ModelVerifiedExternal

Gr00t N16 Base Cabinet Pickup Joint 3cam

NVIDIA Isaac GR00T-N1.6-3B fine-tuned on the ManiGuard cabinet-pickup base task (sim Franka Panda). Part of the ManiGuard VLA benchmark - GR00T vs pi0.5 on the same task families with identical data, cameras, and...

Sourcehuggingface.co/IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam/tree/1650c058a88426c6d9321c4ec609b2a170745eb8

KindOtherBase modelnvidia/GR00T-N1.6-3BVersionRevision 1650c05Publisher@IDEAS-Lab-NorthwesternCgrade

Model source

Kind: Other
File size: 4.49–4.65 GB
Base model: nvidia/GR00T-N1.6-3B
Version: vRevision 1650c05
Updated: Oct 25, 2026
Publisher: @IDEAS-Lab-Northwestern
Downloads: 5
Parameters: 3B
Source: Hugging Face

Source description

GR00T-N1.6 - Cabinet-Pickup (joint, 3-cam)

NVIDIA Isaac GR00T-N1.6-3B fine-tuned on the ManiGuard cabinet-pickup base task (sim Franka Panda). Part of the ManiGuard VLA benchmark - GR00T vs pi0.5 on the same task families with identical data, cameras, and controller.

Model

Base: nvidia/GR00T-N1.6-3B - Cosmos-Reason VLM + flow-matching DiT action head
Embodiment: NEW_EMBODIMENT - Franka Panda, 8-D joint state/action (7 arm joints + 1 gripper)
Cameras (3): image_left, image_right, wrist (256x256)
Action: arm = state-relative chunks, gripper = absolute; 16-step horizon; NON_EEF (joint space)
Tuning: GR00T-N1.6 default - VLM (LLM + visual) frozen, train projector + diffusion action head (no LoRA)

Training

1x H100, bf16, global batch 64, 3545 steps (~10 epochs over 22,684 frames), cosine LR (peak 1e-4)
Data: IDEAS-Lab-Northwestern/sim-cabinet-pickup-30-joint-3cam; videos decoded as H.264 for GR00T's torchcodec loader

Usage

Load with Gr00tPolicy from Isaac-GR00T (n1.6-release), --embodiment-tag NEW_EMBODIMENT. The included experiment_cfg/ carries the modality config + normalization stats.

WARNING - Convention (must match at eval): joint-space JointController (absolute joint targets, NON_EEF) + 3 cameras (image_left, image_right, wrist). A mismatched controller or camera set silently feeds an out-of-distribution input.

Sources

1 source

Hugging FacePrimaryVersion 1650c058a88426c6d9321c4ec609b2a170745eb8

huggingface.co/IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam/tree/1650c058a88426c6d9321c4ec609b2a170745eb8

Verified Oct 25

Download files

Model artifacts

2 artifacts

model-00001-of-00002.safetensors

safetensors · 4.65 GB · SHA-256 2dc2707be2aa…b7bd · Hugging Face

Download shard

model-00002-of-00002.safetensors

safetensors · 4.49 GB · SHA-256 555cdb28572f…0197 · Hugging Face

Download shard

Source excerpts

3 excerpts

ReadmeJul 11huggingface.co/IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam

--- license: apache-2.0 base_model: nvidia/GR00T-N1.6-3B pipeline_tag: robotics tags: [robotics, vla, gr00t, gr00t-n1.6, manipulation, maniguard, franka] --- # GR00T-N1.6 - Cabinet-Pickup (joint, 3-cam) NVIDIA Isaac **GR00T-N1.6-3B** fine-tuned on the ManiGuard **cabinet-pickup** base task (sim Franka Panda). Part of the ManiGuard VLA benchmark - GR00T vs pi0.5 on the same task families with identical data, cameras, and controller. ## Model - **Base:** [nvidia/GR00T-N1.6-3B](https://huggingface.co/nvidia/GR00T-N1.6-3B) - Cosmos-Reason VLM + flow-matching DiT action head - **Embodiment:** NEW_EMBODIMENT - Franka Panda, **8-D joint** state/action (7 arm joints + 1 gripper) - **Cameras (3):** image_left, image_right, wrist (256x256) - **Action:** arm = state-relative chunks, gripper = absolute; 16-step horizon; NON_EEF (joint space) - **Tuning:** GR00T-N1.6 default - VLM (LLM + visual) **frozen**, train projector + diffusion action head (**no LoRA**) ## Training - 1x H100, bf16, global batch 64, 3545 steps (~10 epochs over 22,684 frames), cosine LR (peak 1e-4) - Data: [IDEAS-Lab-Northwestern/sim-cabinet-pickup-30-joint-3cam](https://huggingface.co/datasets/IDEAS-Lab-Northwestern/sim-cabinet-pickup-30-joint-3cam); videos decoded as H.264 for GR00T's torchcodec loader ## Usage Load with `Gr00tPolicy` from [Isaac-GR00T (n1.6-release)](https://github.com/NVIDIA/Isaac-GR00T/tree/n1.6-release), `--embodiment-tag NEW_EMBODIMENT`. The included `experiment_cfg/` carries the modality config + normalization stats. > WARNING - Convention (must match at eval): joint-space JointController (absolute joint targets, NON_EEF) + 3 cameras (image_left, image_right, wrist). A mismatched controller or camera set silently feeds an out-of-distribution input.

NoteOct 25huggingface.co/IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam

Source context: 5 downloads · 0 likes · Pipeline robotics · Repo IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam

NoteJul 17huggingface.co/IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam

Source context: 6 downloads · 0 likes · Pipeline robotics · Repo IDEAS-Lab-Northwestern/gr00t-n16-base-cabinet-pickup-joint-3cam