|
|
--- |
|
|
datasets: |
|
|
- yelp_review_full |
|
|
language: |
|
|
- en |
|
|
metrics: |
|
|
- accuracy |
|
|
- code_eval |
|
|
pipeline_tag: text-classification |
|
|
--- |
|
|
# Model Card for SentimentTensor |
|
|
|
|
|
This modelcard provides details about the SentimentTensor model, developed by Saish Shinde, for sentiment analysis using LSTM architecture. |
|
|
|
|
|
## Model Details |
|
|
|
|
|
### Model Description |
|
|
|
|
|
The SentimentTensor model is a deep learning model based on LSTM architecture, developed by Saish Shinde, for sentiment analysis tasks. It achieves an accuracy of 81% on standard evaluation datasets. The model is designed to classify text data into three categories: negative, neutral, and positive sentiments. |
|
|
|
|
|
- **Developed by:** Saish Shinde |
|
|
- **Model type:** LSTM-based Sequence Classification |
|
|
- **Language(s) (NLP):** English |
|
|
- **License:** No specific license |
|
|
|
|
|
|
|
|
|
|
|
# Dataset Used |
|
|
|
|
|
yelp dataset with 4.04GB compressed,8.65GB uncompressed data |
|
|
|
|
|
## Uses |
|
|
|
|
|
### Direct Use |
|
|
|
|
|
The SentimentTensor model can be directly used for sentiment analysis tasks without fine-tuning. |
|
|
|
|
|
### Downstream Use |
|
|
|
|
|
This model can be fine-tuned for specific domains or integrated into larger NLP applications. |
|
|
|
|
|
### Out-of-Scope Use |
|
|
|
|
|
The model may not perform well on highly specialized or domain-specific text data. |
|
|
|
|
|
## Bias, Risks, and Limitations |
|
|
|
|
|
The SentimentTensor model, like any LSTM-based model, may have biases and limitations inherent in its training data and architecture. It might sometimes struggle with capturing long-range dependencies or understanding context in complex sentences, also it emphasizes less on neutral sentiment |
|
|
|
|
|
### Recommendations |
|
|
|
|
|
Users should be aware of potential biases and limitations and evaluate results accordingly. |
|
|
|
|
|
## How to Get Started with the Model |
|
|
|
|
|
### Loading the Model |
|
|
|
|
|
You can load the SentimentTensor model using the Hugging Face library: |
|
|
|
|
|
# python Code: |
|
|
from transformers import AutoModelForSequenceClassification, AutoTokenizer |
|
|
|
|
|
# Load the model and tokenizer |
|
|
model = AutoModelForSequenceClassification.from_pretrained("your-model-name") |
|
|
tokenizer = AutoTokenizer.from_pretrained("your-tokenizer-name") |
|
|
|
|
|
# Tokenization |
|
|
text = "Your text data here" |
|
|
tokenized_input = tokenizer(text, return_tensors="pt") |
|
|
|
|
|
# Sentiment Analysis |
|
|
#Forward pass through the model |
|
|
outputs = model(**tokenized_input) |
|
|
|
|
|
#Get predicted sentiment label |
|
|
predicted_label = outputs.logits.argmax().item() |
|
|
|
|
|
# Example Usage |
|
|
```python |
|
|
|
|
|
from transformers import AutoModelForSequenceClassification, AutoTokenizer |
|
|
|
|
|
# Load the model and tokenizer |
|
|
model = AutoModelForSequenceClassification.from_pretrained("saishshinde15/SentimentTensor") |
|
|
tokenizer = AutoTokenizer.from_pretrained("saishshinde15/SentimentTensor") |
|
|
|
|
|
# Tokenize text data |
|
|
text = "This is a great movie!" |
|
|
tokenized_input = tokenizer(text, return_tensors="pt") |
|
|
|
|
|
# Perform sentiment analysis |
|
|
outputs = model(**tokenized_input) |
|
|
predicted_label = outputs.logits.argmax().item() |
|
|
|
|
|
# Print predicted sentiment |
|
|
sentiment_labels = ["negative", "neutral", "positive"] |
|
|
print(f"Predicted Sentiment: {sentiment_labels[predicted_label]}") |
|
|
|
|
|
|
|
|
``` |
|
|
# Model Architecture and Objective |
|
|
|
|
|
The SentimentTensor model is based on LSTM architecture, which is well-suited for sequence classification tasks like sentiment analysis. It uses long short-term memory cells to capture dependencies in sequential data. |
|
|
|
|
|
# Model Card Authors |
|
|
Saish Shinde |
