==============================
-
technical problem: automate the evaluating phase in essay questions.
-
buissnes challenge: The goal is to reduce grading time and costs, enabling essays to be included in testing, a crucial measure of student learning often avoided due to grading challenges.
-
i have third approaches:
-
using a classical models with some heavey and smart feature engineering ( Done )
-
using LLMs, fine-tune in the dataset and some expermentations in the best and cost, tests in things like LoRa or full or some specific layers to be fine-tuned for fine-tune, like how to trate the problem in a classification way with 6 labels or regression task, and more ( working on it )
-
useing the feature from the first approache and using a LLM as a feature extraction then pass all to classic model ( i had made a notebook as a prove of concept )
-
- python 3.11 or later
- Download and install Anaconda from here
- Create a new environment using the following command:
$ conda create -n essay-scoring python=3.11
- activate the environment:
$ conda activate essay-scoring
export PS1="\[\033[01;32m\]\u@\h:\w\n\[\033[00m\]\$ "
$ pip install -r requirements.txt
$ cp .env.example .env
all notebooks was tested, runned and experimented in Kaggle.