g1_fdm_500

g1_fdm_500 — LingBot-VA G1 post-trained transformer (step 500, FDM)

Fine-tuned transformer for LingBot-VA on Unitree G1 (Dex1) data (bobchenyx/g1_dex1_pick_pink_lerobot033, task: "pick up the pink object and place it on the blue cross mark").

• Base: robbyant/lingbot-va-base
• Post-training: 50 demos, lr 1e-5, FDM loss enabled (lambda_fdm=1.0, Eq. 13 of the LingBot-VA paper), checkpoint at optimizer step 500.
• This repo contains only the transformer/ — vae/, text_encoder/, and tokenizer/ are unchanged from robbyant/lingbot-va-base.

Assemble an eval-ready checkpoint

hf download robbyant/lingbot-va-base --local-dir lingbot-va-base
hf download armanakbari4/g1_fdm_500  --local-dir g1_fdm_500_dl

mkdir -p g1_fdm_500
ln -sf $(realpath g1_fdm_500_dl/transformer)        g1_fdm_500/transformer
ln -sf $(realpath lingbot-va-base/vae)              g1_fdm_500/vae
ln -sf $(realpath lingbot-va-base/text_encoder)     g1_fdm_500/text_encoder
ln -sf $(realpath lingbot-va-base/tokenizer)        g1_fdm_500/tokenizer

Serve with CONFIG_NAME=g1 MODEL_PATH=g1_fdm_500.

transformer/config.json has attn_mode: torch (inference-ready; flex attention is training-only).