File size: 1,067 Bytes
4bc50fd f256245 4bc50fd f256245 4bc50fd f256245 4bc50fd f256245 4bc50fd f256245 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | # v3-valid eval — fling-excluded, MPC vs pure-VLA (latest code 815ea9c)
Latest vla_ppc (bbox-centre-min distance + completion_frames=2). 50 *valid*
(non-flung) episodes per task for MPC; flung episodes (task object ejected by a
physics glitch, GT speed > 2.0 m/s) are dropped and shown separately. pure-VLA
ran on those same MPC-valid seeds. SmolVLA, max_steps 200, chunk_exec 50.
## Results (env_success)
| task | MPC v3-valid | pure-VLA (same seeds) |
|---|---|---|
| place_bread_skillet | 25/50 (50%) | 25/46 (54%, 4 flung) |
| move_can_pot | 38/50 (76%) | 41/49 (84%, 1 flung) |
**pure-VLA ≥ MPC on both tasks** — MPC correction does not help (slightly hurts)
these tasks; SmolVLA alone is best. The latest distance/de-bounce code did not
change this.
Layout per task: `mpc/` (poselib + MPC: bread=economicgrasp, canpot=object_center)
and `pure_vla/` (passthrough, no MPC). Each has `videos/` (counted episodes) and
`flung_videos/` (dropped; 15-tick grace so the fling is visible). bread flings
often (~34% of seeds); canpot almost never (~2%).
|