metadata

tags:
  - flan-t5
  - essay-scoring
  - education
  - grading
  - gpt-evaluation
license: apache-2.0
language:
  - en
pipeline_tag: text2text-generation

flan-t5-base-pt2 — Essay Trait Scorer

This model is a fine-tuned version of google/flan-t5-base designed for trait-based essay scoring. It was trained on Sets 7 & 8 of the ASAP AES dataset, extended with floating-point scoring and GPT validation.

Each input includes:

a trait (e.g., Ideas, Organization),
the relevant rubric definition,
and the student's essay.

The model outputs a trait-specific score between 0.0 and 3.0, which can be scaled to a final grade out of 100.

🧠 Scoring is optionally cross-validated with OpenAI’s GPT for more reliable results.

Usage

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("srutiii/flan-t5-base-pt2")
model = AutoModelForSeq2SeqLM.from_pretrained("srutiii/flan-t5-base-pt2")