Downloads · 30 days
3
20% of all-time downloads
jren313/starvla_gr00t_n1d7_pipette_3view_nohist_plus21p12_hil46_ft14
starvla_gr00t_n1d7_pipette_3view_nohist_plus21p12_hil46_ft14 is a machine learning model from jren313. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
The warm-start arm of the round-15 A/B: round 14's served checkpoint (steps15000) fine-tuned on the same merged teleop+HIL dataset that round 15 trained on from scratch.
Downloads · 30 days
3
20% of all-time downloads
All-time downloads
15
Public
Repo size
9.5 GB
Likes
0
Public
Click a slice to open those files.
.pt9.5 GB · 100%
From the Hugging Face model README
The warm-start arm of the round-15 A/B: round 14's served checkpoint
(steps_15000) fine-tuned on the same merged teleop+HIL dataset that
round 15 trained on from scratch.
steps_15000_pytorch_model.pt;
fresh optimizer, step 0 (is_resume false).jren313/g1-pipette-3view-plus21p12-hil46
(124 eps / 112,593 frames), same union q99 stats (byte-identical
dataset_statistics.json), uniform frame sampling, no HIL upweighting.checkpoints/steps_3000_pytorch_model.pt (final; best by
in-sample mse 0.000481 — directly comparable to round 15's 0.000507:
same data, same stats; ~1/10 the compute).config.yaml +
dataset_statistics.json sit two directories above the .pt. Views
[rgb, wrist_left, wrist_right].