Downloads · 30 days
0
cloudwalk-research/gr00t-n17-g1-grab-bottle-rh-417ep-v4-finetune
gr00t-n17-g1-grab-bottle-rh-417ep-v4-finetune is a robotics model from cloudwalk-research. 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 gr00t. The card lists the license as other.
v4 fine-tune of NVIDIA GR00T N1.7-3B (paper) on a Unitree G1 humanoid driven by the SONIC whole-body controller.
Downloads · 30 days
0
Access
Public
Updated Oct 7, 2026
Repo size
20.7 GB
Likes
0
Public
Click a slice to open those files.
.safetensors6.9 GB · 100%
From the Hugging Face model README
v4 fine-tune of NVIDIA GR00T N1.7-3B (paper) on a Unitree G1 humanoid driven by the SONIC whole-body controller.
Single-handed pick task ("grab the bottle", RIGHT hand). Trained at the CloudWalk Robotics Lab (CW-RL), 2026-06, on an aggressively curated 417-episode real-robot teleoperation dataset, for the UNITREE_G1_SONIC embodiment. Released as a reference fine-tune for teams building manipulation policies on the GR00T + SONIC + MuJoCo/G1 stack.
Predecessors: v2 (…-210ep-v2-finetune) — the former production champion (superseded by v10 …-371ep-v10-finetune, 12/12) (checkpoint-20000, grasps from poses where v1 failed) — and v1 (…-105ep-v1-finetune). v4 re-trains the same recipe on a more aggressively curated dataset: per the dataset name (zero-wandering-smooth-radius-5), wandering segments (drift without progress toward the grasp) were removed, trajectories smoothed, and a radius-5 filter applied — yielding 417 short windows.
This is a behavior-cloning fine-tune of the full 3B model. Status: trained 2026-06-19, closed-loop eval complete (6/12) — checkpoint-20000 is published; it did not beat the v2 champion, so v2 remains in production (see Evaluation).
| Base | GR00T-N1.7-3B — Qwen3-VL vision-language backbone + flow-matching diffusion-transformer (DiT) action head |
| Parameters | 3.14 B total / 1.62 B trainable (51.5%) |
| Dataset | cloudwalk-research/gr00t-g1-grab-bottle-right-hand-zero-wandering-smooth-radius-5 — 417 episodes, 48,577 frames @ 50 Hz, 640×480 ego_view camera (no wrist cams); aggressively curated (zero-wandering + smoothing, radius 5 → short windows, ~116 frames/ep) |
| Robot target | Unitree G1 (29-DoF body) + Inspire FTP hands (6 finger joints + 1 constant-zero 7th channel for Dex3-layout compatibility, via InspireFTPGripMapper) + SONIC whole-body controller |
| Embodiment tag | UNITREE_G1_SONIC (unitree_g1_sonic) |
| State space | 43-D (left_leg 6 + right_leg 6 + waist 3 + left_arm 7 + left_hand 7 + right_arm 7 + right_hand 7) |
| Action space | [40 × 78] = 40-step horizon × (64 motion_token + 7 left_hand_joints + 7 right_hand_joints) |
| Hardware | 6× NVIDIA B200 (sm_100 / Blackwell) |
| Mixed precision | bf16 |
| Optimizer | AdamW, lr 1e-4 cosine, warmup_ratio 0.05, weight_decay 1e-5 |
| Steps / batch | 20,000 / global batch 48 (8 per GPU × 6 GPUs); checkpoints saved every 5,000 |
| Epochs | ~19.76 @ 20k (960k frame-views ÷ 48,577 frames); 10k ≈ 9.9, 15k ≈ 14.8 |
| Augmentation | color jitter (brightness 0.3, contrast 0.4, saturation 0.5, hue 0.08) |
| Wall-clock | ~67.5 min (4052.6 s · 6× B200 @ 4.94 it/s) |
| Final train loss | 0.0788 (mean over all 20k steps, warmup included); per-step @20k: 0.0275 |
checkpoint-20000/ # 20k steps (~19.8 ep) — full training run
README.md # this file
Each checkpoint-NNNNN/ is a self-contained, deploy-ready snapshot:
model-00001-of-00002.safetensors + model-00002-of-00002.safetensors (~6.5 GB, bf16)model.safetensors.index.jsonconfig.jsonembodiment_id.json (contains unitree_g1_sonic)processor_config.jsonstatistics.json (dataset normalization stats)experiment_cfg/ (training config snapshot)DeepSpeed ZeRO partition states, optimizer, scheduler and RNG files are intentionally omitted — they are needed only to resume training and would add several GB per checkpoint.
Why checkpoint-20000. The v4 dataset is smaller than v2 (48,577 vs 62,772 frames), so 20,000 steps is ≈19.8 epochs vs v2's validated 15.3. Checkpoints were saved every 5k during training (5k/10k/15k/20k), but only checkpoint-20000 (≈19.8 ep) is published; the production checkpoint is chosen by closed-loop comparison, not by training loss. There is no held-out split; train loss is a fit probe, not a generalization measure.
Closed-loop evaluation — completed (real G1). checkpoint-20000 was run closed-loop on the physical G1 + SONIC stack and measured against the v2 checkpoint-20000 champion. Low BC loss is a healthy prerequisite, not evidence of a good policy — this closed-loop result is the load-bearing signal.
Procedure. The bottle was placed at 12 fixed positions spanning one half-side of the table (the table is fixed relative to the robot). These are the same 12 positions used for data collection (see the source dataset card). One trial per position (12 trials) gives the grid result. Legend: 🟥 red = knocked the bottle over · 🟧 orange = stuck in an "indecision" loop, never reaching the bottle · 🟩 green = successfully grasped the bottle. The policy was invoked with the task prompt "grab the bottle" — the same prompt used to collect the data.

Results — checkpoint-20000.
Grid (12 fixed positions, 1 trial each):
| Outcome | Count | Rate |
|---|---|---|
| 🟩 Success | 6/12 | 50% |
| 🟧 Non-convergence (indecision loop) | 2/12 | 17% |
| 🟥 Error (knocked bottle over) | 4/12 | 33% |
⚠️ Single-trial methodology. Each of the 12 grid positions is tested once (1 trial/position); repeated placements at the same spot can succeed or fail differently — the fixed-position repeat test, where run, shows this per-position variance. Treat the grid score as a single-pass signal, not a repeated-trial mean.
Conclusion. The most-aggressive curation did not help. v4 has the highest knock-over rate in the family (4/12, 33%) and middling success (6/12, 50%). Radius-5 zero-wandering cuts episodes into the shortest windows, which the lineage already flagged as overfit-prone (~19.8 epochs) — the closed-loop result bears that out. v4 does not beat the v2 champion; v2 remains in production.
from huggingface_hub import snapshot_download
local = snapshot_download(
repo_id="cloudwalk-research/gr00t-n17-g1-grab-bottle-rh-417ep-v4-finetune",
repo_type="model",
allow_patterns=["checkpoint-20000/*"],
)
print(local)
Start the GR00T policy server (from an Isaac-GR00T environment) pointing at the downloaded checkpoint:
python -m gr00t.eval.run_gr00t_server \
--model-path <local>/checkpoint-20000 \
--embodiment-tag UNITREE_G1_SONIC \
--device cuda:0 --host 0.0.0.0 --port 5550
Closed-loop control of the G1 (sim or real) is driven by the SONIC whole-body controller in GR00T-WholeBodyControl: the policy emits motion_token + hand-joint targets that the SONIC WBC decodes into whole-body joint commands. The server must be launched with the same UNITREE_G1_SONIC embodiment tag used in training. See the NVlabs VLA inference tutorial. This checkpoint is not plug-and-play on hardware — it requires the SONIC C++ deploy stack and the matching G1 setup.
ego_view (head) camera was recorded — no wrist cameras. The policy has never seen wrist views.motion_token + hand joints for UNITREE_G1_SONIC; it only produces robot motion through the SONIC WBC + ZMQ deploy pipeline. It is not directly executable on a bare G1.NCCL_IB_DISABLE=1 NCCL_P2P_LEVEL=NVL (InfiniBand off, P2P over NVLink) and W&B in offline mode (synced post-run). These affect only the training run, not the weights.| Version | Dataset | Episodes | Frames | Epochs @20k | Closed-loop | Notes |
|---|---|---|---|---|---|---|
| v1 | 105ep-v1 | 105 | 70,680 | 13.6 | ✅ 4/12 (33%) | First GR00T N1.7 + SONIC fine-tune at CW-RL; right-hand bottle pick. |
| v2 | right-hand-v2 | 210 | 62,772 | 15.3 | ✅ 10/12 (83%) — former champion | Curated (episodes split into shorter windows, bad segments removed). checkpoint-20000 = former production champion (10/12), superseded by v10 (12/12). |
| v4 (this) | radius-5 | 417 | 48,577 | 19.8 | ✅ 6/12 (50%) | Zero-wandering, most aggressive curation (radius 5 → short windows). |
| v5 | radius-20 | 314 | 50,496 | 19.0 | ✅ 9/12 (75%) | Zero-wandering, least aggressive curation (radius 20 → longer windows, ~161 f/ep). |
| v6 | radius-20-merged | 502 | 120,017 | 8.0 | ✅ 8/12 (67%) | Merged (105-ep + 115-ep "worst-positions"), radius-20 with grasp-frame preservation. Largest set → 20k = only ~8 ep (underfit risk). |
| v7 | speedup-3mm-v1 | 220 | 60,163 | ~16.0 | 7/12 (58%) | DP speedup resampling (wrist-Cartesian, 3 mm/frame, max_K=40) of the merged set; no segment removal — reference baseline. ≈16 ep @20k lands on v2's sweet spot. |
cloudwalk-research/gr00t-g1-grab-bottle-right-hand-zero-wandering-smooth-radius-5 (CloudWalk Research, 2026), PICO 4 Ultra teleoperation on the Unitree G1 with SONIC WBC; aggressive re-curation of the right-hand bottle data.Developed by cloudwalk-research in the CloudWalk Robotics Lab (CW-RL). Fine-tuned from nvidia/GR00T-N1.7-3B using Isaac-GR00T; targets the Unitree G1 with the SONIC whole-body controller.
@misc{cwrl_gr00t_grab_bottle_v4_2026,
title = {GR00T N1.7 Fine-Tune v4 --- Unitree G1 "grab the bottle" (right hand, SONIC WBC, zero-wandering curated 417-ep dataset)},
author = {{CloudWalk Robotics Lab}},
year = {2026},
howpublished = {Hugging Face model repository},
url = {https://huggingface.co/cloudwalk-research/gr00t-n17-g1-grab-bottle-rh-417ep-v4-finetune}
}
Inherits the NVIDIA Open Model License Agreement of the base model nvidia/GR00T-N1.7-3B — see the license terms. This is a research preview: not intended for safety-critical use; closed-loop deployment on a physical humanoid requires human oversight.