tartanimu-public3072-tomjerry

Tom&Jerry: credited V25b continuation

Reproducible single-model inference for TartanIMU 2026 submission 56333155, Kaggle account tomjerry101, public score 0.42731 (lower is better). This entry preserves the participant's previously verified predictions following an organizer-approved account migration reported by the participant. It is not a claim to top rank or a verified private score.

The model derives from DBX / leaonwang's public V25b code and checkpoint release. The upstream architecture and weights are credited to their publisher. Code retains Apache-2.0 and weights, including derivatives, retain CC BY-NC 4.0; see both license files. Challenge data is not included.

Run

Use Python 3.11 and install requirements.txt. The recorded reproduction used Python 3.11.15, PyTorch 2.5.1+cu118, NumPy 2.4.3 and a GTX 1650 Ti. Install the appropriate official PyTorch wheel for your device; cross-runtime bit identity is not guaranteed. The runner supports cpu and cuda:0.

python inference.py --trajectory-root /path/to/test --windows /path/to/index/test_windows.csv --out /new/output/folder --device cuda:0

The output folder must not exist. Outputs are submission.csv, unrounded predictions.npz, and a receipt containing input and artifact hashes. No network connection is needed during inference. Load only the numeric NPZ weight archive; no pickled model object is required.

Inputs and model

One shared TransportState checkpoint contains 4,199,322 parameters in 166 FP32 tensors. It combines a temporal convolution encoder, rotary attention, learned sensor calibration and gyro transport. Four internal velocity offsets and a learned gate are trained within the same model. The gate receives raw IMU features, with no externally supplied platform identity or separately selected models.

The only signal input is 200 Hz six-axis IMU (accelerometer followed by gyroscope, SI units, gravity retained). The runner decodes only the imu array. Window and trajectory identifiers are used for ordering and grouping. Ground-truth orientation, position, velocity and platform metadata are not model inputs. Frozen sensor normalization is in inference_plan.json.

Inference is offline and noncausal: 32-second contexts, 16-second overlap, eight deterministic yaw views of the same weights, inverse rotation and edge-weighted context blending. Trajectory-averaged logits softly mix the internal offsets. Twenty dense estimates form each one-second output. Short sequences repeat the final input frame with a validity mask. This is not a single forward pass and not a causal real-time deployment claim.

Training and evaluation

The publisher reports fitting all 395 original training and 80 validation trajectories, with checkpoint averaging and EMA in its upstream process. Our continuation starts from those released weights and fits only the 395 original training trajectories for exactly 3,072 updates, with a fresh AdamW optimizer, cosine learning rate 5e-6 to 5e-7, weight decay 0.01, gradient clipping 1, batch size 4 and seed 2026091702. One trajectory per platform and a valid 32-second crop are sampled uniformly. Training uses yaw and small constant sensor-bias perturbations, inherited normalization, and corrected end-position/duration trajectory supervision. Our continuation introduces no additional EMA or checkpoint averaging.

Validation has already influenced the upstream model and repeated development decisions. The development composite improved from 0.1152257841 to 0.1109796961 versus the unchanged parent, with improvement on each platform. This is reused development evidence, not an independent generalization estimate. Test labels and official report-service per-sequence metrics were not used to fit this model.

Fresh raw validation reproduced all 23,714 reference predictions. Two independent full test processes produced identical 30,644-window arrays and CSV bytes on the recorded runtime. These are inference reproduction checks, not a full retraining replication. Model/source hashes are verified by the runner.

CSV MD5: a5261507784dd335d2b41e37430f2c5a. A prediction CSV is not bundled or read by inference. Private ranking remains unknown until the organizers release and verify final results.

Scope and limitations

Research use under the stated noncommercial weight license. Unseen sensors, mounting orientations and motion regimes may degrade accuracy. No independent validation set remains for this checkpoint. Long temporal context, eight views and full-trajectory gating impose latency and memory costs. No uncertainty calibration, physical-safety guarantee or autonomous-navigation readiness is claimed.

Credit the TartanIMU challenge and data authors and the original TartanIMU research, in addition to the V25b publisher. The challenge technical report and final verification are separate from this model artifact.