Downloads · 30 days
0
flashwam-arx-student-from-t3000
hz1919810/flashwam-arx-student-from-t3000
flashwam-arx-student-from-t3000 is a robotics model from hz1919810. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for diffusers. The card lists the license as other.
Modality-aware consistency-distilled student of the LingBot-VA ARX world-action model, produced with Flash-WAM (official code implementation, paper-style 2000-step duration).
Access
Public
Updated Sep 3, 2026
Repo size
20.4 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors20.4 GB · 100%
At a glance
- Task
- Robotics
- Library
- diffusers
- License
- other
- Access
- Public
- Created
- Sep 3, 2026
- Updated
- Sep 3, 2026
- SHA
- 4f3f96f6
Model card
From the Hugging Face model README
Flash-WAM Distilled Student (ARX) — from teacher step 3000
Modality-aware consistency-distilled student of the LingBot-VA ARX world-action model, produced with Flash-WAM (official code implementation, paper-style 2000-step duration).
Teacher: LingBot-VA ARX SFT v2 split checkpoint step 3000 (160-episode train split). Distilled on the same 160 train episodes.
Contents
target_student/transformer/— EMA target (recommended for inference).online_student/transformer/— trainable online student at step 2000.
Both are diffusers-format WanTransformer3DModel (config.json + bf16
safetensors, ~9.5 GB each).
Distillation recipe (per collaborator spec)
| mode | flashwam (video + action consistency + action-aware reg) |
| steps | 2000 (paper duration), save 500/1000/1500/2000 |
| lr / betas / wd / warmup | 5e-6 / (0.9, 0.999) / 0 / 100 |
| ema_decay | 0.995 |
| loss / huber_c | huber / 0.001 |
| action_loss_weight / action_aware_weight | 1.0 / 0.01 |
| num_ddim_timesteps (video / action) | 2 / 2 |
| cfg range | [2.0, 10.0] |
| effective global batch | 16 (4 GPU x accum 4) |
| final loss (video / action) | 0.2877 / 0.0120 |
ARX contract (unchanged from teacher)
3 cameras (cam_high, cam_left_wrist, cam_right_wrist); 256x256; K2 profile
video_7p5hz_action_15hz_k2; 14-D absolute joint target -> 30-D layout
[14,15,16,17,18,19,28,21,22,23,24,25,26,29]; dataset q01/q99;
env_type=none, action_per_frame=8, attn_window=30.
Caveat
This is a fast (1-2 step/modality) student. Low distillation loss does not guarantee multi-step rollout fidelity or real-robot task success. Evaluate with rollout / real-robot before deployment; final selection (online vs EMA, and vs teacher) is the data owner's call.
- Task
- Robotics
- Library
- diffusers
- License
- other
- Created
- Sep 3, 2026
- Updated
- Sep 3, 2026