Trained with LeRobot.
Open-weight model · Robotics
eqm-aloha_transfer_cube-seed3-18sep2026_1pm
by Kazi Abrar Mahmud iFaz/eqm-aloha_transfer_cube-seed3-18sep2026_1pm
Trained with LeRobot.
Parameters19M
Context—
Weights74.8 MB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads—
Runs On
What it takes to serve eqm-aloha_transfer_cube-seed3-18sep2026_1pm (19M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
Model Card
By Kazi Abrar Mahmud, published under apache-2.0, revision 1b1bbba0ea04.
Trained with LeRobot.
Read Kazi Abrar Mahmud's full model card
EQM Policy — eqm_aloha_transfer_cube_seed3_18sep2026_1pm
Trained with LeRobot.
Date: 2026-09-18 13:43
Policy type: eqm | Device: cuda
Dataset
| Parameter | Value |
|---|---|
dataset.repo_id |
lerobot/aloha_sim_transfer_cube_human |
Training Config
| Parameter | Value |
|---|---|
steps |
10000 |
batch_size |
8 |
eval_freq |
0 |
save_freq |
5000 |
num_workers |
4 |
seed |
3 |
eval.n_episodes |
1 |
eval.batch_size |
1 |
eval.use_async_envs |
True |
Policy Architecture
| Parameter | Value |
|---|---|
ebm |
dot |
jacobian_reg_weight |
1e-05 |
jacobian_reg_probes |
4 |
fixed_point_anchor_weight |
0.1 |
use_adaptive_compute |
False |
Eval Config
| Parameter | Value |
|---|---|
env.type |
aloha |
env.task |
AlohaTransferCube-v0 |
eval.n_episodes |
1 |
eval.batch_size |
1 |
eval.use_async_envs |
False |
policy.path |
/content/outputs/train/aloha_transfer_cube_seed3/checkpoints/last/pretrained_model |
policy.ood_logging_enabled |
True |
policy.ood_calibration_stats_path |
/content/eqm_calibration.json |
policy.ood_log_path |
/content/outputs/eqm_ood_log.csv |
policy.ood_z_threshold |
3.0 |
policy.sample_stepsize |
0.17 |
Eval Results
| Metric | Value |
|---|---|
| Episodes | 1 |
| Success rate | 0.0% |
| Avg sum reward | 6.00 |
| Avg max reward | 3.00 |
| Eval time (s) | 38.3 |
Citation
@misc{cadene2024lerobot,
author = {Cadene, Remi and Alibert, Simon and others},
title = {LeRobot},
year = {2024},
url = {https://github.com/huggingface/lerobot}
}
Configuration
- Hidden size
- 256
Identity and Version
- Repository
- iFaz/eqm-aloha_transfer_cube-seed3-18sep2026_1pm
- Publisher
- Kazi Abrar Mahmud
- Task
- Robotics
- Modality
- Control
- Library
- lerobot
- Parameters
- 19M parameters
- Languages
- en
- Revision
- 1b1bbba0ea04500981ace9a00dc426632c80855b
- First published
- 2026-09-18
- Last updated
- 2026-09-18
Files and Weights
9 files, 74.9 MB in total. The weights are 3 files totalling 74.8 MB in safetensors.
Weights3 files · 74.8 MB
Configuration4 files · 11.1 KB
Documentation1 file · 2.0 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 74.8 MB | 07e8e232dd27 |
| policy_postprocessor_step_0_unnormalizer_processor.safetensors | Weights | 3.9 KB | 797777e6e870 |
| policy_preprocessor_step_3_normalizer_processor.safetensors | Weights | 3.9 KB | 797777e6e870 |
| config.json | Configuration | 2.6 KB | — |
| policy_postprocessor.json | Configuration | 659 B | — |
| policy_preprocessor.json | Configuration | 1.2 KB | — |
| train_config.json | Configuration | 6.7 KB | — |
| README.md | Documentation | 2.0 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 74.8 MB
Download from Kazi Abrar Mahmud
Released by Kazi Abrar Mahmud through its official repository on Hugging Face. Read the license.
Built From
- Trained on (disclosed) lerobot/aloha_sim_transfer_cube_human
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 74.8 MB |
| 16-bit | 0.0 GB |
| 8-bit | 0.0 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About eqm-aloha_transfer_cube-seed3-18sep2026_1pm
How much GPU memory does eqm-aloha_transfer_cube-seed3-18sep2026_1pm need?
About 0 GB at 16-bit and 0 GB at 4-bit: the weights (19M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run eqm-aloha_transfer_cube-seed3-18sep2026_1pm on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use eqm-aloha_transfer_cube-seed3-18sep2026_1pm commercially?
Yes. eqm-aloha_transfer_cube-seed3-18sep2026_1pm is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
Similar Models
This policy has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs. For a complete walkthrough, see the training guide. Below is the short version on how to train and run inference/eval: Writes checkpoints to outputs/train/ /checkpoints/. Prefix the dataset repo with eval\ and supply --policy.path pointing to a local or hub checkpoint.
This policy has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs. For a complete walkthrough, see the training guide. Below is the short version on how to train and run inference/eval: Writes checkpoints to outputs/train/ /checkpoints/. Prefix the dataset repo with eval\ and supply --policy.path pointing to a local or hub checkpoint.
This policy has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs. For a complete walkthrough, see the training guide. Below is the short version on how to train and run inference/eval: Writes checkpoints to outputs/train/ /checkpoints/. Prefix the dataset repo with eval\ and supply --policy.path pointing to a local or hub checkpoint.
LIBERO 4in1（liberospatial / liberoobject / liberogoal / libero10）共 53.19 GB 的 Wan2.2-VAE 编码 latent 缓存： 训练 LIBERO policy（action head / VLA）时直接读取 latent 缓存，避免重复 VAE 编码。 - 窗口模式: windowed（--windowed） - 输出:.pt 文件，每 episode 一个 - dataset（推荐）: https://huggingface.co/datasets/MangoGoes/libero4in1wan2.2vaelatentdataset - model（本仓库）: https://huggingface.co/MangoGoes/libero4in1wan2.2vaelatentcosmosstyle
Wan2.2-TI2V-5B video DiT + 48-joint action head, trainingmode=joint. The base is suhyeok's finalized B-method recipe: a teacher-forced (sigma=0.25) self-EMA teacher plus an iBOT prototype loss at L18 L18, gamma=0.01, two-view. On top of it the 3 PAST cond latent frames are pooled into one motion frame before a chosen block. These are NOT the surrogate ctxpool runs. The surrogate line (older base, pd8 x GA1) lives in hmkang/wamctxpoolxattn and hmkang/wamctxpoolavg. Do not compare across the two sets. Geometry: 4-latin (numframesin=25, numframesout=41, fdf 2) = 4 cond + 2 future latent slots, 96 tokens per latent frame, 576 tokens per row. Effective batch 16 clips x GA 2 x 2 views = 64 rows…