Upload 6 files

BertEmotionClassifier.py

@@ -0,0 +1,22 @@
+# ---------------------------
+# Model
+# ---------------------------
+import torch.nn as nn
+from transformers import AutoModel
+
+class BertEmotionClassifier(nn.Module):
+    def __init__(self, model_name: str = "roberta-base", num_labels: int = 5, dropout: float = 0.3):
+        super().__init__()
+        self.encoder = AutoModel.from_pretrained(model_name)
+        self.dropout = nn.Dropout(dropout)
+        self.classifier = nn.Linear(self.encoder.config.hidden_size, 128)
+        self.classifier1 = nn.Linear(128, num_labels)
+
+    def forward(self, input_ids, attention_mask):
+        out = self.encoder(input_ids=input_ids, attention_mask=attention_mask)
+        cls = out.last_hidden_state[:, 0, :]
+        cls = self.classifier(cls)
+        cls = self.dropout(cls)
+        logits = self.classifier1(cls)
+        return logits
+    

README.md

@@ -1,19 +1,33 @@
----
-title: Dl Project
-emoji: 🚀
-colorFrom: red
-colorTo: red
-sdk: docker
-app_port: 8501
-tags:
-- streamlit
-pinned: false
-short_description: This is my DL-genAI project sep-2025
----
-
-# Welcome to Streamlit!
-
-Edit `/src/streamlit_app.py` to customize this app to your heart's desire. :heart:
-
-If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
-forums](https://discuss.streamlit.io).
+# 🎭 Emotion Classifier — RoBERTa Based
+A deep-learning model that classifies text into **five emotion categories**:
+**anger, fear, joy, sadness, surprise**.
+
+Built with **PyTorch, RoBERTa transformer, Streamlit UI** and deployed on **Hugging Face Spaces**.
+
+---
+
+## 🧠 Model Details
+| Component | Description |
+|-----------|-------------|
+| Base model | `roberta-base` |
+| Task | Single-label Emotion Classification |
+| Input | Raw text |
+| Output | Softmax probability distribution (5 emotions) |
+| Framework | PyTorch + Transformers |
+
+### Load Model from Hugging Face
+```python
+from huggingface_hub import hf_hub_download
+import torch
+from BertEmotionClassifier import BertEmotionClassifier
+
+model_path = hf_hub_download(repo_id="aadhi3/RoBert_Model", filename="model.pth")
+
+from transformers import AutoTokenizer
+tokenizer = AutoTokenizer.from_pretrained("roberta-base")
+
+model = BertEmotionClassifier(model_name="roberta-base", num_labels=5)
+state_dict = torch.load(model_path, map_location="cpu")
+model.load_state_dict(state_dict)
+model.eval()
+```

app.py

@@ -0,0 +1,153 @@
+import streamlit as st
+import torch
+import pandas as pd
+import matplotlib.pyplot as plt
+from BertEmotionClassifier import BertEmotionClassifier
+from transformers import AutoTokenizer
+from huggingface_hub import hf_hub_download
+import torch.nn.functional as F
+
+# ----------------------------
+# Config
+# ----------------------------
+EMOTIONS = ['anger', 'fear', 'joy', 'sadness', 'surprise']
+MODEL_NAME = "roberta-base"
+
+st.set_page_config(page_title="Emotion Classifier", page_icon="🎭", layout="wide")
+
+# ----------------------------
+# Custom Styling
+# ----------------------------
+st.markdown("""
+    <style>
+        .emotion-card {
+            padding: 15px;
+            border-radius: 18px;
+            text-align: center;
+            font-size: 18px;
+            font-weight: 600;
+            color: white;
+            margin-bottom: 10px;
+            box-shadow: 0px 0px 10px rgba(0,0,0,0.4);
+        }
+        .dominant {
+            background: linear-gradient(135deg, #00c6ff, #0072ff);
+        }
+        .sub {
+            background: linear-gradient(135deg, #434343, #000000);
+        }
+        .main-container {
+            max-width: 900px;
+            margin: auto;
+            padding-top: 30px;
+        }
+    </style>
+""", unsafe_allow_html=True)
+
+# ----------------------------
+# Load Model
+# ----------------------------
+@st.cache_resource
+def load_model():
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+
+    model = BertEmotionClassifier(model_name=MODEL_NAME, num_labels=len(EMOTIONS))
+
+    # Download model weights from HuggingFace Hub
+    model_path = hf_hub_download(repo_id="aadhi3/RoBert_Model", filename="model.pth")
+
+    # Load state dict from downloaded path
+    state_dict = torch.load(model_path, map_location="cpu")
+    state_dict = {k.replace('module.', ''): v for k, v in state_dict.items()}
+    model.load_state_dict(state_dict)
+    model.eval()
+
+    return tokenizer, model
+
+tokenizer, model = load_model()
+
+# ----------------------------
+# Prediction Function
+# ----------------------------
+def predict_emotions(text: str):
+    encoding = tokenizer(
+        text,
+        add_special_tokens=True,
+        max_length=256,
+        padding='max_length',
+        truncation=True,
+        return_tensors='pt'
+    )
+
+    with torch.no_grad():
+        logits = model(**encoding)
+        probs = F.softmax(logits, dim=-1)[0].cpu()
+
+    return {emo: round(float(probs[i]), 4) for i, emo in enumerate(EMOTIONS)}
+
+# ----------------------------
+# UI Layout
+# ----------------------------
+st.title("🎭 Emotion Detection AI")
+st.markdown("### Understand emotions in text using **AI-driven emotion analysis**")
+st.write("")  # spacing
+
+input_text = st.text_area(
+    "Enter your text here:",
+    placeholder="Type something like: 'I am extremely happy today!' 😄",
+    height=150
+)
+
+predict_btn = st.button("🔮 Analyze Emotions", use_container_width=True)
+
+if predict_btn and input_text.strip() != "":
+    with st.spinner("AI thinking..."):
+        results = predict_emotions(input_text.strip())
+
+        df = pd.DataFrame(results.items(), columns=["Emotion", "Probability"])
+        dominant = df.iloc[df["Probability"].idxmax()]["Emotion"]
+
+        st.markdown(f"### 🎯 Dominant Emotion: **{dominant.upper()}**")
+
+        col_chart, col_cards = st.columns([1.2, 1])
+
+        # ---------------------------
+        # Matplotlib Bar Chart
+        # ---------------------------
+        with col_chart:
+            fig, ax = plt.subplots()
+            ax.bar(df["Emotion"], df["Probability"])
+            ax.set_ylim(0, 1)
+            ax.set_ylabel("Probability Score")
+            ax.set_title("Emotion Prediction Distribution")
+            st.pyplot(fig)
+
+        # ---------------------------
+        # Emotion Cards
+        # ---------------------------
+        with col_cards:
+            st.markdown("### 📊 Emotion Strength")
+            for emo, val in results.items():
+                style = "dominant" if emo == dominant else "sub"
+                st.markdown(
+                    f"<div class='emotion-card {style}'>{emo.capitalize()} — {val}</div>",
+                    unsafe_allow_html=True
+                )
+
+        # Save Prediction History
+        row = {"text": input_text, **results}
+        try:
+            old_df = pd.read_csv("./predictions.csv")
+        except FileNotFoundError:
+            old_df = pd.DataFrame(columns=["text"] + EMOTIONS)
+
+        new_df = pd.concat([old_df, pd.DataFrame([row])], ignore_index=True)
+        new_df.to_csv("./predictions.csv", index=False)
+
+        with st.expander("📂 View Recent Predictions"):
+            st.dataframe(new_df.tail(10), use_container_width=True)
+
+        st.success("Result saved successfully! ✨")
+
+st.markdown("---")
+st.caption("Built with ❤️ using Streamlit & PyTorch — deployed on HuggingFace Spaces")

predictions.csv

@@ -0,0 +1,3 @@
+text,anger,fear,joy,sadness,surprise
+Hi Today is a lucky day,0.0,0.0,1.0,0.0,0.0
+What is the day of the week,0.0397,0.025,0.0156,0.0431,0.8766

requirements.txt

@@ -1,3 +1,6 @@
-altair
-pandas
-streamlit
+streamlit
+torch
+transformers
+huggingface_hub
+pandas
+matplotlib