Deprecated / not recommended for new deployments.
This checkpoint remains available for reproducibility of earlier experiments.
Use Cosmos3-Edge-Policy-DROID-GenW8A8 instead.
This is a historical Cosmos Lite W4-GenW8 artifact for the Edge DROID policy. It is not a recommended new deployment target.
At A Glance
Quant.
Denoise
Guidance
Peak VRAM
Request latency
Banana SR
W4-GenW8
2
3.0
8.79 GB
570 ms p50
80% (40/50)
The preview above is a historical RoboLab Banana rollout associated with this artifact.
Model Overview
nvidia/Cosmos3-Edge-Policy-DROID is a 4B Cosmos3
Mixture-of-Transformers policy that predicts DROID robot actions from language,
RGB observations, and robot state. Cosmos Lite changes its inference
representation and runtime; it does not retrain or distill the policy.
This historical gen_branch_w8 artifact uses W4-GenW8 across 336 (168 / 168 W4 / W8). Activations
and non-targeted weights remain BF16. The bundle is self-contained and does not
require the BF16 source checkpoint at deployment time.
Calibration And Evaluation
No additional policy training is performed. W4 scales use activation
statistics from 128 successful episodes in the official nvidia/Cosmos3-DROID
training split. Calibration does not update model parameters and uses no
RoboLab evaluation episode. Closed-loop quality in this card is the historical
50-episode Banana comparison.
Policy Interface
The validated RoboLab client composes three camera views into a 640x540 RGB
input and maps it to the model's 736x544 inference bucket. It returns a 32x8
DROID joint-position action chunk; the standard integration executes all 32
actions before requesting another chunk.
Model precision is fixed by this artifact. Guidance, denoise steps, and shift
are independent runtime controls. The release config defaults to guidance 3,
two denoise steps, and shift 5.
Benchmark Results
This artifact is retained for reproducibility of the earlier gen_branch_w8 experiment. The historical Banana result above is a 50-episode paired task-local comparison, not a substitute for the RoboLab-120 release benchmark. See the legacy ablation tables for the complete comparison and the main benchmark for recommended artifacts.
The artifact uses the Cosmos Lite packed weight layout and must be loaded
through the Cosmos Lite runtime.
Quickstart
Install the Hugging Face CLI, clone the runtime, and download this self-contained bundle:
This is a Cosmos Lite deployment bundle, not a generic GPTQ or AWQ checkpoint.
The packed weights, precision map, residual tensors, processor assets, and
runtime manifest must be loaded through Cosmos Lite. Weight-only artifacts keep
activations in BF16; GenW8A8 artifacts use FP8 W8A8 only in the generation
branch and retain the documented BF16 paths elsewhere.
The default server binds to 127.0.0.1:8000 and has no built-in TLS or
authentication. Do not expose it directly to an untrusted network.
Bundle contents include runtime configuration, source revisions, file sizes,
and SHA256 hashes.
Limitations And Safety
Simulation results do not establish transfer to a new robot, camera layout,
task distribution, or action contract. Validate those changes with paired
closed-loop rollouts. This policy is not real-robot safety certified. Real
deployment requires an independent E-stop, watchdog, motion limits,
stale-command rejection, and operator supervision.
This community quantization is not an NVIDIA product.
License
The upstream model and this derivative are distributed under
OpenMDW-1.1. Review the upstream model card
and repository notices before use.
