instructions

README.md

@@ -1,2 +1,13 @@
 # experiments_emolit
-Experiments for the emotions in literature paper.
+Train a model like in the emotions in literature paper.
+
+Data: https://zenodo.org/record/7883954
+
+Extract the data into the `data` dir so it looks like this: `data/emolit`.
+
+Install the requirements in `requirements.txt` (consider virtualenv).
+
+Change any parameters in the `soft_train.py` file (e.g. encoder model, batch
+size, number of epochs, ...).
+
+Run: `python soft_train.py`. This should train a model and save it to the `model` directory.

data/.gitignore

@@ -0,0 +1,4 @@
+# Ignore everything in this directory
+*
+# Except this file
+!.gitignore

model/.gitignore

@@ -0,0 +1,5 @@
+# Ignore everything in this directory
+*
+# Except this file
+!.gitignore
+

soft_train.py

@@ -17,28 +17,30 @@
 from mutils import compute_metrics_soft, write_metrics
 from loss import SoftBCETrainer
 
-logger = logging.getLogger(__name__)
-logging.basicConfig(
-    format="%(asctime)s - %(levelname)s - %(name)s -   %(message)s",
-    datefmt="%m/%d/%Y %H:%M:%S",
-    level=logging.INFO,
-)
 
+# Data
 TRAIN_FILE = "trn.tsv"
 VAL_FILE = "val.tsv"
-
 DATA_DIR = "./data/emolit"
 TST_FILE = "./data/emolit/gold.tsv"
 output_dir = "./model"
-NUM_EPOCHS = 10
-BS = 16
-LOAD_BEST = True
-
 
-SEQLEN = 64
+# Params
 MODEL_NAME = "roberta-large"
+SEQLEN = 64
+BS = 16
+NUM_EPOCHS = 10
+LOAD_BEST = True
 os.makedirs(output_dir, exist_ok=True)
 
+# logging
+logger = logging.getLogger(__name__)
+logging.basicConfig(
+    format="%(asctime)s - %(levelname)s - %(name)s -   %(message)s",
+    datefmt="%m/%d/%Y %H:%M:%S",
+    level=logging.INFO,
+)
+
 
 # Create the tokenizer
 os.environ["TOKENIZERS_PARALLELISM"] = "true"  # ensure parallel