train_pool8_msg_len_6.slurm

#! /bin/sh
#SBATCH --job-name=pool8_msg_len_6
#SBATCH --output="outs/create_language_data_%j.out"
#SBATCH --error="outs/create_language_data_%j.err"
#SBATCH --partition=studentbatch
#SBATCH --gpus=1
#SBATCH --mail-user=shahafgoren@mail.tau.ac.il
#SBATCH --mail-type=ALL

python model_symmetric.py -experiment_name pool8_msg_len_6 -batch_size 128 -batch_size_dev 50 -save_interval 1000 -save_distinct_interval 50000 -m_dim 6 -h_dim 100 -desc_dim 100 -num_classes 10 -learning_rate 1e-4 -entropy_agent1 0.01 -entropy_agent2 0.01 -use_binary -max_epoch 10000 -top_k_dev 1 -top_k_train 1 -dataset_path ./data/oneshape/oneshape_simple_textselect -dataset_indomain_valid_path ./data/oneshape_valid/oneshape_simple_textselect -dataset_outdomain_valid_path ./data/oneshape_valid_all_combos/oneshape_simple_textselect -dataset_name oneshape_simple_textselect -dataset_size_train 5000 -dataset_size_dev 1000 -wv_dim 100 -glove_path ./glove/glove-100d.txt -log_path ./logs/ -debug_log_level INFO -cuda -log_interval 1000 -log_dev 5000 -log_self_com 25000 -reward_type "cooperative" -randomize_comms -random_seed 17 -agent_pools -num_agents 8 -check_accuracy_interval 10000