Skip to content

Summarizing Emotion Triggers with TRansformers.

Notifications You must be signed in to change notification settings

kimbochen/settr

Repository files navigation

Summarizing Emotion Triggers with TRansformers

Install Packages

pip install -r requirements.txt

Data Setup

Put the data in the folder data in the settr directory. Data is not released and will be provided upon request.

Training

See launch_train.sh for examples.

Evaluation

./launch_eval.sh <Experiment Name>

Useful Commands

Connect to a Slurm Cluster Node

# Open a bash shell
salloc -p <partition> --gres gpu:1 --mem 12g -t 3:00:00

# Run an interactive job
srun -p <partition> --gres gpu:1 --mem 12g -t 3:00:00

# Run another job within the same node
srun --jobid <job_id>

# Check job status and up time
squeue -u $USER

# Show node status
sinfo

# Show node info
scontrol show nodes <node_hostname>

# Show partition info
scontrol show partition <partition_name>

Connect to Jupyter Notebooks on Slurm Cluster Nodes

# Find the IP <ip> of the node
cat /etc/hosts

# Listen to all ports
jupyter notebook --no-browser --port=8888 --ip=0.0.0.0

# On the local computer
ssh cs260 -NL 8888:<ip>:8888

About

Summarizing Emotion Triggers with TRansformers.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published