Finetune slurm script and loss plot

requirements.txt

@@ -9,3 +9,4 @@ requests
 tqdm
 gensim
 accelerate
+matplotlib

scripts/NEC_eval_cl.sh

+#!/bin/bash
+
+#SBATCH --job-name=NEC_evaluation
+#SBATCH --output=logs/NEC_evaluation_%j.txt
+#SBATCH --ntasks=1
+#SBATCH --time=24:00:00
+#SBATCH --mem=8000
+#SBATCH --mail-type=ALL
+#SBATCH --mail-user=kupper@cl.uni-heidelberg.de
+#SBATCH --partition=students
+#SBATCH --cpus-per-task=4
+#SBATCH --qos=batch
+#SBATCH --gres=gpu
+
+export PYTHONUNBUFFERED=1
+srun python3 -m src.experiments.NEC_evaluation.evaluation

scripts/T5_finetune_cl.sh

@@ -3,7 +3,7 @@
 #SBATCH --job-name=finetune_T5
 #SBATCH --output=logs/finetune_T5_%j.txt
 #SBATCH --ntasks=1
-#SBATCH --time=1:00:00
+#SBATCH --time=24:00:00
 #SBATCH --mem=8000
 #SBATCH --mail-type=ALL
 #SBATCH --mail-user=kupper@cl.uni-heidelberg.de
@@ -13,4 +13,4 @@
 #SBATCH --gres=gpu
 
 export PYTHONUNBUFFERED=1
-srun python3 -m src.experiments.finetune_T5
+srun python3 -m src.experiments.finetune_T5.finetune_T5

src/experiments/finetune_T5/results/plot.py

+import pandas as pd
+import matplotlib.pyplot as plt
+
+df = pd.read_csv("eval_loss.csv")
+
+print(df.head())
+
+plt.figure(figsize=(10,5))
+plt.plot(df["epoch"], df["eval_loss"])
+
+plt.xlabel("Epoch")
+plt.ylabel("Eval Loss")
+plt.title("T5 finetuning training curve")
+plt.legend()
+plt.grid(True)
+
+plt.savefig("eval_loss.pdf")