduck sim renders, external MuJoCo camera: GT teacher vs v4@20k side by side, offset-from-STAND action convention 8 days ago
renders_microduck_sim_duckft
sim renders: GT vs duck-FT policy (FIXED expert), external camera 6 days ago
renders_microduck_step10k
microduck zero-shot renders + metrics (v4 ckpt step_10000): all combos worse than zero-baseline, honest negative 9 days ago
scripts
VLA closed-loop driver (fixed head camera + auto-gain): falls <1s all behaviors — BC covariate shift on unstable system 6 days ago
src
fix: HubEpisodeStream.set_epoch was missing — first worker-recycle epoch boundary crashed the run 8 days ago
B200.md
19.2 kB
wds mixture: unitree+nav converters (gnm/vamos), wds source branch in train_fast, AMP autoselect (V100 fp16), embedding expand 16->32, push/fetch/validate tooling, DATA.md 11 days ago
DATA.md
7.39 kB
wds mixture: unitree+nav converters (gnm/vamos), wds source branch in train_fast, AMP autoselect (V100 fp16), embedding expand 16->32, push/fetch/validate tooling, DATA.md 11 days ago
README.md
13.2 kB
tinyvla_b200: B200 streaming training on nvidia/physical-ai (745 samples/s e2e, 7x ViT fast path, hub episode streaming, full instructions) 12 days ago
requirements.txt

1.45 kB
requirements: py3.12, torch/tv pair, lerobot from git tag, triton>=3.7.1 (fla hopper bwd bug) 12 days ago