TSAI Group Assignment

Group Members:

1. Arjun Gupta
2. Himanshu
3. Aeshna Singh
4. Palash Baranwal

SESSION 5 - TorchText & Advanced Concepts

ASSIGNMENT

1. Look at the given code: https://colab.research.google.com/drive/1z2N6ipI6n6t1ivprJtugDeqKrCM2NPFy?usp=sharing
2. Pick any 2 datasets (except AG_NEWS) from torchtext.datasets and train your model on them achieving 50% more accuracy than random prediction.
3. Upload to Github with a proper readme file describing your datasets, and showing your logs as well.

---

SogouNews DATASET

The labels includes:

• 0 : Sports
• 1 : Finance
• 2 : Entertainment
• 3 : Automobile - 4 : Technology

Number of lines per split:

• train: 450000
• test: 60000

Number of classes: 5

Examples:

• train_dataset, test_dataset = torchtext.datasets.SogouNews(ngrams=3)

ACCURACY: 94.6%

---

YelpReviewPolarity DATASET

The labels includes:

• 0 : Negative polarity.
• 1 : Positive polarity.

Number of lines per split:

• train: 560000
• test: 38000

Number of classes: 2

Examples:

• train_dataset, test_dataset = torchtext.datasets.YelpReviewPolarity(ngrams=3)

ACCURACY: 91.0%

---

LOGS SCREENSHOTS

SogouNews

YelpReviewPolarity

---