Downloads · 30 days
16
22% of all-time downloads
MolmoAct2-BimanualYAM-oranges-ultrawide-12k
ASethi04/MolmoAct2-BimanualYAM-oranges-ultrawide-12k
MolmoAct2-BimanualYAM-oranges-ultrawide-12k is a robotics model from ASethi04. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for lerobot. The card lists the license as apache-2.0.
Fine-tune of allenai/MolmoAct2-BimanualYAM on brandonyang/yam-ultrawide-teleop (75 episodes, 108,726 frames, 30 fps — ultrawide wrist cameras at 312×416, top at 270×480).
All-time downloads
73
Public
Parameters
5.6B
11.5 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors11.5 GB · 100%
Parameter types
How the weights are stored.
BF165.4B · 97%
Base models
Datasets
Model card
From the Hugging Face model README
MolmoAct2-BimanualYAM — "Put all oranges in the bowl" (ultrawide, 3 cameras, 12k steps)
Fine-tune of allenai/MolmoAct2-BimanualYAM on brandonyang/yam-ultrawide-teleop (75 episodes, 108,726 frames, 30 fps — ultrawide wrist cameras at 312×416, top at 270×480).
Uses all three cameras: top + both ultrawide wrist cams. Camera keys: top, left, right (order matters).
Trained with the lerobot PR#4249 fixes (fp32 flow targets, DDP-correct loss normalization,
episode-boundary supervision, llm_residual_dropout=0.1). inference_action_mode=continuous
is baked into the config.
| Steps / epochs | 12,000 / ~7.1 |
| Global batch | 64 (8 GPUs × 8) |
| VLM | LoRA r=64 @ 5e-5; action expert full FT @ 1e-4 |
| Schedule | cosine, 600 warmup, decay ratio 0.1 |
| Augmentation | brightness/contrast/saturation 0.8–1.2, hue ±0.02, sharpness 0.5–1.5, affine ±5°/5% |
| Action space | 14-D absolute joint pose, chunk 30/30 @ 30 Hz |
| Normalization | quantile q01/q99; grippers raw |
| Split / seed | 100/0 / 1000 |
Final training step: step:12K smpl:768K ep:530 epch:7.06 loss:1.130 grdn:1.412 lr:5.0e-06 updt_s:2.355 data_s:0.075 smp/s:26 mem_gb:25.86 discrete_ce_loss:1.127 discrete_z_loss:0.000 action_flow_loss:0.004
Deployment
- Do not pass
norm_tag. - Task string exactly:
Put all oranges in the bowl. --robot.max_gripper_delta=0.05(data needs 0.05/tick).- Camera resolutions must match training (wrists 312×416, top 270×480 if used).
Limitations
Single task, 75 demos, no held-out val. Validate with a no-motion action probe before arming.
- Task
- Robotics
- Library
- lerobot
- License
- apache-2.0
- Created
- Aug 6, 2026
- Updated
- Aug 6, 2026