first commit

app.py

@@ -0,0 +1,26 @@
+import streamlit as st
+import torch
+from model import (DistilBertTokenizerFast, TARGET_IND2LABEL, DistilBERTClassifier,
+                    load_model, load_tokenizer,
+                    get_preds, get_verdict)
+
+
+st.markdown("## Hello, my dear friend!")
+st.markdown("### This service can classify article topic by it's title and abstract")
+st.markdown("##### You can specify only title or only abstract, just leave the field empty, but it can work worse in that case.")
+
+title = st.text_area("Title:")
+abstract = st.text_area("Abstract:")
+
+device = 'cpu'
+tokenizer = load_tokenizer()
+model = load_model(device)
+
+text = title + abstract
+if (not text):
+    verdict = "Both fields are empty"
+else:
+    verdict = get_verdict(get_preds(text, model, tokenizer, device))
+
+st.markdown("#### Verdict:")
+st.markdown(verdict)

model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0883bdb976c57c2ca825375fd530c836f5d231ff323fb7f23f6fc14189db57a
+size 263204889

model.py

@@ -0,0 +1,65 @@
+import torch
+import torch.nn as nn
+import numpy as np
+import streamlit as st
+from transformers import DistilBertModel, DistilBertTokenizerFast
+
+
+TARGET_IND2LABEL = {
+    0: 'Computer Science',
+    1: 'Economics',
+    2: 'Electrical Engineering and Systems Science',
+    3: 'Mathematics',
+    4: 'Physics',
+    5: 'Quantitative Biology',
+    6: 'Quantitative Finance',
+    7: 'Statistics',
+}
+
+class DistilBERTClassifier(nn.Module):
+    def __init__(self, num_classes=8):
+        super().__init__()
+        self.encoder = DistilBertModel.from_pretrained("distilbert-base-cased")
+        self.pre_classifier = nn.Linear(768, 768)
+        self.gelu = nn.GELU()
+        self.dropout = nn.Dropout(0.1)
+        self.classifier = nn.Linear(768, num_classes)
+
+    def forward(self, input_ids, attention_mask, labels):
+        output = self.encoder(input_ids=input_ids, attention_mask=attention_mask)
+        hidden_state = output[0]
+        pooler = hidden_state[:, 0]
+        pooler = self.dropout(self.gelu(self.pre_classifier(pooler)))
+        preds = self.classifier(pooler)
+        return preds
+    
+@st.cache_resource
+def load_tokenizer():
+    return DistilBertTokenizerFast.from_pretrained('distilbert-base-cased')
+
+@st.cache_resource
+def load_model(device):
+    model = torch.load('model.pt', map_location=torch.device('cpu')).to(device)
+    model.eval()
+    return model
+
+def get_verdict(preds):
+    inds = np.argsort(preds)[::-1]
+    sum_prob = 0.0
+    verdict = []
+    for ind in inds:
+        prob = preds[ind]
+        sum_prob += prob
+        verdict.append(f"{TARGET_IND2LABEL[ind]}: {prob}")
+        if (sum_prob >= 0.95):
+            break
+    return "\n\n".join(verdict)
+
+def get_preds(text, model, tokenizer, device):
+    tokens = tokenizer(text, padding=True, truncation=True, return_tensors='pt')
+    tokens['input_ids'] = tokens['input_ids'].to(device)
+    tokens['attention_mask'] = tokens['attention_mask'].to(device)
+    tokens['labels'] = None # made for training convinience
+    with torch.no_grad():
+        preds = torch.softmax(model(**tokens)[0], 0).cpu().numpy()
+    return preds

requirements.txt

@@ -0,0 +1,3 @@
+numpy
+torch
+transformers