Skip to main contentSkip to navigation
MachineHack Gen AI Logo
oneAPI Hackathon: The LLM Challenge Logo

oneAPIHackathon:The
LLMChallenge

Expired
Start: September 14, 2023Ends: October 21, 2023
Participants
1,092
Time Left
Ended
Subs/day
3
Challenge Overview

About oneAPI


oneAPI is an open, cross-industry, standards-based, unified, multi-architecture, multi-vendor programming model that delivers a common developer experience across accelerator architectures – for faster application performance, more productivity, and greater innovation. The oneAPI initiative encourages collaboration on the oneAPI specification and compatible oneAPI implementations across the ecosystem.

 

1. About the Hackathon

While text-based tasks are present everywhere, one of the most compelling objectives is the development of a question-answering system tailored to textual data. Imagine a system capable of sifting through vast datasets, identifying 'span_start' and 'span_end' positions within the 'Story' text, extracting the relevant 'span_text,' and generating responses that align perfectly with the provided 'Answer' for each question.


Problem statement


This is precisely the challenge that awaits participants in the oneAPI Hackathon: The LLM Challenge. The goal is to use LLMs, leveraging their text understanding and generation capabilities to put up accurate, context-aware answers. Moreover, it needs to be done with a level of accuracy that pushes the boundaries of what language models can do. Beyond just creating a robust LLM, participants will be challenged to build an application that utilises their language model effectively and pen down their insights in a blog post.
 

  • The Contest begins on Sep 14th, 2023 6:00 PM IST and ends on Oct 29th, 2023 6:00 PM IST.
  • Hands-on workshop date: Sep 21st, 2023 [Mandatory for participants to join - ** more details below]

 

  • Phase #1:  Metric-based evaluation - Sep 14th -  Oct 21st, 2023 IST
  • Phase #2: Application building + GitHub/Blog writing - Oct 15th - Oct 29th, 2023 IST[The GitHub repository must be clonable and executable on the Intel® Developer Cloud]
  • Announcement of winners & Prize distribution: 15th November.

 

2. Register on the Intel® Developer Cloud platform

*** NOTE: Signing up and utilizing the Intel® Developer Cloud is mandatory for joining this hackathon.

For accessing and training the model, participants must follow these steps:

 

3. Hands-on workshop on the Intel® Developer Cloud platform: 

Topic: Intel® AI Software Optimization Using Intel® AI Analytics Toolkit with Demo on Intel® Developer Cloud

Registration Link: https://us06web.zoom.us/webinar/register/WN_y6XjTZWhSc2c4T9ZUXYiQw#/registration

** NOTE:  Mandatory for participants to join the workshop
**NOTE: The Q&A will happen on Discord: https://discord.com/invite/ycwqTP6

Training resources- workshops on Intel® Developer Cloud

 

4. Sample Resources for submission: 

Utilize the Intel® oneAPI Toolkits available at: https://www.intel.com/content/www/us/en/developer/tools/oneapi/toolkits.html#gs.4ptqsw.

Explore the AI Reference Kits provided at: https://www.intel.com/content/www/us/en/developer/topic-technology/artificial-intelligence/reference-kit.html.

 

5. After training the model on Intel® Developer Cloud, participants should submit their final predictions in a file named "submission.csv" along with the Intel® Developer Cloud screenshot on the MachineHack hackathon page. This submission will contribute to the final leaderboard rankings.

** NOTE: In order to improve the ranking, participants need to:

Write a comprehensive blog post on the usage of the various components from the Intel® AI Analytics Toolkit and Intel® Developer Cloud & a GitHub repository detailing their approach

Develop an AI application using the Intel® provided Language Models (LLMs) tailored to the dataset.

The weightage of the final score will be judged based on 

  • 40% on #Phase 1
    • Score on MachineHack Leaderboard
  • 60% on #Phase 2 - Application building [Use open-source large language models to build a custom chatbot tailored to the dataset] + GitHub + Blog writing.The GitHub repository must be clonable and executable on the Intel® Developer Cloud.
    • Usage of Intel oneAPI optimizations in the final application
    • Re-usability and ease of deploying the application on Intel® Developer Cloud
    • Structure of Github Repo and ReadMe description
    • Short blog highlighting the features of the application

 

6. Training Resources 

Problem Statement

This challenge focuses on building advanced machine learning models to solve real-world problems. Participants will work with carefully curated datasets and compete to achieve the best performance metrics.

Target Column: Answer
Metric: accuracy_score
Level: Advanced
Submissions: 3/day
Top Submissions

No leaderboard data available

Check back later for updates

oneAPI Hackathon: The LLM Challenge

View competition details

Similar Challenges

Discover similar AI and data science competitions

No sponsored hackathons available at the moment.

Never Miss a Hackathon

Get notified about new AI hackathons, data science competitions, and exclusive opportunities. Join 50,000+ developers staying ahead of the curve.

No spam, unsubscribe at any time. We respect your privacy.