---
tags:
- robotics
- humanoid
- imitation-learning
---
# Delta Flow — Humanoid Benchmark

Delta Flow (ImageWAM / HumanWAM whole-body delta-flow policy) checkpoints trained on Humanoid Benchmark tasks (40 epochs; the last two saved checkpoints are published).

## Checkpoints

100 demonstrations per task. Every row is one training run; the epoch folders are its published checkpoints.

| Folder | Task | Robot | Cameras (model input) | Global batch | Proprio augmentation | Checkpoints |
|---|---|---|---|---|---|---|
| `b6-pedal-bin/` | B6 — pick up a chip barrel, step on the pedal bin, drop it in | Unitree G1 (Inspire hands) | ego_view, chest | — | aligned (GR00T A20 recipe) | `epoch-030`, `epoch-040` |

## What an epoch folder contains

Every checkpoint folder (`epoch-XXX/`, or `step-XXXXXX/` for runs trained for a fixed number of steps) is self-contained for evaluation and deployment: weights, model configuration, normalization
statistics and `protocol.json` (task, data, input/output dimensions, cameras, batch, steps, augmentation, file list).
Simple training records are included for reproducibility (training plan, training log, configs, metrics); optimizer /
scheduler / RNG / DeepSpeed resume state is not. `code/` holds the adapter code used for training and inference
(`code/VERSIONS.json`: upstream commits and package versions).

## Inputs and augmentation

Model inputs are identical across all benchmark models: G1 state 44-D (29 body joint angles + 6 + 6 measured hand values
+ 3 projected gravity), H2 state 78-D (75 joint angles + 3 projected gravity), no joint velocities. Cameras: G1 ego view +
chest; H2 head + chest (A14: head + chest + wrist). The third-person camera recorded in the datasets is for data review
only and is never a model input. Each camera is a separate image input, except for Delta Flow and DiT4DiT, which
place the views side by side in one frame.
Training uses the Humanoid Benchmark online data augmentation spec v2 (15 photometric / geometric / sensor / lighting
effects, one of them boosted to 1.5x per sample) with the model's own image augmentation turned off, plus joint-angle
bias and noise on the proprioception (column "Proprio augmentation"; the earlier recipe also delayed the
proprioception and dropped action-history rows).

`legacy/` holds earlier experiments that do NOT follow the benchmark protocol; see `legacy/README.md`.