surabhi08
/

dataset-genome-agriculture-benchmark-model

scientific-reasoning

Model card Files Files and versions

Dataset Genome Agriculture Benchmark Model

Model Overview

This repository contains a PEFT (LoRA) fine-tuned adapter for GPT-OSS-20B trained using the AutoScientist pipeline from Adaption Labs.

The model is designed for scientific reasoning tasks in agriculture, including:

Observation analysis
Research gap identification
Hypothesis generation
Experimental design
Scientific conclusion generation

Base Model

Base Model: togethercomputer/gpt-oss-20b-bf16
Fine-tuning Method: PEFT (LoRA)
Training Framework: AutoScientist (Adaption Labs)

Training Dataset

The model was trained using the Dataset Genome – Agriculture Benchmark dataset.

Dataset Repository:

https://huggingface.co/datasets/surabhi08/dataset-genome-agriculture-benchmark

Dataset Characteristics:

Domain: Agriculture
Records: 20
Language: English
Format: JSONL
License: MIT

Intended Use

This model is intended for research and educational purposes involving:

Scientific reasoning
Agricultural AI
Dataset benchmarking
Hypothesis generation
Experimental planning

Repository

GitHub:

https://github.com/HackIndiaXYZ/adaption-autoscientist-challenge-part-2-60000-prize-pool-surabhicodes

Author

Surabhi M. S.

Created as part of the HackIndia 2026 AutoScientist Challenge.

License

MIT License

Downloads last month: 4

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for surabhi08/dataset-genome-agriculture-benchmark-model

Base model

togethercomputer/gpt-oss-20b-bf16

Adapter

(22)

this model