raw/run_width32_aborted.log

[16:46:18] INVOCATION: bash /workspace/tom_channel_full.sh (stages 2+3, encoder=/workspace/large_artifacts/ii_mats/tom_channel_v1/distill/encoder.pt, NEED_MIB=60000)
[16:46:48] GPU has 182632 MiB free (>= 60000); starting
[16:46:52] INVOCATION: python -m tom.nbs_channel_eval --bank instances_nbs_confirmatory_v1 --encoder-checkpoint /workspace/large_artifacts/ii_mats/tom_channel_v1/distill/encoder.pt --out /workspace/large_artifacts/ii_mats/tom_channel_v1/campaign --model Qwen/Qwen3-8B --model-thinking off --seeds 0 1 2 3 4 5 6 7 8 9 --max-pool-episodes 32
[16:46:52] bank instances_nbs_confirmatory_v1: 48 games | playing 24 HELD-OUT games x 10 seeds x 4 arms = 960 episodes
[16:46:52] encoder: /workspace/large_artifacts/ii_mats/tom_channel_v1/distill/encoder.pt | n_soft=38 hidden=512 gate_ratio=0.2783 (trained on 24 games)

Loading checkpoint shards:   0%|          | 0/5 [00:00<?, ?it/s]
Loading checkpoint shards: 100%|██████████| 5/5 [00:00<00:00, 112.83it/s]
[16:47:08] encoder fingerprint 8c7709c64f65988f (identical for B_soft and B_shuffled)
[16:58:53] A_text seeds 0-9: 32/32 episodes done | registry 0 prompts | fabricated 0/407 turns | 722s elapsed