clean 2

app.py

@@ -1,87 +0,0 @@
-import gradio as gr
-import torch
-from transformers import AutoTokenizer, AutoModelForSequenceClassification
-from pathlib import Path
-
-# Load model - check both local and deployed paths
-MODEL_PATHS = ["outputs/final_model", "model", "/tmp/model"]
-
-tokenizer = None
-model = None
-MODEL_LOADED = False
-
-for model_path in MODEL_PATHS:
-    if Path(model_path).exists():
-        try:
-            tokenizer = AutoTokenizer.from_pretrained(model_path)
-            model = AutoModelForSequenceClassification.from_pretrained(model_path)
-            model.eval()
-            MODEL_LOADED = True
-            print(f"✓ Model loaded from: {model_path}")
-            break
-        except Exception as e:
-            print(f"Failed to load from {model_path}: {e}")
-
-if not MODEL_LOADED:
-    print("⚠ Model not found. Using demo mode.")
-
-def predict_genre(track_name):
-    """Predict genre for a track name"""
-    if not MODEL_LOADED:
-        return "Model not found. Please train first."
-    
-    if not track_name:
-        return "Please enter a track name"
-    
-    # Tokenize
-    inputs = tokenizer(track_name, return_tensors='pt', padding=True, truncation=True, max_length=256)
-    
-    # Predict
-    with torch.no_grad():
-        outputs = model(**inputs)
-        probs = torch.softmax(outputs.logits, dim=-1)
-        pred_id = torch.argmax(probs, dim=-1).item()
-        confidence = probs[0, pred_id].item()
-    
-    # Get label
-    pred_label = model.config.id2label.get(pred_id, f"Class_{pred_id}")
-    
-    return f"**Genre:** {pred_label}\n\n**Confidence:** {confidence:.2%}"
-
-# Create Gradio interface
-with gr.Blocks(title="Spotify Genre Classifier", theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 🎵 Spotify Genre Classifier")
-    gr.Markdown("Enter a song track name to predict its genre using a fine-tuned GPT-2 model.")
-    
-    with gr.Row():
-        with gr.Column():
-            track_input = gr.Textbox(
-                label="Track Name",
-                placeholder="e.g., Bohemian Rhapsody",
-                lines=1
-            )
-            predict_btn = gr.Button("🔮 Predict Genre", variant="primary")
-        
-        with gr.Column():
-            output = gr.Textbox(label="Prediction")
-    
-    # Examples
-    gr.Examples(
-        examples=[
-            "Bohemian Rhapsody",
-            "Shape of You",
-            "Old Town Road",
-            "Blinding Lights",
-            "Bad Guy",
-            "Stairway to Heaven",
-            "Smells Like Teen Spirit",
-            "Billie Jean",
-        ],
-        inputs=track_input
-    )
-    
-    predict_btn.click(fn=predict_genre, inputs=track_input, outputs=output)
-    track_input.submit(fn=predict_genre, inputs=track_input, outputs=output)
-
-if __name__ == "__main__":
-    demo.launch()

configs/config.yaml

@@ -1,19 +1,18 @@
-# Default Configuration
+# GPT-2 on Spotify Dataset
 
 model:
-  name: "bert-base-uncased"
+  name: "gpt2"
 
 dataset:
-  name: "spotify"
-  config: null
-  text_column: "text"
-  label_column: "label"
-  max_length: 512
+  name: "maharshipandya/spotify-tracks-dataset"
+  text_column: "track_name"
+  label_column: "track_genre"
+  max_length: 256
 
 training:
-  epochs: 3
-  batch_size: 32
-  learning_rate: 2e-5
+  epochs: 5
+  batch_size: 8
+  learning_rate: 5e-5
   weight_decay: 0.01
   warmup_ratio: 0.1
 

configs/gpt2_spotify.yaml

@@ -1,26 +0,0 @@
-# GPT-2 on Spotify Dataset
-
-model:
-  name: "gpt2"
-
-dataset:
-  name: "maharshipandya/spotify-tracks-dataset"
-  text_column: "track_name"
-  label_column: "track_genre"
-  max_length: 256
-
-training:
-  epochs: 5
-  batch_size: 8
-  learning_rate: 5e-5
-
-hardware:
-  mixed_precision: "fp16"
-
-output:
-  dir: "./outputs"
-
-evaluation:
-  metrics:
-    - "accuracy"
-    - "f1"

configs/spotify.yaml

@@ -1,27 +0,0 @@
-# Spotify Dataset Configuration
-
-model:
-  name: "bert-base-uncased"
-
-dataset:
-  name: "maharshipandya/spotify-tracks-dataset"
-  text_column: "track_name"
-  label_column: "track_genre"
-  max_length: 256
-
-training:
-  epochs: 5
-  batch_size: 16
-  learning_rate: 3e-5
-
-hardware:
-  mixed_precision: "fp16"
-
-output:
-  dir: "./outputs"
-  save_strategy: "epoch"
-
-evaluation:
-  metrics:
-    - "accuracy"
-    - "f1"

run.sh

@@ -1,61 +1,33 @@
 #!/bin/bash
-#
-# HF Training Pipeline - Run Script
-# Usage: ./run.sh [config_name]
-#        ./run.sh test [model_path]
-#
-# Examples:
-#   ./run.sh              # Run with default config
-#   ./run.sh spotify      # Run with Spotify config
-#   ./run.sh gpt2_spotify # Run with GPT-2 model
-#   ./run.sh test         # Test trained model
-#
+# Usage: ./run.sh
 
 set -e
 
-# Colors
-RED='\033[0;31m'
 GREEN='\033[0;32m'
 YELLOW='\033[1;33m'
 BLUE='\033[0;34m'
 NC='\033[0m'
 
-SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
-cd "$SCRIPT_DIR"
+cd "$(dirname "${BASH_SOURCE[0]}")"
 
-# Handle test command
-if [ "$1" = "test" ]; then
-    MODEL_PATH="${2:-outputs/final_model}"
-    echo -e "${BLUE}========================================${NC}"
-    echo -e "${BLUE}  Testing Model${NC}"
-    echo -e "${BLUE}  Path: ${YELLOW}${MODEL_PATH}${NC}"
-    echo -e "${BLUE}========================================${NC}"
-    echo ""
-    python3 test_model.py "$MODEL_PATH"
-    exit 0
+# Install uv if not exists
+if ! command -v uv &> /dev/null; then
+    echo -e "${YELLOW}Installing uv...${NC}"
+    curl -LsSf https://astral.sh/uv/install.sh | sh
 fi
 
-CONFIG_NAME="${1:-spotify}"
+# Sync dependencies (creates .venv if needed)
+echo -e "${GREEN}✓ Syncing dependencies with uv...${NC}"
+uv sync
 
-echo -e "${BLUE}========================================${NC}"
-echo -e "${BLUE}  Hugging Face Training Pipeline${NC}"
-echo -e "${BLUE}========================================${NC}"
-echo ""
+# Activate venv
+source .venv/bin/activate
 
-# Activate venv if exists
-if [ -f ".venv/bin/activate" ]; then
-    echo -e "${GREEN}✓ Activating virtual environment...${NC}"
-    source .venv/bin/activate
-fi
-
-# Check Python
-if ! command -v python3 &> /dev/null; then
-    echo -e "${RED}❌ Python3 not found${NC}"
-    exit 1
+# Check .env
+if [ ! -f ".env" ]; then
+    cp .env.example .env
 fi
 
-echo -e "${GREEN}✓ Python found: $(python3 --version)${NC}"
-
 # Check GPU
 echo ""
 echo -e "${YELLOW}Checking GPU availability...${NC}"
@@ -69,38 +41,19 @@ else:
     print('⚠ No CUDA available - training will use CPU (slower)')
 "
 
-# Check .env
-echo ""
-if [ -f ".env" ]; then
-    echo -e "${GREEN}✓ .env file found${NC}"
-else
-    echo -e "${YELLOW}⚠ .env not found. Creating...${NC}"
-    cp .env.example .env
-fi
-
-# Run training
-echo ""
 echo -e "${BLUE}========================================${NC}"
-echo -e "${BLUE}  Starting Training${NC}"
-echo -e "${BLUE}  Config: ${YELLOW}${CONFIG_NAME}${NC}"
+echo -e "${BLUE}  Training GPT-2 on Spotify Dataset${NC}"
 echo -e "${BLUE}========================================${NC}"
-echo ""
 
-python3 src/training_pipeline.py "$CONFIG_NAME"
+python3 src/training_pipeline.py config
 
 echo ""
-echo -e "${GREEN}========================================${NC}"
-echo -e "${GREEN}  Training Complete!${NC}"
-echo -e "${GREEN}========================================${NC}"
-echo ""
+echo -e "${GREEN}✓ Training Complete!${NC}"
 echo -e "Model saved to: ${YELLOW}outputs/final_model${NC}"
 echo ""
 
-# Ask to run test
-echo -e "${YELLOW}Would you like to test the model? (y/n)${NC}"
-read -r -n 1 -s response
-echo ""
-if [[ $response =~ ^[Yy]$ ]]; then
-    echo ""
-    python3 test_model.py outputs/final_model
-fi
+echo -e "${BLUE}========================================${NC}"
+echo -e "${BLUE}  Testing Model${NC}"
+echo -e "${BLUE}========================================${NC}"
+
+python3 test_model.py outputs/final_model