This model is a fine-tuned ProphetNet designed for predicting stock market trends. It takes historical stock data as input and generates forecasts for future price movements, including upward, downward, or stable trends. Trained on extensive financial datasets, it provides probabilistic predictions to aid in investment decisions.
The model utilizes the ProphetNet architecture, which is an encoder-decoder model optimized for sequence-to-sequence tasks. It features 12 encoder and 12 decoder layers, each with a hidden size of 1024. The model incorporates n-gram attention mechanisms for improved long-range dependency handling in time-series data.
It can be used with the Hugging Face Transformers library. Example usage:
1from transformers import pipeline
2
3predictor = pipeline("text2text-generation", model="your-username/stock-market-predictor")
4result = predictor("Historical data: AAPL prices from 2020-2025.")
5print(result)
