This paper introduces RetLLM-E, a method that enhances Large Language Models (LLMs) for answering student questions on large forums like Piazza and EdSTEM. RetLLM-E combines text-retrieval and prompting techniques to improve response quality. The system retrieves relevant context from past Q&As and course materials, then uses this context to generate precise answers. Evaluation shows that RetLLM-E provides higher-quality responses compared to LLMs without context or those using only retrieval-based methods.
-
Mitra, Chancharik, et al. "RetLLM-E: Retrieval-Prompt Strategy for Question-Answering on Student Discussion Forums." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 38. No. 21. 2024.
-
Members
- JoelR
- Chris Anderson
- Myr
- Live Games
- IC Essentials
- Nathan Explosion
- Square Wheels
- bernhara
- Auto Evoke
- opentype
- ReyDev
- send2yoni
- Brian
- A Zayed
- Adriano Faria
- terabyte
- Dilip
- ZLTRGO
- adik
- master963
- DawPi
- eivindsimensen
- envy
- onlyME
- V0RT3X
- GazzaGarratt
- Analog
- Voyage
- Paul Kaiser
- Como
- N700
- Paul
- TracyIsland
- Andy Y
- Omar Barbeytia carretero
- JoeyM
- Ryancoolround
- rainx
- YourSharona
- Kentraiyle Robinson
- MichaelR
- Edward Ellas
- IPS THEME
- aXenDev
- PrettyPixels
- Denis Dyack
- Labis
- DursunKaptan
- MissB
- TheLlamaman
- aLEX49566
- Codepixel
- alsl sndnxnx
- burnyourfeelings
- isvans
- Marius
- Matt
- Thomas Taschler
- Surpac
- JoshB
- Ioannis D
- abobader
- Richard Arch
- bdmusic 24
- Majster87
- TomCat
- Pmw
- Torgeir Rui
- Kammer et
- Nicolas PC
- XwReK
- Claudia999
- Kirill Gromov
- Synergy
- bing11
- Marcin Martyniak
- ArashDev
- ali hagi
- StevenM
- NewVicious
- lukash
- Andhrafriends Admin
- Daffy
- hyprem
- GuitarGathering
- Tripp
- Askancy
- MLK
- Jelly Belly
- eveneme eveneme
- Nomad
- Morphe
- lordi
- shahed
- John Horton
- PayMap
- Serval
- Nomer3
- Dennis Maidon
- Zennuie