Added documentation of process of setting up the environment to run o…

…n compute nodes. Added python script to download and save models from Huggingface.

AttendExcite/save_model.py

@@ -0,0 +1,8 @@
+from pipeline_attend_and_excite import AttendAndExcitePipeline
+
+model = input("Huggingface model to save: ")
+modelname = input("Model name to save to ./models: ")
+
+stable = AttendAndExcitePipeline.from_pretrained(model)
+
+stable.save_pretrained(f"./models/{modelname}")

BYURC_setup.md

@@ -0,0 +1,19 @@
+## Environment setup for sending batch Attend and Excite jobs
+
+If the `miniconda3` module is not yet loaded, run:
+`module load miniconda3`
+
+Create the latent diffusion model environment, `ldm`, by running:
+`conda env create -f environment.yaml`
+
+Add the following to `~/.bashrc`
+
+```bash
+# Activate ldm conda environment.
+conda activate ldm
+
+# Add LD_LIBRARY_PATH
+export LD_LIBRARY_PATH=/home/tysweat0/.conda/envs/ldm/lib
+```
+
+Run `StableComics/AttendExcite/save_model.py` to download the model locally. The compute nodes do not allow access to the Internet.

run_attend_excite.sh

@@ -7,7 +7,7 @@
 #SBATCH -C pascal
 
 
-cd ~/AttendExcite/
+cd ~/StableComics/AttendExcite/
 
 nvidia-smi --list-gpus
 nvidia-smi --query-gpu=memory.total --format=csv