General humanoid motion tracking in minutes on one GPU. PartyMimic trains one whole-body tracking policy for a 41 h motion library on a single RTX 4090 and tracks motions it has never seen, on two humanoids (Unitree G1 and RoboParty RP1). This repository is the method: the training code, the data recipe built only from public datasets, the RP1 retargeting, the diversity analysis and the evaluation. Two trained checkpoints are included.
- Fast: 26 min of training on 1× RTX 4090 (24 GB).
- Diverse data: 41 h from LAFAN1, a diversity-selected 10k subset of BONES-SEED and KungfuAthlete.
- Generalizes: zero-shot on five held-out test sets (Loco, Mani, Dance, Ground, Extreme martial arts).
- Precise: 20.5 mm on the held-out manipulation clips (G1), the lowest error of every method we compared.
- Two architectures: an MLP reference encoder (default) or a cross-attention encoder (
--ref_enc xattn). - Two modalities: joint-angle references (default) or the pelvis and five keypoints (
--interface sparse5).
PartyMimic is a model-based learner in the SuperMimic family. It fits an online MLP world model, trains the policy
by backpropagating through 32 imagined steps of that model, and uses a critic beyond the horizon. Project page:
roboparty.github.io/PartyMimic (docs/index.html),
with videos, the data-diversity analysis, the recipe ablation, the SuperMimic full-factor ablation and a
sim2real-oriented evaluation.
Body position error (mm, lower is better), scored by the sim2real leaderboard sim2sim evaluator (MuJoCo, every clip from its first frame). Loco, Mani, Dance and Ground are the MotionDecode splits; Extreme is the held-out 30% of KungfuAthlete (whole source videos held out). Mean is the mean of the five.
| G1 | Training | Loco | Mani | Dance | Ground | Extreme | Mean |
|---|---|---|---|---|---|---|---|
| SONIC (official) | 21,000 GPU-h | 28.7 | 45.5 | 48.7 | 80.3 | 88.7 | 58.4 |
| Mimic Lite v1.1 (official) | 50 GPU-h | 25.0 | 28.2 | 43.3 | 145.2 | 95.7 | 67.5 |
| SuperMimic | 52 min | 34.9 | 34.3 | 55.5 | 119.2 | 95.5 | 67.9 |
| NablaMimic | 120 min | 32.9 | 30.3 | 46.9 | 107.8 | 81.6 | 59.9 |
| BlinkMimic | 59 min | 32.9 | 30.9 | 54.1 | 100.1 | 81.4 | 59.9 |
| XMimic | 50 min | 32.3 | 30.6 | 53.5 | 101.3 | 83.1 | 60.2 |
| PartyMimic (MLP) | 26 min | 25.7 | 20.5 | 36.8 | 80.6 | 62.0 | 45.1 |
| PartyMimic (attention) | 41 min | 25.1 | 21.1 | 35.0 | 76.2 | 60.3 | 43.5 |
| RP1 | Training | Loco | Mani | Dance | Ground | Extreme | Mean |
|---|---|---|---|---|---|---|---|
| SuperMimic | 149 min | 34.8 | 29.1 | 53.4 | 162.5 | 90.2 | 74.0 |
| BlinkMimic | 68 min | 33.7 | 27.3 | 50.3 | 110.5 | 80.3 | 60.4 |
| XMimic | 59 min | 33.8 | 29.1 | 49.7 | 121.3 | 79.5 | 62.7 |
| PartyMimic (MLP) | 26 min | 23.4 | 16.5 | 35.3 | 105.1 | 58.9 | 47.8 |
| PartyMimic (attention) | 42 min | 22.9 | 16.1 | 33.4 | 94.3 | 56.2 | 44.6 |
| PartyMimic (sparse5, MLP) | 29 min | 24.7 | 19.1 | 38.3 | 100.9 | 60.7 | 48.7 |
SuperMimic (paper hyperparameters), NablaMimic (official config), BlinkMimic (main preset), XMimic (its defaults) and PartyMimic were all trained by us on the same library (G1: LAFAN1 + BONES-SEED subset + KungfuAthlete train split; RP1: the same clips retargeted to RP1), each run on one RTX 4090; 2-seed means. Time is training wall-clock. SONIC and Mimic Lite are official leaderboard checkpoints with other data and hardware. Completion rates and the full protocol are on the project page.
Two checkpoints, the default recipe (MLP encoder, joint references), seed 0, 12,000 iterations:
| File | Robot | Training | Loco | Mani | Dance | Ground | Extreme | Mean |
|---|---|---|---|---|---|---|---|---|
checkpoints/g1_mlp.pt |
G1 (--robot g1mj) |
27 min | 25.5 | 20.6 | 36.7 | 82.6 | 61.9 | 45.5 |
checkpoints/rp1_mlp.pt |
RP1 (--robot rp1) |
29 min | 23.3 | 16.5 | 35.5 | 102.3 | 59.2 | 47.4 |
Each holds the policy, world model and critic (no optimizer state). Export for the sim2real controller (sim2sim and real robot), score, or fine-tune:
mkdir -p deploy && uv run python tools/export_deploy_onnx.py checkpoints/g1_mlp.pt deploy/g1_mlp.onnx # + .yaml
SIM2REAL=<patched sim2real checkout> bash scripts/eval.sh checkpoints/g1_mlp.pt eval/g1_mlp g1 # see Evaluate
bash scripts/train.sh logs/g1_ft --robot g1mj --motion_dir <a few clips> --init checkpoints/g1_mlp.pt --iterations 2000The export checks the ONNX against the PyTorch policy on a few training clips (at least 4). Without the training
library it uses the LAFAN1 clips for G1; for RP1 point MOTION_DIR at a directory of RP1 motion npz clips.
Linux x86_64, an NVIDIA GPU with a CUDA 12.8 driver, and uv.
git clone git@github.com:Roboparty/PartyMimic.git partymimic && cd partymimic
uv sync # mjlab, MuJoCo Warp, torch 2.9.0 (CUDA 12.8)
tools/download_data.sh # G1 description + 40 LAFAN1 G1 clips, RP1 meshes + MJCF, G1 MJCF of the diversity analysisdownload_data.sh gets the RP1 meshes from Roboparty/rp1_description
(pinned commit) and copies the capsule-collision MJCF of assets/rp1_28dof_mjlab next to
them (data/rp1_28dof_mjlab/), which is the model PartyMimic and the evaluator simulate.
Every clip comes from a public dataset:
| Dataset | Used for | Source | License |
|---|---|---|---|
| LAFAN1, G1 retarget | training (40 clips) | lvhaidong/LAFAN1_Retargeting_Dataset (LAFAN1) | CC BY-NC-ND 4.0 |
| BONES-SEED, G1 retarget | training (10,000 clips) | bones-studio/seed (gated) | BONES-SEED license |
| KungfuAthlete, G1 | training (635 clips), Extreme test (308 clips) | LuluCao/KungfuAthleteBot, NPCLEI/KungFuAthleteBot | Apache-2.0 |
| MotionDecode | Loco / Mani / Dance / Ground tests (228 clips) | CMRobot/MotionDecode, as packaged by sim2real | non-commercial access terms |
G1 library. tools/data/build_library.sh g1 builds it from the sources (23.5 GB BONES-SEED download; accept
its terms on Hugging Face and log in with hf auth login or set HF_TOKEN):
- LAFAN1: the 40 csv clips of
download_data.shto motion npz (tools/data/prep_motions.py --format lafan). - BONES-SEED: the 10,000 clips of
manifests/bones_seed_div10k.txt, 120 Hz csv to 50 Hz npz. The list was selected bytools/data/bones_subset.py(below). - KungfuAthlete:
tools/data/kungfu.py qascores every smoothed G1 clip (floor offset, ground and self penetration, joint limits, speeds);writedrops the failing clips, corrects the floor and writes the train split and the Extreme test set ofmanifests/kungfu_split.json(a 30% test share per subset and style, drawn by whole source videos, so no test clip shares a video with a training clip). - All clips are linked into
data/motions/g1_mix; the Extreme set isdata/testsets/extreme_g1.
The selection of the BONES-SEED subset can be rerun (feat needs BONES-SEED's metadata/seed_metadata_v004.parquet,
select a GPU) and gives the same list:
uv run --with pyarrow python tools/data/bones_subset.py feat data/raw/bones data/raw/bones/feat.npz --meta <seed_metadata_v004.parquet>
uv run python tools/data/bones_subset.py select data/raw/bones/feat.npz div10k.txt --n 10000It clusters 1-s windows of all non-mirror clips (without clips more than 3 cm below the floor or faster than 25 rad/s at a joint) and greedily picks clips that fill the clusters evenly, every content once before repeats.
RP1 library. The same clips, retargeted G1 → RP1 with the robot-to-robot IK of
human-humanoid-tools (commit ed8b350; uv sync --extra all
in its checkout):
HHT=<human-humanoid-tools checkout> P=8 GPUS="0 1" tools/data/build_library.sh rp1It writes the G1 clips as csv (tools/data/rp1_data.py g1csv), retargets them (tools/data/rp1_retarget.sh: profile
mimic, Newton backend, 24 IK iterations, 50 Hz, the zero-pose calibration and the robot presets of
assets/hhtools_robots), converts the RP1 csv to npz (data/motions/rp1_mix) and moves the
KungfuAthlete clips to the G1 floor height (tools/data/rp1_floor.py; R2R leaves some jump clips 2–5 cm above the
ground). The Extreme set is retargeted the same way (data/testsets/extreme_rp1). R2R takes about 8 s per clip per
process; P processes run in parallel over the GPUs in GPUS. The RP1 copy of the MotionDecode tests:
HHT=... SIM2REAL=... tools/data/build_library.sh rp1_md (into $SIM2REAL/datasets/motiondecode_rp1).
Motion format. --motion_dir takes a directory of .npz clips, each with fps (50), joint_pos [T, J] in
the robot's joint order (src/partymimic/<robot>/), and the pelvis pose as the first body of body_pos_w
[T, B, 3] and body_quat_w [T, B, 4] (wxyz). PartyMimic computes the other bodies by forward kinematics.
The defaults of partymimic.train are the PartyMimic recipe:
bash scripts/train.sh logs/g1_s0 --robot g1mj --motion_dir data/motions/g1_mix --seed 0 # 12k iterations, ~26 min
bash scripts/train.sh logs/rp1_s0 --robot rp1 --motion_dir data/motions/rp1_mix --seed 0
# two GPUs: collection + world model on cuda:0, policy + critic on cuda:1, overlapped (1.6x the iterations per second)
bash scripts/train.sh logs/g1_p0 --robot g1mj --motion_dir data/motions/g1_mix --seed 0 --pipelineThe MLP recipe fits a standard 24 GB RTX 4090: with only 24.1 GB free, G1 and RP1 runs kept their speed and peaked
at 22.6 and 23.4 GB. Checkpoints are written every 2,000 iterations (ckpt_002000.pt, ...) and last.pt.
Options.
- Defaults include
--muon_wd 0.1and--std_scale 0.6(see The recipe). The shipped checkpoints were trained withstd_scale 0.35and no Muon weight decay. --ref_enc xattn: the cross-attention reference encoder. Lower error on Dance, Ground and Extreme (tables above) for 1.6x the training time. It needs more than 24 GB; on a 24 GB card add--pi_batch 1024(G1: 29 min).--interface sparse5: the policy gets only the pelvis, head, wrist and ankle poses of the reference, no joint angles (one command format for a person, a VR tracker set or another robot). A keypoint-to-joint network is fit to the library before training and restores the reference-action skip (--kp_skip), with an auxiliary joint-angle loss (--kp_aux 1.0). Export withtools/export_deploy_onnx.py(inputref_kp).--init <ckpt>: fine-tune a trained policy on a few clips, e.g. one hard clip before it is deployed. Three held-out G1 ground transitions went from 120–415 mm to 24–36 mm in 2k iterations. It does not help on short, highly dynamic clips.
uv run python -m partymimic.train -h lists the remaining options.
# sim2real leaderboard protocol (ONNX + deploy YAML, MuJoCo CPU sim2sim on the four MotionDecode splits)
SIM2REAL=<checkout of github.com/EGalahad/sim2real> bash scripts/eval.sh logs/g1_s0/ckpt_012000.pt eval/g1_s0 g1
# the Extreme set (any any4hdmi dataset directory)
SET=data/testsets/extreme_g1 SIM2REAL=... bash scripts/eval.sh logs/g1_s0/ckpt_012000.pt eval/g1_s0 g1
# side-by-side videos of the median clip per split: robot solid, reference ghost, both at global poses
MUJOCO_GL=egl python tools/sim2sim_video.py eval/g1_s0 g1 <robot MJCF> videos --sim2real <sim2real checkout>Apply tools/sim2real.patch to the sim2real checkout first (git apply, tested on its
commits 287ac15 and 4a777c8). It adds the observation terms the PartyMimic deploy YAML reads
(supermimic.*: proprioceptive history, reference joints and root, root orientation history) and, for RP1, the
robot config (limits, armatures, standing pose, joint friction) and RP1's tracked bodies in the metrics; G1 scoring
is unchanged. scripts/eval.sh ... rp1 points it at data/rp1_28dof_mjlab and the RP1 test sets.
Also included: tools/eval_batch.py and tools/eval_metrics.py for fast in-simulator evaluation on the training
clips, and tools/export_onnx.py / tools/play_onnx.py for a single-motion ONNX in the BeyondMimic layout.
tools/diversity measures how much of a library is distinct motion and how new each test set is
(page). Each 1-s window is described by heading-invariant root, joint and hand / foot
kinematics, z-scored and projected to 32 PCA dimensions, and the library is clustered into 2,000 motion clusters
(datacard/). Per source it reports clips, hours and the effective number of motion types (exponential of the
cluster entropy); for the cumulative mixes, the share of each test set's windows that lie within the median
cross-source distance of a training window.
uv run --with scikit-learn python tools/diversity/analyze.py work/div out/div \
--source lafan:LAFAN1=data/motions/g1/lafan --source bones:BONES-SEED=data/motions/g1/bones \
--source kungfu:KungfuAthlete=data/motions/g1/kungfu \
--md <sim2real>/datasets/motiondecode --test Extreme=data/testsets/extreme_g1
uv run --with scikit-learn python tools/diversity/figure.py work/div out/div out/div/data.pngXMimic's protocol on our RTX 4090s: all runs train on its 40 LAFAN1 clips (SHA-256 checked) and are scored on the same clips, 4 replicas every 100 iterations, 3 seeds. Minutes to BlinkMimic's target; compilation excluded by XMimic's rule, added back in the start-up column.
| Method | GPUs | it/s | Return ≥ 0.82 | + 0 failed, held ×3 | Held, with start-up | Return at 3,500 it |
|---|---|---|---|---|---|---|
| BlinkMimic (main) | 1 | 7.8 | 2.93 | 3.45 | 3.62 | 0.865 |
| XMimic | 1 | 9.3 | 2.32 | 3.71 | 3.96 | 0.860 |
| PartyMimic | 1 | 8.2 | 2.45 | 5.17 | 5.57 | 0.883 |
| XMimic (actor | learner) | 2 | 13.2 | 1.71 | 2.89 | 3.17 | 0.863 |
PartyMimic (--pipeline) |
2 | 12.7 | 1.66 | 3.41 | 3.82 | 0.883 |
PartyMimic reaches return 0.82 in the fewest iterations and ends highest. It is slower to zero failures: one or two hard training clips keep failing for a while; BlinkMimic and XMimic search for those clips with evaluation probes, which PartyMimic dropped because they did not improve held-out tracking.
What matters most, measured by the change in G1 held-out Loc/Man error when a factor is removed (ablations on an earlier 9,173-clip library; the page explains each factor):
| Factor | Setting | Effect when removed |
|---|---|---|
| Muon for policy and world model | --pi_opt muon --wm_opt muon |
Adam instead: +10.7 mm |
| Muon weight decay | --muon_wd 0.1 |
five-set mean +2.4 mm |
| Tracking-reward kernels | --std_scale 0.6 |
1.0: +5.5 mm; 0.35: more slip / jerk |
| 32-step imagination | --horizon 32 |
16 steps: +5.9 mm |
| Reference-action skip | --pi_ref_skip |
+5.6 mm |
| Policy EMA | --pi_ema 0.995 |
+3.4 mm |
| Evaluator's collision model | --robot g1mj |
LAFAN falls 14/40 instead of 3/40 |
| Value-gradient consistency | --v_gcons 0.5 |
+1.0 mm |
| Diversity-balanced resets | --div_k 24 |
Ground +6–19 mm |
| Self-imitation | --sil 1 |
+0.5 mm early in training |
Of the official SuperMimic design choices ported onto this recipe, only --muon_wd 0.1 clearly helps tracking.
Widening the kernels from 0.35 to 0.6 keeps the five-set mean within noise and reduces foot slip under soft contact
and other unseen perturbations (page: Sim2real). The shipped checkpoints were trained
with std_scale 0.35; retrain for the new defaults.
PartyMimic is derived from BlinkMimic, which builds on supermimic_vibed and InstantTrack. All three are MIT-licensed; their notices are in THIRD_PARTY_NOTICES.md. Methods: SuperMimic (Liao et al., arXiv:2610.28480, 2026), SHAC (Xu et al., 2022), BeyondMimic (Liao et al., 2025) and Muon (Jordan et al., 2024). Simulation: mjlab and MuJoCo Warp. Retargeting: human-humanoid-tools (Apache-2.0). Robots: Unitree G1 (unitree_mujoco, BSD-3-Clause) and RoboParty RP1 (rp1_description).
Code: MIT (LICENSE). The checkpoints, videos and built motion libraries are derived from the motion datasets above and follow their licenses (LAFAN1: CC BY-NC-ND 4.0; BONES-SEED: its own license; MotionDecode: non-commercial terms); they are not covered by the MIT License.