tags:
- generated_from_trainer
metrics:
- accuracy
- f1
widget:
- text: >-
Absolutely thrilled with my new wireless earbuds! The sound quality is
exceptional, and they stay securely in my ears during workouts. Plus, the
charging case is so convenient for on-the-go
- text: >-
Absolutely disappointed with this product! It arrived damaged and looked
nothing like the picture. Total waste of money.
- text: >-
This coffee maker has truly simplified my mornings. It brews quickly and
the programmable features allow me to wake up to the aroma of freshly
brewed coffee. Plus, the sleek design looks great on my countertop.
- text: >-
Do not buy this item! It broke within a week of use. Poor quality and not
worth the price at all.
- text: >-
I'm impressed with the durability of this laptop backpack. It comfortably
fits my 15-inch laptop, charger, and other essentials without feeling
bulky. The USB charging port is a lifesaver for staying connected on the
move.
- text: >-
Terrible experience with this purchase. The product had a weird smell and
caused skin irritation. Highly regret buying it.
- text: >-
As someone who loves to cook, this chef's knife is a game-changer. The
sharpblade effortlessly cuts through vegetables, meats, and herbs, making
prepwork a breeze. The ergonomic handle ensures comfort even during
longchopping sessions.
- text: >-
Extremely misleading description! The size was way smaller than
advertised, and the material felt cheap. Save your money and look
elsewhere.
- text: >-
This smart thermostat has made managing my home's temperature a breeze.
Theintuitive app allows me to adjust settings remotely, and the
energy-savingfeatures have noticeably reduced my utility bills.
Installation was also abreeze thanks to clear instructions.
- text: >-
Worst purchase ever! Not only did it not work as described, but the
customer service was also non-existent when I tried to resolve the issue.
Avoid at all costs.
model-index:
- name: gpt2-amazon-sentiment-classifier-V1.0
results: []
license: mit
datasets:
- McAuley-Lab/Amazon-Reviews-2023
language:
- en
library_name: transformers
gpt2-amazon-sentiment-classifier-V1.0
This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0320
- Accuracy: 0.9680
- F1: 0.9680
Model description
Hi! I'd be happy to share some insights about the Amazon Sentiment Analysis model I developed. The model is based on GPT-2, a transformer-based language model, which I fine-tuned using Amazon user reviews from 2023. The purpose of fine-tuning GPT-2 was to adapt it specifically for understanding and generating text related to sentiment analysis in Amazon reviews.
During the fine-tuning process, I trained the model to recognize different sentiments (positive, negative, neutral) by leveraging real user feedback. The fine-tuned GPT-2 model can now predict the sentiment of new reviews by generating relevant responses or categorizing them based on the emotions conveyed in the text.
You can use my model by using API
import transformers from transformers import pipeline sentiment_model = pipeline(model="ashok2216/gpt2-amazon-sentiment-classifier")
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure -->
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
Training results
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2