PLEASE DO NOT MODIFY THIS PAGE!

#tag1, #tag2, ... (Add relevant tags to categorize your project.)

Name 1, Name 2, Name 3 (List all members in the project.)

Abstract

(~3–5 sentences, max 150 words)

Your abstract should:

• Motivate the problem (Why is this problem important?)
• Describe your goals (What are you trying to achieve?)
• Highlight key findings (What have you discovered so far?)

Since your project is still in progress, it’s fine if your findings are incomplete. However, try to frame your work in a concise, high-level, and convincing way.

What this project is about

(~300 words; guideline, not a hard limit--adjust as necessary)

Provide a clear, self-contained explanation of your project. While you can reuse parts of your project proposal (e.g., goal, task, data, methods, etc.), ensure the content is understandable by a broader audience.

• Explain the motivation for your project.
• Clearly define the task and scope.
• Outline the data and methodology at a high level.

Ideally, anyone in this class—or even those with limited exposure to NLP, outside of this class—would be able to understand your work.

Progress made so far

Describe what you've accomplished since the original proposal and remaining tasks.

Note that depending on your project type, you might want to modify the order of the contents (e.g., talk about metrics first instead of models, if your project is proposing a new metric). Please consider this template as a guideline, not a hard requirement.

Approach

(~300 words, with subsections as needed; guideline, not a hard limit--adjust as necessary)

This section details your approach to the problem.

• Main approach: Clearly describe the techniques and models you are using. Include key equations, figures, or diagrams if helpful (though time-consuming figures can be deferred to the final report).
• Baselines: Describe the baselines you are using for comparison. If it is a standard/common/well-established approach, you may reference existing work instead of providing extensive details.
• Novelty: If any part of your approach is original, explicitly highlight what is new and how it differs from prior work. Provide proper citations for models or techniques that are not yours.

Experiments

(~300–800 words, with subsections as needed; guideline, not a hard limit--adjust as necessary)

Your experiments should cover the following:

• Data
  • Describe the dataset(s) you are using.
  • Provide references and explain why the dataset is appropriate for your task.
  • Clearly describe the task associated with the dataset.
• Evaluation method
  • Define the evaluation metric(s) you are using.
  • Provide any necessary details to understand your evaluation.
• Experimental details
  • Explain how you ran your experiments, including model configurations, hyperparameters, training setup, and hardware specifications if relevant.
• Results
  • Report quantitative or qualitative results collected so far.
  • Use tables or plots to compare different results and baselines.
  • If your results are preliminary, discuss what you expect to improve or refine.

Remaining tasks

(~100–300 words)

• Outline the remaining work you plan to complete in the next one to two weeks.
• Prioritize tasks that are essential for your final presentation and report.
• If applicable, briefly mention any additional experiments, refinements, or analyses you intend to conduct.

(The content is based on Stanford CS224N’s Custom Final Project.)