lossisnotanumber/browsecomp-plus-100k-corpus-as-local-folder

BrowseComp-Plus 100K Corpus — as local folder

The 100K-document subset of the BrowseComp-Plus benchmark corpus, processed into a plain document-directory form using the browsecomp-plus processing code from DCI-Agent-Lite.

This repository stores the corpus exactly as it is expected on local disk: a flat tree of <domain>/<title>.txt files, ready to be pointed at by --corpus-dir. It is the form consumed by the RARG / DCI-Agent retrieval-augmented agent during the BrowseComp-Plus evaluation.

Source of the processing pipeline: https://github.com/DCI-Agent/DCI-Agent-Lite

Repository contents

File	Size	Description
bc_plus_100k.zip	~1.1 GB	The full 100K corpus, zipped. Extracts to bc_plus_100k/.

Corpus structure (after unzipping)

bc_plus_100k/
├── <domain>/
│   ├── <title>.txt
│   └── ...
├── en.wikipedia.org/
│   ├── ...
└── ...

• 100,195 plain-text .txt files
• 34,506 distinct source domains (top: en.wikipedia.org 8,524, www.reddit.com 3,553, pmc.ncbi.nlm.nih.gov 1,679, www.imdb.com 1,550, www.youtube.com 1,267)
• ~3.2 GB uncompressed

Each file is a single captured web document. The folder name is the source domain and the file name is the page title. The directory layout matches what DCI-Agent-Lite / RARG expect under corpus/bc_plus_100k.

How to load

from huggingface_hub import hf_hub_download
import zipfile

# Download the single zip (HF resolves it via git-lfs)
zip_path = hf_hub_download(
    repo_id="lossisnotanumber/browsecomp-plus-100k-corpus-as-local-folder",
    filename="bc_plus_100k.zip",
    repo_type="dataset",
)

# Extract into corpus/bc_plus_100k so it matches RARG's --corpus-dir convention
with zipfile.ZipFile(zip_path) as z:
    z.extractall("corpus")

# Now usable as:  --corpus-dir corpus/bc_plus_100k

Or with the hf CLI:

hf download lossisnotanumber/browsecomp-plus-100k-corpus-as-local-folder bc_plus_100k.zip --repo-type dataset

Using with RARG / DCI-Agent-Lite

Both projects expect the corpus at corpus/bc_plus_100k. After extracting, the run scripts (e.g. scripts/sra_bench/run_bcplus_100k_ts_mirror_agent_*.sh) work unchanged with --corpus-dir corpus/bc_plus_100k.