Stock Market QA Chatbot with Text-to-Text Transfer Transformer(T5)
π Overview
This repository hosts the quantized version of the T5 model fine-tuned for question-answer tasks related to stock market. The model has been trained on the stock_trading_QA dataset from Hugging Face. The model is quantized to Float16 (FP16) to optimize inference speed and efficiency while maintaining high performance.
π Model Details
- Model Architecture: t5-base
- Task: QA Chatbot for Stock Market
- Dataset: Hugging Face's
stock_trading_QA
- Quantization: Float16 (FP16) for optimized inference
- Fine-tuning Framework: Hugging Face Transformers
π Usage
Installation
pip install transformers torch
Loading the Model
from transformers import T5Tokenizer, T5ForConditionalGeneration
import torch
device = "cuda" if torch.cuda.is_available() else "cpu"
model_name = "AventIQ-AI/t5-stockmarket-qa-chatbot"
model = T5ForConditionalGeneration.from_pretrained(model_name).to(device)
tokenizer = T5Tokenizer.from_pretrained(model_name)
Question Answer Example
question = "How can I start investing in stocks?"
input_text = "question: " + question
input_ids = tokenizer.encode(input_text, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(input_ids, max_length=50)
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(f"Question: {question}")
print(f"Predicted Answer: {answer}")
π Evaluation Metric: BLEU Score
For question answer tasks, a high BLEU score indicates that the modelβs corrected sentences closely match human-annotated corrections.
Interpreting Our BLEU Score
Our model achieved a BLEU score of 0.7888, which indicates:
β
Good answer generating ability
β
Moderate sentence fluency
BLEU is computed by comparing the 1-gram, 2-gram, 3-gram, and 4-gram overlaps between the modelβs output and the reference sentence while applying a brevity penalty if the model generates shorter sentences.
BLEU Score Ranges for Chatbot
BLEU Score | Interpretation |
---|---|
0.8 - 1.0 | Near-perfect corrections, closely matching human annotations. |
0.7 - 0.8 | High-quality corrections, minor variations in phrasing. |
0.6 - 0.7 | Good corrections, but with some grammatical errors or missing words. |
0.5 - 0.6 | Decent corrections, noticeable mistakes, lacks fluency. |
Below 0.5 | Needs improvement, frequent incorrect corrections. |
β‘ Quantization Details
Post-training quantization was applied using PyTorch's built-in quantization framework. The model was quantized to Float16 (FP16) to reduce model size and improve inference efficiency while balancing accuracy.
π Repository Structure
.
βββ model/ # Contains the quantized model files
βββ tokenizer_config/ # Tokenizer configuration and vocabulary files
βββ model.safetensors/ # Quantized Model
βββ README.md # Model documentation
β οΈ Limitations
- The model may struggle with highly ambiguous sentences.
- Quantization may lead to slight degradation in accuracy compared to full-precision models.
- Performance may vary across different writing styles and sentence structures.
π€ Contributing
Contributions are welcome! Feel free to open an issue or submit a pull request if you have suggestions or improvements.
- Downloads last month
- 57