/
cmd.cache
41 lines (28 loc) · 5.53 KB
/
cmd.cache
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
# pretrain detector
## training
### 4 gpus distributed
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --master_port 10001 --nproc_per_node=4 tools/detector_pretrain_net.py --config-file "configs/e2e_relation_detector_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 8 TEST.IMS_PER_BATCH 4 DTYPE "float16" SOLVER.MAX_ITER 40000 SOLVER.STEPS "(20000, 27000)" SOLVER.VAL_PERIOD 2000 SOLVER.CHECKPOINT_PERIOD 2000 MODEL.RELATION_ON False OUTPUT_DIR /home/kaihua.tkh/projects/benchmark_debug SOLVER.PRE_VAL False SOLVER.UPDATE_SCHEDULE_DURING_LOAD True
### 2 gpus distributed
CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --master_port 10002 --nproc_per_node=2 tools/detector_pretrain_net.py --config-file "configs/e2e_relation_detector_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 4 TEST.IMS_PER_BATCH 2 DTYPE "float16" SOLVER.MAX_ITER 80000 SOLVER.STEPS "(40000, 57000)" SOLVER.VAL_PERIOD 5000 SOLVER.CHECKPOINT_PERIOD 5000 MODEL.RELATION_ON False OUTPUT_DIR /home/kaihua.tkh/projects/benchmark_debug SOLVER.PRE_VAL False SOLVER.UPDATE_SCHEDULE_DURING_LOAD True
### 1 gpu
CUDA_VISIBLE_DEVICES=0 python tools/detector_pretrain_net.py --config-file "configs/e2e_relation_detector_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 2 DTYPE "float16" SOLVER.MAX_ITER 160000 SOLVER.STEPS "(80000, 108000)" SOLVER.VAL_PERIOD 5000 SOLVER.CHECKPOINT_PERIOD 5000 MODEL.RELATION_ON False OUTPUT_DIR /home/kaihua.tkh/projects/benchmark_debug SOLVER.UPDATE_SCHEDULE_DURING_LOAD True
## testing
### 4 gpus distributed
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --master_port 10003 --nproc_per_node=4 tools/detector_pretest_net.py --config-file "configs/e2e_relation_detector_R_101_FPN_1x.yaml" DTYPE "float16" OUTPUT_DIR /home/kaihua.tkh/projects/benchmark_debug MODEL.RELATION_ON False
# relation prediction
## training
### 4 gpus distributed
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --master_port 10004 --nproc_per_node=4 tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 8 TEST.IMS_PER_BATCH 4 DTYPE "float16" SOLVER.MAX_ITER 25000 SOLVER.STEPS "(8000, 16000)" GLOVE_DIR /home/kaihua.tkh/projects/glove MODEL.PRETRAINED_DETECTOR_CKPT /home/kaihua.tkh/projects/model_final.pth OUTPUT_DIR /home/kaihua.tkh/checkpoints/relation_motif SOLVER.PRE_VAL False
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --master_port 10004 --nproc_per_node=4 tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 16 TEST.IMS_PER_BATCH 4 DTYPE "float16" SOLVER.MAX_ITER 24000 SOLVER.STEPS "(12000, 16000)" GLOVE_DIR /data/share/glove.6B MODEL.PRETRAINED_DETECTOR_CKPT /data/sjx/exp/GQA/pretrain_detector/model_final.pth OUTPUT_DIR /data/sjx/exp/GQA/relation_motif SOLVER.PRE_VAL False
### 2 gpus distributed
CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --master_port 10005 --nproc_per_node=2 tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 4 TEST.IMS_PER_BATCH 2 DTYPE "float16" SOLVER.MAX_ITER 80000 SOLVER.STEPS "(40000, 54000)" GLOVE_DIR /home/kaihua.tkh/projects/glove MODEL.PRETRAINED_DETECTOR_CKPT /home/kaihua.tkh/projects/model_final.pth OUTPUT_DIR /home/kaihua.tkh/checkpoints/relation_motif SOLVER.PRE_VAL False
### 1 gpu
CUDA_VISIBLE_DEVICES=0 python tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 3 TEST.IMS_PER_BATCH 1 DTYPE "float16" SOLVER.MAX_ITER 120000 SOLVER.STEPS "(60000, 81000)" GLOVE_DIR /home/kaihua.tkh/projects/glove MODEL.PRETRAINED_DETECTOR_CKPT /home/kaihua.tkh/projects/model_final.pth OUTPUT_DIR /home/kaihua.tkh/checkpoints/relation_motif SOLVER.PRE_VAL False
CUDA_VISIBLE_DEVICES=0 python tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 4 TEST.IMS_PER_BATCH 1 DTYPE "float16" SOLVER.MAX_ITER 90000 SOLVER.STEPS "(45000, 60000)" GLOVE_DIR /data/share/glove.6B MODEL.PRETRAINED_DETECTOR_CKPT /data/sjx/exp/GQA/pretrain_detector/model_final.pth OUTPUT_DIR /data/sjx/exp/GQA/relation_motif SOLVER.PRE_VAL False
## need to check the performance of model without frequence_bias
## added MODEL.ROI_RELATION_HEAD.PREDICT_USE_BIAS = False for debug
CUDA_VISIBLE_DEVICES=2,3 python -m torch.distributed.launch --master_port 10002 --nproc_per_node=2 tools/relation_train_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" MODEL.ROI_RELATION_HEAD.PREDICT_USE_BIAS False MODEL.ROI_RELATION_HEAD.USE_GT_BOX True MODEL.ROI_RELATION_HEAD.USE_GT_OBJECT_LABEL True SOLVER.IMS_PER_BATCH 8 DTYPE "float16" SOLVER.MAX_ITER 40000 SOLVER.STEPS "(20000, 27000)" SOLVER.VAL_PERIOD 5000 SOLVER.CHECKPOINT_PERIOD 5000 GLOVE_DIR /data4/jiaxin/glove MODEL.PRETRAINED_DETECTOR_CKPT /data4/jiaxin/benchmark_debug/model_final.pth OUTPUT_DIR /data4/jiaxin/relation_debug
# testing
CUDA_VISIBLE_DEVICES=0 python tools/relation_test_net.py --config-file "configs/e2e_relation_R_101_FPN_1x.yaml" TEST.IMS_PER_BATCH 1 DTYPE "float16" GLOVE_DIR /home/kaihua/projects/glove SOLVER.PRE_VAL False OUTPUT_DIR /home/kaihua/checkpoints/TYPE_PATH_HERE
# image retrieval
CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --master_port 10005 --nproc_per_node=2 tools/image_retrieval_main.py --config-file "configs/e2e_relation_X_101_32_8_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 12 SOLVER.MAX_ITER 61 SOLVER.STEPS "(20, 50)" SOLVER.VAL_PERIOD 5 SOLVER.CHECKPOINT_PERIOD 5 SOLVER.SCHEDULE.TYPE "WarmupMultiStepLR" SOLVER.WARMUP_FACTOR 1.0 SOLVER.BASE_LR 0.01 SOLVER.PRE_VAL False OUTPUT_DIR /home/kaihua/checkpoints/TYPE_PATH_HERE