TikTok TechJam 2026 — OpenFake train slice
Filtered, re-encoded train split used by Seer. This is not a copy of
ComplexDataLab/OpenFake
(3.44 TB). It is the on-disk mixture source openfake/train.
Do not use core/test or reddit/test from this repo — those holdouts stay
out of training.
Contents
439,523 JPEGs, ~102 GB. Images were streamed from OpenFake core/train,
filtered to 30 generators + two real pools, resized to max_side=1536,
JPEG quality 95.
| split | count | folders |
|---|---|---|
| fake | 309,523 | fake/<generator>/ (30 models) |
| real | 130,000 | real/pexels (80,000) + real/laion (50,000) |
Generator caps came from rank.json (recall of an earlier Seer checkpoint)
with tiers 0.70=25000 0.95=15000 0.98=10000 and
--cap-model pexels=80000 laion=50000. Some models have fewer images than
their cap (upstream pool exhausted). _meta/counts.json and _meta/plan.json
are the exact plan and realized counts.
Load
from huggingface_hub import snapshot_download
root = snapshot_download(
"glennwuwu/tiktok-techjam-2026-openfake-train",
repo_type="dataset",
)
# then point Seer at:
# fake_dirs: [f"{root}/fake"]
# real_dirs: [f"{root}/real"]
Licence / attribution
Redistribution of a subset of OpenFake
(cc-by-sa-4.0). Proprietary-generator subsets are non-commercial.
Cite ComplexDataLab/OpenFake if you use this slice.
- Downloads last month
- 26