diff --git "a/train_job_output.txt" "b/train_job_output.txt" --- "a/train_job_output.txt" +++ "b/train_job_output.txt" @@ -1,4 +1,4 @@ -slurm submission log: 2024-05-20 10:11:57.803652 +slurm submission log: 2024-05-20 23:23:27.080509 created following sbatch script: ############################### @@ -7,13 +7,13 @@ created following sbatch script: #SBATCH --account=nlp #SBATCH --cpus-per-task=16 -#SBATCH --dependency=afterok:7635783 +#SBATCH --dependency=afterok:7637747 #SBATCH --gres=gpu:2 -#SBATCH --job-name=tthrush-job-1992241 +#SBATCH --job-name=tthrush-job-4852181 #SBATCH --mem=100G #SBATCH --nodelist=sphinx2 #SBATCH --open-mode=append -#SBATCH --output=/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/70m_llms_more_data_test_normalized/pythia-70m_lambada_1/train_job_output.txt +#SBATCH --output=/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/test_ordinal_clipped_scaled/llms/pythia-70m_lambada_1/train_job_output.txt #SBATCH --partition=sphinx #SBATCH --time=14-0 @@ -24,7 +24,7 @@ created following sbatch script: cd . # launch commands -srun --unbuffered run_as_child_processes 'torchrun --master_port 29506 --nproc_per_node=2 train_llm.py --dataset_id /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/train_data_more_data_test_normalized/lambada --output_dir /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/70m_llms_more_data_test_normalized/pythia-70m_lambada_1 --output_hub_id pythia-70m_lambada --model_id EleutherAI/pythia-70m --learning_rate 1e-3 --warmup_ratio=0.1 --gradient_accumulation_steps 2 --per_device_train_batch_size 256 --seed 1 --num_train_epochs 1' +srun --unbuffered run_as_child_processes 'torchrun --master_port 29526 --nproc_per_node=2 train_llm.py --dataset_id /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/test_ordinal_clipped_scaled/data/lambada --output_dir /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/test_ordinal_clipped_scaled/llms/pythia-70m_lambada_1 --output_hub_id pythia-70m_lambada --model_id EleutherAI/pythia-70m --learning_rate 1e-3 --warmup_ratio=0.1 --gradient_accumulation_steps 2 --per_device_train_batch_size 256 --seed 1 --num_train_epochs 1' ############################### @@ -34,13 +34,13 @@ submission to slurm complete! ############################### slurm submission output -Submitted batch job 7635784 +Submitted batch job 7637748 ############################### -/var/lib/slurm/slurmd/job7635784/slurm_script: line 16: /nlp/scr/tthrush/miniconda3/envs/pretraining-coreset-selection/etc/profile.d/conda.sh: No such file or directory +/var/lib/slurm/slurmd/job7637748/slurm_script: line 16: /nlp/scr/tthrush/miniconda3/envs/pretraining-coreset-selection/etc/profile.d/conda.sh: No such file or directory CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'. To initialize your shell, run @@ -61,454 +61,455 @@ IMPORTANT: You may need to close and restart your shell after running 'conda ini ############################### -start time: 2024-05-20 14:03:29.621062 +start time: 2024-05-21 09:17:53.339239 machine: sphinx2 conda env: pretraining-coreset-selection ############################### running following processes - torchrun --master_port 29506 --nproc_per_node=2 train_llm.py --dataset_id /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/train_data_more_data_test_normalized/lambada --output_dir /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/70m_llms_more_data_test_normalized/pythia-70m_lambada_1 --output_hub_id pythia-70m_lambada --model_id EleutherAI/pythia-70m --learning_rate 1e-3 --warmup_ratio=0.1 --gradient_accumulation_steps 2 --per_device_train_batch_size 256 --seed 1 --num_train_epochs 1 + torchrun --master_port 29526 --nproc_per_node=2 train_llm.py --dataset_id /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/test_ordinal_clipped_scaled/data/lambada --output_dir /juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/test_ordinal_clipped_scaled/llms/pythia-70m_lambada_1 --output_hub_id pythia-70m_lambada --model_id EleutherAI/pythia-70m --learning_rate 1e-3 --warmup_ratio=0.1 --gradient_accumulation_steps 2 --per_device_train_batch_size 256 --seed 1 --num_train_epochs 1 ############################### command outputs: -[2024-05-20 14:03:32,107] torch.distributed.run: [WARNING] -[2024-05-20 14:03:32,107] torch.distributed.run: [WARNING] ***************************************** -[2024-05-20 14:03:32,107] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. -[2024-05-20 14:03:32,107] torch.distributed.run: [WARNING] ***************************************** -05/20/2024 14:03:39 - INFO - __main__ - Script parameters ScriptArguments(seed=1, dataset_id='/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/train_data_more_data_test_normalized/lambada', output_dir='/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/70m_llms_more_data_test_normalized/pythia-70m_lambada_1', output_hub_id='pythia-70m_lambada', hf_hub_token=True, model_id='EleutherAI/pythia-70m', per_device_train_batch_size=256, num_train_epochs=1.0, learning_rate=0.001, gradient_accumulation_steps=2, from_scratch=True, warmup_ratio=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, weight_decay=0.01, lr_scheduler_type='cosine', local_rank=0, resume_from_checkpoint=False, deepspeed=None, peft=False) -05/20/2024 14:03:41 - INFO - __main__ - Script parameters ScriptArguments(seed=1, dataset_id='/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/train_data_more_data_test_normalized/lambada', output_dir='/juice5/scr5/tthrush/pretraining-coreset-selection/llm_pretraining/70m_llms_more_data_test_normalized/pythia-70m_lambada_1', output_hub_id='pythia-70m_lambada', hf_hub_token=True, model_id='EleutherAI/pythia-70m', per_device_train_batch_size=256, num_train_epochs=1.0, learning_rate=0.001, gradient_accumulation_steps=2, from_scratch=True, warmup_ratio=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, weight_decay=0.01, lr_scheduler_type='cosine', local_rank=0, resume_from_checkpoint=False, deepspeed=None, peft=False) - 0%| | 0/10699 [00:00