-
Notifications
You must be signed in to change notification settings - Fork 8
/
run_cc.sh
31 lines (29 loc) · 869 Bytes
/
run_cc.sh
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
#!/bin/bash
#SBATCH --gres=gpu:1 # Request GPU "generic resources"
#SBATCH --cpus-per-task=6 # Cores proportional to GPUs: 6 on Cedar, 16 on Graham.
#SBATCH --mem=32000M # memory (per node)
#SBATCH --time=0-6:00
#SBATCH --output=%N-%j.out
module load python/3.7
virtualenv --no-download $SLURM_TMPDIR/env
source $SLURM_TMPDIR/env/bin/activate
pip install --upgrade pip
make requirements
make train PROCESSED_DATA_PATH="/home/albanmdb/scratch/datasets/NTU_RGB+D/" \
EVALUATION_TYPE=cross_subject \
MODEL_TYPE=FUSION \
USE_POSE=True \
USE_IR=False \
PRETRAINED=False \
USE_CROPPED_IR=False \
LEARNING_RATE=1e-4 \
WEIGHT_DECAY=0.0 \
GRADIENT_THRESHOLD=10 \
EPOCHS=15 \
BATCH_SIZE=16 \
ACCUMULATION_STEPS=1 \
SUB_SEQUENCE_LENGTH=20 \
AUGMENT_DATA=False \
MIRROR_SKELETON=False \
EVALUATE_TEST=True \
SEED=1