Code summary writing

chevron-icon
RFP Proposals
Top
chevron-icon
project-presentation-img
Expert Rating 2.6
Tofara Moyo
Project Owner

Code summary writing

Expert Rating

2.6

Overview

We plan to have a reinforcement algorithm learn to write comments on the code such as what each part does, its bigO evaluation as well as any other useful information that will help an llm evaluate the candidates using software that is meant to summarize documents for efficient search with search engines. This time the reward will be based on the success of the text in causing the llm to suggest certain children. So given a program the llm must write comments for it that best cause another llm to generate the most successful children. The first llms reward will be the success rate.

RFP Guidelines

Utilize LLMs for modeling within MOSES

Complete & Awarded
  • Type SingularityNET RFP
  • Total RFP Funding $150,000 USD
  • Proposals 10
  • Awarded Projects 1
author-img
SingularityNET
Oct. 9, 2024

This RFP invites proposals to explore the integration of LLMs into the MOSES evolutionary algorithm. Researchers can pursue one of several approaches, including generation modeling, fitness function learning, fitness estimation, investigation into domain-independent “cognitively motivated” fitness functions, or propose new innovative ways to leverage LLMs to enhance MOSES's capabilities within the OpenCog Hyperon framework.

Proposal Description

Open Source Licensing

GNU GPL - GNU General Public License

Links and references

https://github.com/LittleYUYU/CoaCor/tree/master

Proposal Video

Not Avaliable Yet

Check back later during the Feedback & Selection period for the RFP that is proposal is applied to.

  • Total Milestones

    3

  • Total Budget

    $72,000 USD

  • Last Updated

    5 Dec 2024

Milestone 1 - Created software

Description

Created the software needed to demonstrate the method

Deliverables

Fully produced software ready for training with no bugs.

Budget

$24,000 USD

Success Criterion

pipleine runs successfully to completion on sample data

Milestone 2 - Trained algorithm

Description

Trained the algorithm to produce code from scratch that it annotates and uses together with the annotations to predict the next generation of code.

Deliverables

Results of comparing standalone method of deploying moses to using our algorithm

Budget

$24,000 USD

Success Criterion

tabular data showing different metrics comparing our method to standalone moses

Milestone 3 - Integrated with Moses

Description

demonstrated the software as it will function integrated in with moses as well as tweaked it for better performance

Deliverables

software showing it intergrated with moses

Budget

$24,000 USD

Success Criterion

showing it operating under the framework of moses.

Join the Discussion (0)

Expert Ratings

Reviews & Ratings

Group Expert Rating (Final)

Overall

2.6

  • Feasibility 3.0
  • Desirabilty 3.0
  • Usefulness 2.0
  • Expert Review 1

    Overall

    2.0

    • Compliance with RFP requirements 2.0
    • Solution details and team expertise 2.0
    • Value for money 2.0
    This proposes a novel and potentially interesting algorithm but it's described loosely/vaguely and not easy to tell what's going on

    What is proposed is a creative interesting algorithm for doing search using LLMs and MOSES , however it's described in a confusing and incomplete way so that it's hard to tell for sure if it really. makes sense as an application of MOSES.

  • Expert Review 2

    Overall

    3.0

    • Compliance with RFP requirements 4.0
    • Solution details and team expertise 4.0
    • Value for money 3.0
    Short but sensible proposal to adapt their code retrieval and annotion system to program search.

    The authors wishes to adapt CoaCor, an invention of theirs, an LLM-based code retrieval and annotation technique that work in tandem, to guide search in program space. The proposal is succinct, just one paragraph as well as a link to their work, but it is sensible nonetheless.

  • Expert Review 3

    Overall

    3.0

    • Compliance with RFP requirements 3.0
    • Solution details and team expertise 3.0
    • Value for money 1.0

    Interesting concept “to have a reinforcement algorithm learn to write comments on” code, but does not provide nearly enough detail on steps involved.

feedback_icon