metadata
language: en
tags:
- sentiment-analysis
- flan-t5
- text-classification
license: apache-2.0
datasets:
- imdb
FLAN-T5 Small - Sentiment Analysis
Fine-tuned version of google/flan-t5-small for sentiment analysis on IMDB reviews.
Model Details
- Base Model: google/flan-t5-small
- Task: Binary sentiment classification (positive/negative)
- Dataset: IMDB movie reviews (300 training samples)
- Accuracy: 85.00%
Usage
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
tokenizer = AutoTokenizer.from_pretrained("usef310/flan-t5-small-sentiment")
model = AutoModelForSeq2SeqLM.from_pretrained("usef310/flan-t5-small-sentiment")
text = "This movie was amazing!"
inputs = tokenizer("sentiment: " + text, return_tensors="pt")
outputs = model.generate(**inputs)
prediction = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(prediction) # Output: positive or negative
Training Details
- Epochs: 3
- Batch size: 4 (with gradient accumulation)
- Learning rate: 5e-5
- Optimizer: AdamW