Modelpublic

XUUUUSID/gpt2-124m-variants-pre

sourceHugging Faceapache-2.0updated 23d agoView on Hugging Face

0likes

Model Card

gpt2-124m-variants-pre

Research artifacts: GPT-2 (124M, trained from scratch) pretraining variants produced under controlled data-composition differences; per-variant, per-seed subfolders; intended for reproducibility and membership/contamination research. Details and provenance will be documented in a future update.

Subfolders

—clean_seed20260901
—clean_seed20260902
—inj1x_seed20260901
—inj1x_seed20260902
—inj16x_seed20260901
—inj16x_seed20260902

Loading

python

from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "XUUUUSID/gpt2-124m-variants-pre"
sub = "clean_seed20260901"
tok = AutoTokenizer.from_pretrained(repo, subfolder=sub)
model = AutoModelForCausalLM.from_pretrained(repo, subfolder=sub)