The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/libs/libcommon/src/libcommon/packaged_modules.py", line 18, in _refuse_lance
raise NotImplementedError(LANCE_DISABLED_MESSAGE)
NotImplementedError: The Lance format is not supported.
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Quarry Prompt Datasets
This folder contains a collection of LanceDB tables holding prompts, captions, and metadata for images, video, music, and prose. Most were imported from publicly available datasets on the Hugging Face Hub.
How to use these files
These tables are consumed by the SwarmUI-Quarry extension:
Drop this folder where the extension expects its data and it will index the .lance tables to
provide prompt search, browsing, and suggestion features inside SwarmUI. The files are standard
LanceDB tables, so they can also be opened directly with the lancedb
Python or Node libraries if you want to query them yourself.
Attribution & data sources
Each .lance file is named after the Hugging Face dataset it was built from, with the / in the
repository id replaced by a .. For example, Gustavosta.Stable-Diffusion-Prompts.lance comes
from Gustavosta/Stable-Diffusion-Prompts.
All credit for the underlying data belongs to the original dataset authors. Please refer to
each source repository for its full license, terms of use, and citation requirements. Links in
the table below were verified against the Hugging Face Hub on 2026-06-13, except the
ThetaCursed, llaa33219, quarterturn, Falah/new_photorealistic_prompts, and jgreely sources,
which were checked on 2026-09-12, the webshart sources, which were checked on 2026-09-13,
and the sources added on 2026-09-15: CaptionEmporium, Doub7e, Felldude, GHonem, RicemanT,
SilentAntagonist/vintage-artworks-60k-captioned, codeShare, juliaturc, nixiieee, opendiffusionai,
ossaili, quik-models, unography, victorzarzu, and visheratin (jgreely is hosted on GitHub).
The License column reflects the source's declared license
at the corresponding verification date. Image-prompt tables live in nl/ or tags/;
older entries below show only the table name.
| LanceDB file | Source dataset | License |
|---|---|---|
Aconexx.CivitAI-Flux-Prompts.lance |
Aconexx/CivitAI-Flux-Prompts | Apache-2.0 |
AlekseyKorshuk.midjourney-prompts-text-dedup.lance |
AlekseyKorshuk/midjourney-prompts-text-dedup | Not specified |
AlekseyKorshuk.product-photography-v1-tiny-prompts-tasks-collage-filtered.lance |
AlekseyKorshuk/product-photography-v1-tiny-prompts-tasks-collage-filtered | Not specified |
AliN96.midjourney-prompts-embeddings.lance |
AliN96/midjourney-prompts-embeddings | MIT |
BaiqiL.GenAI-Bench.lance |
BaiqiL/GenAI-Bench | Apache-2.0 |
nl/CaptionEmporium.flickr-megalith-10m-internvl2-multi-caption.lance |
CaptionEmporium/flickr-megalith-10m-internvl2-multi-caption | CC-BY-SA-4.0 |
nl/CaptionEmporium.laion-pop-llama3.2-11b.lance |
CaptionEmporium/laion-pop-llama3.2-11b | CC-BY-SA-4.0 |
nl/CaptionEmporium.midjourney-niji-1m-llavanext.lance |
CaptionEmporium/midjourney-niji-1m-llavanext | CC-BY-SA-4.0 |
ChrisGoringe.flux_prompts.lance |
ChrisGoringe/flux_prompts | Apache-2.0 |
CodeGoat24.UniGenBench.lance |
CodeGoat24/UniGenBench | MIT |
CyberHarem.lance |
CyberHarem (organization) | MIT |
DamarJati.SD-Prompts.lance |
DamarJati/SD-Prompts | CC0-1.0 |
DamarJati.mj-disney.lance |
DamarJati/mj-disney | Not specified |
nl/Doub7e.ArtBench-2-gpt4o-captions.lance |
Doub7e/ArtBench-2-gpt4o-captions | Not specified |
Falah.anime_arabic_style_prompts.lance |
Falah/anime_arabic_style_prompts | Not specified |
nl/Falah.new_photorealistic_prompts.lance |
Falah/new_photorealistic_prompts | Not specified |
Falah.varied_portrait_prompts.lance |
Falah/varied_portrait_prompts | Not specified |
nl/Felldude.84k_40GB_Captioned_2k_XXX_Images.lance |
Felldude/84k_40GB_Captioned_2k_XXX_Images | CreativeML-OpenRAIL-M |
nl/Felldude.Human_Corpus_Small.lance |
Felldude/Human_Corpus_Small | CreativeML-OpenRAIL-M |
nl/Felldude.xxx_hdr_13GB.lance |
Felldude/xxx_hdr_13GB | CreativeML-OpenRAIL-M |
FredZhang7.anime-prompts-180K.lance |
FredZhang7/anime-prompts-180K | CreativeML-OpenRAIL-M |
nl/GHonem.fashion_image_caption-3500.lance |
GHonem/fashion_image_caption-3500 | Not specified |
Geonmo.midjourney-prompts-only.lance |
Geonmo/midjourney-prompts-only | Not specified |
Gustavosta.Stable-Diffusion-Prompts.lance |
Gustavosta/Stable-Diffusion-Prompts | Unknown |
KBlueLeaf.danbooru2023-florence2-caption.lance |
KBlueLeaf/danbooru2023-florence2-caption | Apache-2.0 |
Mediocreatmybest.Miscellany_of_Australian_Historical_Photography.lance |
Mediocreatmybest/Miscellany_of_Australian_Historical_Photography | CC0-1.0 |
MingSafeR.anime_porn_image_prompts.lance |
MingSafeR/anime_porn_image_prompts | Apache-2.0 |
MingSafeR.photo_porn_image_prompts.lance |
MingSafeR/photo_porn_image_prompts | Apache-2.0 |
MohamedRashad.midjourney-detailed-prompts.lance |
MohamedRashad/midjourney-detailed-prompts | Apache-2.0 |
Nelis.Midjourney_Stable_Diffusion_prompts.lance |
Nelis/Midjourney_Stable_Diffusion_prompts | Unknown |
Nelis.prompts.lance |
Nelis/prompts | Unknown |
NewEden.CivitAI-SD-Prompts.lance |
NewEden/CivitAI-SD-Prompts | Apache-2.0 |
Owen777.9M_77Token_SDXL_Prompts.lance |
Owen777/9M_77Token_SDXL_Prompts | Not specified |
Pratofeitoo.fashion_photography_prompts_SDXL.lance |
Pratofeitoo/fashion_photography_prompts_SDXL | Not specified |
nl/RicemanT.Anime-Background-Finetuning-V1.1.screencap.lance |
RicemanT/Anime-Background-Finetuning-V1.1 | Apache-2.0 |
nl/RicemanT.booru-essence-2026.danbooru.lance |
RicemanT/booru-essence-2026 | CC-BY-NC-SA-4.0 |
nl/RicemanT.booru-essence-2026.e621.lance |
RicemanT/booru-essence-2026 | CC-BY-NC-SA-4.0 |
nl/SilentAntagonist.vintage-artworks-60k-captioned.lance |
SilentAntagonist/vintage-artworks-60k-captioned | CC-BY-SA-4.0 |
SilentAntagonist.vintage-photography-450k-high-quality-captions.lance |
SilentAntagonist/vintage-photography-450k-high-quality-captions | CC-BY-SA-4.0 |
tags/ThetaCursed.danbooru-2026-clean-metadata.lance |
ThetaCursed/danbooru-2026-clean-metadata | Not specified |
WoWoWoWololo.stable_diffusion_female_prompts.lance |
WoWoWoWololo/stable_diffusion_female_prompts | MIT |
X779.Danbooruwildcards.characters_str_tags_danbooru_full.lance |
X779/Danbooruwildcards | Other |
agentlans-stock-photos-asian-people.lance |
agentlans/stock-photos-asian-people | CC0-1.0 |
akba08.ultra-realistic-cinematic-photography.lance |
akba08/ultra-realistic-cinematic-photography | MIT |
caqlayan.prompts2.lance |
caqlayan/prompts2 | Not specified |
chengli-thu.anime-character-prompt-15k.lance |
chengli-thu/anime-character-prompt-15k | Apache-2.0 |
chinoll.Danbooru2024-Webp-4MPixel-NL.lance |
chinoll/Danbooru2024-Webp-4MPixel-NL | Other (see page) |
civitai.author_prompts.lance |
civitai.com site scrape (no Hugging Face source) | Not specified |
civitai.lance |
civitai.com site scrape (no Hugging Face source) | Not specified |
nl/codeShare.chroma_prompts.anime_captions.lance |
codeShare/chroma_prompts | MIT |
nl/codeShare.chroma_prompts.photoreal.lance |
codeShare/chroma_prompts | MIT |
nl/codeShare.chroma_prompts.reddit_captions.lance |
codeShare/chroma_prompts | MIT |
csuhan.midjourney-prompts-FLUX.lance |
csuhan/midjourney-prompts-FLUX | Not specified |
dartags.danbooru-2408-blind-captions.lance |
dartags/danbooru-2408-blind-captions | MIT |
daspartho.stable-diffusion-prompts.lance |
daspartho/stable-diffusion-prompts | Not specified |
deepghs.sankaku_full.lance |
deepghs/sankaku_full | Other |
dnkdnk.CVTG-2K.lance |
dnkdnk/CVTG-2K | Apache-2.0 |
dvilasuero.awesome_hunyuanImage_prompts.lance |
dvilasuero/awesome_hunyuanImage_prompts | Apache-2.0 |
gaodrew.midjourney-prompts-highquality.lance |
gaodrew/midjourney-prompts-highquality | Apache-2.0 |
gnokit.stable-diffusion-prompts.lance |
gnokit/stable-diffusion-prompts | Not specified |
huzaib.stable-diffusion-prompts.lance |
huzaib/stable-diffusion-prompts | Not specified |
isidentical.random-stable-diffusion-prompts.lance |
isidentical/random-stable-diffusion-prompts | Not specified |
nl/jgreely.c1ga.lance |
jgreely/c1ga (GitHub, no Hugging Face source) | MIT |
jorgevipo.image_prompts_dataset.lance |
jorgevipo/image_prompts_dataset | Not specified |
jtatman.stable-diffusion-prompts-stats-full-uncensored.lance |
jtatman/stable-diffusion-prompts-stats-full-uncensored | Not specified |
jtatman.stable-diffusion-prompts-uncensored.lance |
jtatman/stable-diffusion-prompts-uncensored | MIT |
nl/juliaturc.captioned-cartoons.lance |
juliaturc/captioned-cartoons | Not specified |
k-mktr.improved-flux-prompts-photoreal-portrait.lance |
k-mktr/improved-flux-prompts-photoreal-portrait | MIT |
k-mktr.improved-flux-prompts.lance |
k-mktr/improved-flux-prompts | MIT |
k-mktr.trending-text-to-image.lance |
k-mktr/trending-text-to-image | MIT |
k-mktr.z-image-ethnicity-test.lance |
k-mktr/z-image-ethnicity-test | Apache-2.0 |
k-mktr.z-image-examples.lance |
k-mktr/z-image-examples | Apache-2.0 |
kaupane.vintage-photography-captions.lance |
kaupane/vintage-photography-captions | Not specified |
lioooox.T2I-CoReBench.lance |
lioooox/T2I-CoReBench | Apache-2.0 |
tags/llaa33219.danbooru-anime-screenshot.lance |
llaa33219/danbooru-anime-screenshot | Not specified |
mnemic.200k-random-prompts.lance |
mnemic/200k-random-prompts | Not specified |
moescape.lance |
moescape.ai site scrape (no Hugging Face source) | Not specified |
nl/nixiieee.wikiart_captions.lance |
nixiieee/wikiart_captions | Not specified |
nl/opendiffusionai.pexels-woman-croppable.lance |
opendiffusionai/pexels-woman-croppable | Not specified |
nl/ossaili.archdaily_30k_captioned.lance |
ossaili/archdaily_30k_captioned | Not specified |
poloclub.diffusiondb.lance |
poloclub/diffusiondb | CC0-1.0 |
nl/quarterturn.danbooru-1024-eq-captioned.lance |
quarterturn/danbooru-1024-eq-captioned | CC-BY-SA-4.0 (Hub metadata; see note below) |
nl/quik-models.captions-100k.lance |
quik-models/captions-100k | Not specified |
roborovski.upsampled-prompts-parti.lance |
roborovski/upsampled-prompts-parti | Not specified |
sengunsipahi.civitai_top10k.lance |
sengunsipahi/civitai_top10k | Not specified |
succinctly.midjourney-prompts.lance |
succinctly/midjourney-prompts | Apache-2.0 |
suriyagunasekar.midjourney-prompts.lance |
suriyagunasekar/midjourney-prompts | MIT |
svjack-conceptual_captions_3m_en_tiny.lance |
svjack/conceptual_captions_3m_en_tiny | Not specified |
taoki.stair-captions-prompts.lance |
taoki/stair-captions-prompts | CC-BY-4.0 |
tommmyhug.concept-art-female-anime-realism-fantasy-cyberpunk-steampunk-post-apocalypse-art-prompts.lance |
tommmyhug/concept-art-female-anime-realism-fantasy-cyberpunk-steampunk-post-apocalypse-art-prompts | MIT |
nl/unography.shopify-stock-images-2048-captioned.lance |
unography/shopify-stock-images-2048-captioned | Not specified |
nl/victorzarzu.interior-design-edit-captions.lance |
victorzarzu/interior-design-edit-captions | Not specified |
nl/visheratin.unsplash-caption-questions-init.lance |
visheratin/unsplash-caption-questions-init | Not specified |
wtcherr.midjourney-prompts.lance |
wtcherr/midjourney-prompts | Not specified |
nl/webshart.pseudo-camera-10k-structured.lance |
webshart/pseudo-camera-10k-structured | CC (variant unspecified in Hub metadata) |
nl/webshart.terminusresearch-photo-aesthetics.lance |
webshart/terminusresearch-photo-aesthetics | MIT |
nl/webshart.terminusresearch-photo-anatomy.lance |
webshart/terminusresearch-photo-anatomy | MIT |
nl/webshart.terminusresearch-photo-typography.lance |
webshart/terminusresearch-photo-typography | MIT |
Some local table names include a subset suffix after the source repository name:
RicemanT.Anime-Background-Finetuning-V1.1.screencaprefers to the source'sScreencap/subset.RicemanT.booru-essence-2026.danbooruand.e621refer to the source'sdata/danbooru.parquetanddata/e621.parquetfiles.codeShare.chroma_prompts.anime_captions,.photoreal, and.reddit_captionsrefer to subsets withincodeShare/chroma_prompts, not separate Hub repositories.
Notes
- "Not specified" means the source repository did not declare a license on its Hub page. That is not the same as public domain — when in doubt, contact the dataset author before reusing the data.
- Several datasets are themselves derivatives (e.g.
daspartho/stable-diffusion-promptsandgaodrew/...are filtered subsets of other datasets, andkaupane/...re-captionsSilentAntagonist/...). Upstream terms may also apply. quarterturn/danbooru-1024-eq-captioneddeclares CC-BY-SA-4.0 in its Hub metadata, but its dataset card also limits use to personal offline research and states that images remain under individual artist copyright. See the linked source for these additional terms.- These tables are redistributions/derivatives of the datasets above; the original authors retain all rights to their data. If you are a dataset author and would like your data removed, please open an issue on the SwarmUI-Quarry repository.
- Licenses vary per dataset. Always check the linked source before using any data commercially or redistributing it further.
Short-story & roleplay datasets (short-stories/)
The tables in the short-stories/ subfolder are a separate collection of prose datasets —
short stories, roleplay/instruction conversations, and writing-prompt data — rather than the
text-to-image prompts above. They follow the same naming idea, except the / in the source
repository id is replaced by - (so org/repo boundaries are not always obvious from the file name
alone). Every link below was verified against the Hugging Face Hub on 2026-06-27; the License
column reflects the license declared on each dataset's Hub page at that time.
| LanceDB file | Source dataset | License |
|---|---|---|
short-stories/Chat-Error-tinystories-gpt4-eval.lance |
Chat-Error/tinystories-gpt4 | Not specified |
short-stories/Chat-Error-tinystories-gpt4-train.lance |
Chat-Error/tinystories-gpt4 | Not specified |
short-stories/DataMajin-Data-Majin_Short-Stories.lance |
DataMajin/Data-Majin_Short-Stories | OpenRAIL |
short-stories/NewEden-Forge-Erebus-R_ShortStories-Combined.lance |
NewEden-Forge/Erebus-R_ShortStories-Combined | Not specified |
short-stories/SicariusSicariiStuff-No_Robots_ShareGPT.lance |
SicariusSicariiStuff/No_Robots_ShareGPT | Apache-2.0 |
short-stories/SicariusSicariiStuff-Reddit_Dirty_Writing_Prompts_ShareGPT.lance |
SicariusSicariiStuff/Reddit_Dirty_Writing_Prompts_ShareGPT | Apache-2.0 |
short-stories/SicariusSicariiStuff-Short_Stories_ShareGPT1.lance |
SicariusSicariiStuff/Short_Stories_ShareGPT | Apache-2.0 |
short-stories/TinyStoriesV2-GPT4-train.lance |
roneneldan/TinyStories | CDLA-Sharing-1.0 |
short-stories/agentlans-Gemma-2-Ataraxy-9B-short-stories.lance |
agentlans/Gemma-2-Ataraxy-9B-short-stories | Gemma |
short-stories/agentlans-lemonilia-LimaRP-scenario.lance |
agentlans/lemonilia-LimaRP | Apache-2.0 |
short-stories/agentlans-lemonilia-LimaRP-user_description.lance |
agentlans/lemonilia-LimaRP | Apache-2.0 |
short-stories/agentlans-lemonilia-LimaRP.lance |
agentlans/lemonilia-LimaRP | Apache-2.0 |
short-stories/agentlans-literary-genre-examples.lance |
agentlans/literary-genre-examples | CC-BY-NC-SA-4.0 |
short-stories/agentlans-literary-reasoning.lance |
agentlans/literary-reasoning | CC0-1.0 |
short-stories/agentlans-prompt-difficulty.lance |
agentlans/prompt-difficulty | CC-BY-4.0 |
short-stories/allura-org-SynthRP-RpR-converted.lance |
allura-org/SynthRP-RpR-converted | CC-BY-SA-4.0 |
short-stories/allura-org-r_shortstories_24k.lance |
allura-org/r_shortstories_24k | CC-BY-NC-ND-4.0 |
short-stories/athena-ml-gpt4_short_stories_with_tokens.lance |
athena-ml/gpt4_short_stories_with_tokens | Not specified |
short-stories/dltdojo-park-tinystories-533k.lance |
dltdojo/park-tinystories-533k | Not specified |
short-stories/lordgrim18-short-stories.lance |
lordgrim18/short-stories | Not specified |
short-stories/roneneldan-TinyStories.lance |
roneneldan/TinyStories | CDLA-Sharing-1.0 |
short-stories/skeskinen-TinyStories-hf.lance |
skeskinen/TinyStories-hf | Not specified |
A few of these file names are not a direct org-repo split:
Chat-Error-tinystories-gpt4-evaland-trainare the two splits of the single repoChat-Error/tinystories-gpt4(the localevalname corresponds to the Hub'stestsplit).TinyStoriesV2-GPT4-trainis theTinyStoriesV2-GPT4-train.txtfile shipped insideroneneldan/TinyStories, not a standalone repo.SicariusSicariiStuff-Short_Stories_ShareGPT1maps toSicariusSicariiStuff/Short_Stories_ShareGPT(the trailing1is a local shard suffix).agentlans-lemonilia-LimaRP-scenarioand-user_descriptionare column subsets ofagentlans/lemonilia-LimaRP; no separate Hub repos exist for those suffixes.skeskinen/TinyStories-hfis itself a copy ofroneneldan/TinyStories(CDLA-Sharing-1.0) but declares no license of its own.
Video prompt datasets (video/)
The tables in the video/ subfolder contain text-to-video prompts or index metadata. They
follow the same naming idea as the main table (the / in the source repository id replaced by a
.). The links below were verified against the Hugging Face Hub on 2026-07-03 (GokuScraper),
2026-09-13 (webshart), and 2026-09-15 (AdwolfCzar, DSULT-Core, MultimodalFriends, and svjack);
the License column reflects each repository's declaration at that time.
| LanceDB file | Source dataset | License |
|---|---|---|
video/AdwolfCzar.bouncing_body_jiggle_dataset_captioned_30-07.lance |
AdwolfCzar/bouncing_body_jiggle_dataset_captioned_30-07 | Not specified |
video/DSULT-Core.i-love-anime-sakuga.aesthetic.lance |
DSULT-Core/i-love-anime-sakuga | Other (gaysex-dont-be-a-prick-license) |
video/DSULT-Core.i-love-anime-sakuga.full.lance |
DSULT-Core/i-love-anime-sakuga | Other (gaysex-dont-be-a-prick-license) |
video/GokuScraper.seedance-2-prompts-datasets.lance |
GokuScraper/seedance-2-prompts-datasets | CC-BY-4.0 |
video/MultimodalFriends.AcitvityNet-Captions-bootstrapped-5K.lance |
MultimodalFriends/AcitvityNet-Captions-bootstrapped-5K | Not specified |
video/svjack.Goku_MovieGenBench_Videos_Captioned.lance |
svjack/Goku_MovieGenBench_Videos_Captioned | Not specified |
video/webshart.OpenVid-1M-webshart-indices.lance |
webshart/OpenVid-1M-webshart-indices | Not specified (index repository; source videos declare CC-BY-4.0) |
The OpenVid table contains index metadata and captions, not the source video files.
The DSULT-Core.i-love-anime-sakuga.aesthetic and .full tables correspond to
metadata/sakugadataset_train_aesthetic.parquet and metadata/sakugadataset_train_full.parquet
within the same source repository. Its custom license is linked in the table above.
AcitvityNet preserves the spelling used by the MultimodalFriends source repository.
Music datasets (music/)
The music/ folder contains music caption and metadata tables. The source repository below was
verified against the Hugging Face Hub on 2026-09-13.
| LanceDB file | Source dataset | License / rights |
|---|---|---|
music/webshart.suno-various-94k.lance |
webshart/suno-various-94k | Source rights retained; see dataset card |
Publishing / updating this collection
Upload new or changed tables:
hf upload-large-folder jtreminio/prompt-dataset . --repo-type=dataset
Note: uploading only adds and updates files — it never removes them. If you move or
rename a table locally (e.g. into nl/ or tags/), the old copy still lingers on the remote
and must be deleted explicitly so the Hub tree matches local:
# delete a single stale path (recursive)
hf repos delete-files jtreminio/prompt-dataset "old-name.lance/*" --repo-type=dataset
To prune everything on the remote that no longer exists locally in one pass, diff the two
trees (the CLI has no file-listing command, so use the huggingface_hub Python API) and feed
the stale top-level paths to delete-files:
PATTERNS=$(python -c '
import os
from huggingface_hub import HfApi
remote = set(f.split("/")[0] for f in
HfApi().list_repo_files("jtreminio/prompt-dataset", repo_type="dataset"))
local = set(os.listdir("."))
# never delete the Hub-managed .gitattributes
for p in sorted((remote - local) - {".gitattributes"}):
print(f"{p}/*")
')
hf repos delete-files jtreminio/prompt-dataset $PATTERNS --repo-type=dataset
Before deleting, confirm each stale path has really been relocated (e.g. now lives under nl/
or tags/) so you are pruning duplicates, not the only copy.
- Downloads last month
- 1,105