diff --git "a/swap_analysis_synthetic/logs_consistent_ratio_0.5_short/nvila/roborefer.log" "b/swap_analysis_synthetic/logs_consistent_ratio_0.5_short/nvila/roborefer.log" new file mode 100644--- /dev/null +++ "b/swap_analysis_synthetic/logs_consistent_ratio_0.5_short/nvila/roborefer.log" @@ -0,0 +1,1688 @@ +2026-03-12 14:38:30,541 - INFO - [synthetic] Logging to : /data/shared/Qwen/experiments/swap_analysis_synthetic/logs_consistent_ratio_0.5_short/nvila_synthetic.log +2026-03-12 14:38:30,541 - INFO - [synthetic] Counter-only : False +2026-03-12 14:38:30,541 - INFO - [synthetic] Consistent ratio : 0.5 +2026-03-12 14:38:30,541 - INFO - [synthetic] Question type : short_answer +2026-03-12 14:38:30,541 - INFO - [synthetic] VD-only : False +2026-03-12 14:38:30,541 - INFO - [synthetic] Data dir : /data/shared/Qwen/synthetic/2body_bias +2026-03-12 14:38:30,541 - INFO - [synthetic] Output dir : /data/shared/Qwen/experiments/swap_analysis_synthetic/results_consistent_ratio_0.5_short +2026-03-12 14:38:30,542 - INFO - +=== Loading Synthetic Swap Pairs === +2026-03-12 14:38:30,542 - INFO - [mixed] consistent_ratio=0.5 → 100 consistent + 100 counter per category (total 200) +2026-03-12 14:38:32,456 - INFO - Mixed swap pair creation stats: +2026-03-12 14:38:32,457 - INFO - left: 200/200 +2026-03-12 14:38:32,457 - INFO - right: 200/200 +2026-03-12 14:38:32,457 - INFO - above: 200/200 +2026-03-12 14:38:32,457 - INFO - below: 200/200 +2026-03-12 14:38:32,457 - INFO - far: 200/200 +2026-03-12 14:38:32,457 - INFO - close: 200/200 +2026-03-12 14:38:32,457 - INFO - Total pairs: 1200 +2026-03-12 14:38:32,462 - INFO - Synthetic cross-group quads: 400/400 (ambiguous=0) +2026-03-12 14:38:32,462 - INFO - +============================================================ +2026-03-12 14:38:32,462 - INFO - Processing nvila - roborefer +2026-03-12 14:38:32,462 - INFO - Model path: /data/shared/Qwen/mydisk/RoboRefer_model +2026-03-12 14:38:32,462 - INFO - ============================================================ +[2026-03-12 14:38:36,173] [INFO] [real_accelerator.py:110:get_accelerator] Setting ds_accelerator to cuda (auto detect) +2026-03-12 14:38:37,057 - INFO - PyTorch version 2.3.0 available. +2026-03-12 14:38:41,414 - INFO - We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk). +You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`. +2026-03-12 14:39:03,287 - INFO - Loaded RoboRefer from /data/shared/Qwen/mydisk/RoboRefer_model +2026-03-12 14:39:03,287 - INFO - Model has 28 layers. Extracting ALL. +2026-03-12 14:39:03,288 - INFO - +--- Phase A: Extracting swap pair features --- + Swap pairs: 0%| | 0/1200 [00:00