Downloads · 30 days
1
14% of all-time downloads
SleepMastger/pcbnew100_flashwam_ft
pcbnew100_flashwam_ft is a machine learning model from SleepMastger. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
FlashWAM (M1, fused-KV, RoPE-fixed) finetuned from the LIBERO checkpoint on the first 100 trajectories of the placecubeinbowl new dataset (Franka, 10 Hz, 2-cam). 30 epochs, global batch 32 (2 GPU x 16). Slurm job 852…
Downloads · 30 days
1
14% of all-time downloads
All-time downloads
7
Public
Repo size
60.8 GB
Likes
0
Public
Click a slice to open those files.
.pt60.8 GB · 100%
From the Hugging Face model README
FlashWAM (M1, fused-KV, RoPE-fixed) finetuned from the LIBERO checkpoint on the first 100 trajectories of the place_cube_in_bowl new dataset (Franka, 10 Hz, 2-cam). 30 epochs, global batch 32 (2 GPU x 16). Slurm job 852 on gpu-h200-103.
Direct FT-vs-scratch comparison with pcbnew100_flashwam_scratch.
Checkpoints in checkpoints/weights/ at steps 2445/4890/7335/9780/12225/14670 (every 5 epochs; 489 opt-steps/epoch).
wandb: huaweiwam/fastwam-realrobot/runs/d7291eli