initial commit

.gitattributes

@@ -32,3 +32,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+test.csv filter=lfs diff=lfs merge=lfs -text
+train.csv filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,38 @@
+import streamlit as st
+import torch
+import torch.nn.functional as F
+import pandas as pd
+from transformers import AutoTokenizer as AT, AutoModelForSequenceClassifcation as AMFSC
+
+# TODO choose model for use
+modName = "distilbert-base-uncased-finetuned-sst-2-english"
+mod = AMFSC.from_pretrained(modName)
+tokenizer = AT.from_pretrained(modName)
+
+# TODO set up training data
+train = pd.read_csv('./train.csv')
+train_texts = train['text'].values
+train_labels = train['label'].values
+
+# TODO set up test data
+test = pd.read_csv('./test.csv')
+test_texts = test['text'].values
+test_labels = test['label'].values
+
+# TODO working with the model
+batch = tokenizer(train_texts, padding=True, truncation=True, return_tensors="pt")
+
+# For App
+st.title("Finetuning Toxicity Model")
+with torch.no_grad():
+    outs = mod(**batch, labels=torch.tensor([1, 0]))
+    st.write(outs)
+    predicts = F.softmax(outs.logits, dim=1)
+    st.write(predicts)
+    labels = torch.argmax(predicts, dim=1)
+    st.write(labels)
+    labels = [mod.config.id2label[label_id] for label_id in labels.tolist()]
+    st.write(labels)
+
+# TODO deploy app to HuggingFace Streamlit Space
+# TODO add link to readme file

requirements.txt

@@ -0,0 +1,5 @@
+transformers
+torch
+torch.nn.functional
+pandas
+streamlit

test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2513ce4abb98c4d1d216e3ca0d4377d57589a0989aa8c06a840509a16c786e8
+size 60354593

train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd4084611bd27c939ba98e5e63bc3e5a2c1a4e99477dcba46c829e4c986c429d
+size 68802655