diff --git a/smolvla-split-gk_adp-libero-1000/17-08-53_smolvla_gk_adp_libero_object_2e-3_split2/libero_object_task02_065000/eval_task09_a10/eval.log b/smolvla-split-gk_adp-libero-1000/17-08-53_smolvla_gk_adp_libero_object_2e-3_split2/libero_object_task02_065000/eval_task09_a10/eval.log index 2ace43878041c2ee678f1d9163720924d8548a7e..c2448002e21f6fed1ec26a8ae2dea45bf2de75e9 100644 --- a/smolvla-split-gk_adp-libero-1000/17-08-53_smolvla_gk_adp_libero_object_2e-3_split2/libero_object_task02_065000/eval_task09_a10/eval.log +++ b/smolvla-split-gk_adp-libero-1000/17-08-53_smolvla_gk_adp_libero_object_2e-3_split2/libero_object_task02_065000/eval_task09_a10/eval.log @@ -10414,4 +10414,8956 @@ Assets already downloaded at /home/tml_lab/.cache/libero/assets Running rollout with at most 280 steps: 9%|▉ | 25/280 [00:01<00:13, 19.27it/s, running_success_rate=0.0%] Running rollout with at most 280 steps: 9%|▉ | 26/280 [00:01<00:12, 21.14it/s, running_success_rate=0.0%] Running rollout with at most 280 steps: 9%|▉ | 26/280 [00:01<00:12, 21.14it/s, running_success_rate=0.0%] - Running rollout with at most 280 steps: 10%|▉ | 27/280 [00:01<00:11, 21.14it/s, running_success_rate=0.0%] \ No newline at end of file + Running rollout with at most 280 steps: 10%|▉ | 27/280 [00:01<00:11, 21.14it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 10%|█ | 28/280 [00:01<00:11, 21.14it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 10%|█ | 29/280 [00:01<00:11, 22.74it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 10%|█ | 29/280 [00:01<00:11, 22.74it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 11%|█ | 30/280 [00:01<00:10, 22.74it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 11%|█ | 31/280 [00:01<00:10, 22.74it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 11%|█▏ | 32/280 [00:01<00:16, 14.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 11%|█▏ | 32/280 [00:01<00:16, 14.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 12%|█▏ | 33/280 [00:01<00:16, 14.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 12%|█▏ | 34/280 [00:01<00:16, 14.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 12%|█▎ | 35/280 [00:01<00:16, 14.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 13%|█▎ | 36/280 [00:01<00:13, 18.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 13%|█▎ | 36/280 [00:01<00:13, 18.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 13%|█▎ | 37/280 [00:01<00:13, 18.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 14%|█▎ | 38/280 [00:01<00:12, 18.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 14%|█▍ | 39/280 [00:01<00:12, 18.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 14%|█▍ | 40/280 [00:01<00:10, 22.46it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 14%|█▍ | 40/280 [00:02<00:10, 22.46it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 15%|█▍ | 41/280 [00:02<00:10, 22.46it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 15%|█▌ | 42/280 [00:02<00:10, 22.46it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 15%|█▌ | 43/280 [00:02<00:10, 22.46it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 16%|█▌ | 44/280 [00:02<00:21, 10.81it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 16%|█▌ | 44/280 [00:02<00:21, 10.81it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 16%|█▌ | 45/280 [00:02<00:21, 10.81it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 16%|█▋ | 46/280 [00:02<00:21, 10.81it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 17%|█▋ | 47/280 [00:02<00:20, 11.63it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 17%|█▋ | 47/280 [00:02<00:20, 11.63it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 17%|█▋ | 48/280 [00:02<00:19, 11.63it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 18%|█▊ | 49/280 [00:03<00:19, 11.63it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 18%|█▊ | 50/280 [00:03<00:18, 12.19it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 18%|█▊ | 50/280 [00:03<00:18, 12.19it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 18%|█▊ | 51/280 [00:03<00:18, 12.19it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 19%|█▊ | 52/280 [00:03<00:33, 6.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 19%|█▊ | 52/280 [00:03<00:33, 6.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 19%|█▉ | 53/280 [00:03<00:32, 6.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 19%|█▉ | 54/280 [00:03<00:28, 7.95it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 19%|█▉ | 54/280 [00:04<00:28, 7.95it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 20%|█▉ | 55/280 [00:04<00:28, 7.95it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 20%|██ | 56/280 [00:04<00:24, 9.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 20%|██ | 56/280 [00:04<00:24, 9.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 20%|██ | 57/280 [00:04<00:24, 9.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 21%|██ | 58/280 [00:04<00:23, 9.64it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 21%|██ | 58/280 [00:04<00:23, 9.64it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 21%|██ | 59/280 [00:04<00:22, 9.64it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 21%|██▏ | 60/280 [00:04<00:20, 10.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 21%|██▏ | 60/280 [00:05<00:20, 10.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 22%|██▏ | 61/280 [00:05<00:20, 10.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 22%|██▏ | 62/280 [00:05<00:37, 5.79it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 22%|██▏ | 62/280 [00:05<00:37, 5.79it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 22%|██▎ | 63/280 [00:05<00:37, 5.79it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 23%|██▎ | 64/280 [00:05<00:30, 7.10it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 23%|██▎ | 64/280 [00:05<00:30, 7.10it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 23%|██▎ | 65/280 [00:05<00:30, 7.10it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 24%|██▎ | 66/280 [00:05<00:27, 7.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 24%|██▎ | 66/280 [00:05<00:27, 7.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 24%|██▍ | 67/280 [00:05<00:26, 7.93it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 24%|██▍ | 68/280 [00:05<00:22, 9.35it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 24%|██▍ | 68/280 [00:05<00:22, 9.35it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 25%|██▍ | 69/280 [00:05<00:22, 9.35it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 25%|██▌ | 70/280 [00:05<00:19, 10.77it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 25%|██▌ | 70/280 [00:06<00:19, 10.77it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 25%|██▌ | 71/280 [00:06<00:19, 10.77it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 26%|██▌ | 72/280 [00:06<00:40, 5.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 26%|██▌ | 72/280 [00:06<00:40, 5.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 26%|██▌ | 73/280 [00:06<00:40, 5.16it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 26%|██▋ | 74/280 [00:06<00:31, 6.50it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 26%|██▋ | 74/280 [00:06<00:31, 6.50it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 27%|██▋ | 75/280 [00:06<00:31, 6.50it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 27%|██▋ | 76/280 [00:06<00:27, 7.43it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 27%|██▋ | 76/280 [00:06<00:27, 7.43it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 28%|██▊ | 77/280 [00:06<00:27, 7.43it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 28%|██▊ | 78/280 [00:06<00:22, 8.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 28%|██▊ | 78/280 [00:07<00:22, 8.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 28%|██▊ | 79/280 [00:07<00:22, 8.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 29%|██▊ | 80/280 [00:07<00:19, 10.36it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 29%|██▊ | 80/280 [00:08<00:19, 10.36it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 29%|██▉ | 81/280 [00:08<00:19, 10.36it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 29%|██▉ | 82/280 [00:08<00:45, 4.34it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 29%|██▉ | 82/280 [00:08<00:45, 4.34it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 30%|██▉ | 83/280 [00:08<00:45, 4.34it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 30%|███ | 84/280 [00:08<00:39, 4.96it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 30%|███ | 84/280 [00:08<00:39, 4.96it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 30%|███ | 85/280 [00:08<00:36, 5.40it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 30%|███ | 85/280 [00:08<00:36, 5.40it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███ | 86/280 [00:08<00:32, 5.92it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███ | 86/280 [00:08<00:32, 5.92it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███ | 87/280 [00:08<00:29, 6.45it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███ | 87/280 [00:08<00:29, 6.45it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███▏ | 88/280 [00:08<00:27, 6.95it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 31%|███▏ | 88/280 [00:08<00:27, 6.95it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▏ | 89/280 [00:08<00:25, 7.51it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▏ | 89/280 [00:09<00:25, 7.51it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▏ | 90/280 [00:09<00:24, 7.91it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▏ | 90/280 [00:10<00:24, 7.91it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▎ | 91/280 [00:10<01:39, 1.91it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 32%|███▎ | 91/280 [00:10<01:39, 1.91it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 33%|███▎ | 92/280 [00:10<01:17, 2.42it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 33%|███▎ | 92/280 [00:10<01:17, 2.42it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 33%|███▎ | 93/280 [00:10<01:04, 2.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 33%|███▎ | 93/280 [00:11<01:04, 2.88it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▎ | 94/280 [00:11<00:55, 3.34it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▎ | 94/280 [00:11<00:55, 3.34it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▍ | 95/280 [00:11<00:46, 4.02it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▍ | 95/280 [00:11<00:46, 4.02it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▍ | 96/280 [00:11<00:42, 4.37it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 34%|███▍ | 96/280 [00:11<00:42, 4.37it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▍ | 97/280 [00:11<00:35, 5.09it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▍ | 97/280 [00:11<00:35, 5.09it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▌ | 98/280 [00:11<00:34, 5.20it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▌ | 98/280 [00:11<00:34, 5.20it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▌ | 99/280 [00:11<00:34, 5.26it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 35%|███▌ | 99/280 [00:12<00:34, 5.26it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▌ | 100/280 [00:12<00:30, 5.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▌ | 100/280 [00:13<00:30, 5.82it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▌ | 101/280 [00:13<01:48, 1.66it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▌ | 101/280 [00:13<01:48, 1.66it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▋ | 102/280 [00:13<01:25, 2.09it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 36%|███▋ | 102/280 [00:14<01:25, 2.09it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 37%|███▋ | 103/280 [00:14<01:08, 2.57it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 37%|███▋ | 103/280 [00:14<01:08, 2.57it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 37%|███▋ | 104/280 [00:14<00:54, 3.21it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 37%|███▋ | 104/280 [00:14<00:54, 3.21it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 105/280 [00:14<00:47, 3.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 105/280 [00:14<00:47, 3.67it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 106/280 [00:14<00:42, 4.07it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 106/280 [00:14<00:42, 4.07it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 107/280 [00:14<00:36, 4.72it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 38%|███▊ | 107/280 [00:14<00:36, 4.72it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 39%|███▊ | 108/280 [00:14<00:39, 4.36it/s, running_success_rate=0.0%] + Running rollout with at most 280 steps: 39%|███▊ | 108/280 [00:15<00:39, 4.36it/s, running_success_rate=100.0%] + Running rollout with at most 280 steps: 39%|███▉ | 109/280 [00:15<00:33, 5.15it/s, running_success_rate=100.0%] +  Stepping through eval batches: 22%|██▏ | 11/50 [02:04<07:50, 12.06s/it, running_success_rate=33.3%] Stepping through eval batches: 24%|██▍ | 12/50 [02:04<10:50, 17.12s/it, running_success_rate=33.3%] + Running rollout with at most 280 steps: 0%| | 0/280 [00:00 + +GK expert activation by evaluation task +Cells show mean activation weight per router invocation +Expert 0 +Expert 1 +Expert 2 +Expert 3 +Expert 4 +Expert 5 +Expert 6 +Expert 7 +Expert 8 +Expert 9 +libero_object / task 9 + +0.0% + +0.0% + +0.0% + +0.0% + +0.0% + +0.0% + +0.0% + +0.0% + +0.0% + +100.0% +Top-1 produces one-hot activations; weighted routing displays mixture weights. +