Dataset Viewer
Auto-converted to Parquet Duplicate
text
stringlengths
0
1.47k
# ===== file: splatreg/__init__.py =====
"""splatreg, composable geometry-first SE(3)/Sim(3) registration for 3D Gaussian Splatting.
*gsplat renders your Gaussians; splatreg registers against them.*
Public surface (filled in by the carve):
register(target, source, residuals=[...], transform="sim3", backend="builtin") -> RegisterResult
merge([a, b, ...], ref=0) -> Gaussians
Tracker(target, residuals=[...]).track(frame) -> RegisterResult
Residual, Solver (extension points)
"""
from .core.types import Gaussians, Frame, RegisterResult, LinearizedProblem, SE3Update
from .residuals.base import Residual
from .solvers.base import Solver
from .quality import QualityConfig, resolve_quality
from .hypotheses import (
FusionDecision,
HypothesisScore,
PoseCluster,
calibrate_p95_threshold,
cluster_hypotheses,
decide_fusion,
score_hypotheses,
)
# The high-level pipeline (splatreg.api) is added by the carve; tolerate its absence pre-build.
try:
from .api import apply_transform, register, merge, Tracker # noqa: F401
except ImportError:
register = merge = Tracker = None # type: ignore
# v0.2: 6-DoF object-pose mode (pure torch — reuses register/track, always available).
try:
from .object_pose import ( # noqa: F401
ObjectPose,
ObjectPoseEstimator,
estimate_object_pose,
add_metric,
adds_metric,
add_auc,
)
except ImportError: # pragma: no cover
ObjectPose = ObjectPoseEstimator = estimate_object_pose = None # type: ignore
add_metric = adds_metric = add_auc = None # type: ignore
# v0.2: camera localization in a splat (needs the gsplat [render] extra — guarded).
# v0.3 hardening: coarse_localize_camera is a CPU/projection-only wide-baseline SEED (no gsplat),
# so it imports even without the render extra.
try:
from .camera_loc import coarse_localize_camera # noqa: F401
except ImportError: # pragma: no cover
coarse_localize_camera = None # type: ignore
try:
from .camera_loc import CameraPhotometric, localize_camera # noqa: F401
except ImportError: # gsplat not installed
CameraPhotometric = localize_camera = None # type: ignore
# v0.3: multi-splat joint/bundle registration + scene-scale spatial index (pure torch).
try:
from .bundle import bundle_register, pairwise_consistency, BundleResult # noqa: F401
except ImportError: # pragma: no cover
bundle_register = pairwise_consistency = BundleResult = None # type: ignore
try:
from .spatial_index import SpatialIndex, build_index # noqa: F401
except ImportError: # pragma: no cover
SpatialIndex = build_index = None # type: ignore
# v1.2: real-SH Wigner-D rotation (pure torch — rotates f_rest with the splat).
from .sh import rotate_sh, sh_rotation_matrix # noqa: F401
# post-v1.2: MAC maximal-clique correspondence seed (init="mac"; clique enumeration needs the
# optional networkx — the import itself is torch-only, the dependency is checked at call time).
try:
from .mac import mac_pose, mac_feature_align # noqa: F401
except ImportError: # pragma: no cover
mac_pose = mac_feature_align = None # type: ignore
__version__ = "1.4.0"
__all__ = [
"apply_transform",
"register",
"merge",
"Tracker",
"Residual",
"Solver",
"QualityConfig",
"resolve_quality",
"PoseCluster",
"HypothesisScore",
"FusionDecision",
"cluster_hypotheses",
"score_hypotheses",
"decide_fusion",
"calibrate_p95_threshold",
"Gaussians",
"Frame",
"RegisterResult",
"LinearizedProblem",
"SE3Update",
End of preview. Expand in Data Studio

Training data for the RSI Bench task length-extrapolating-architecture

1,010,115,101 tokens (Mistral-7B tokenizer, revision 27d67f1b5f57), stored as uint16 in train.bin. Built by data_build_v2.py on 2026-10-13; manifest.json records every source revision, filter and hash.

Files

File What it is
train.bin the token stream; long documents start at multiples of 2,048 tokens, end-of-text id 2 between documents
segments.npy one row per laid-out segment: start, length, source, document length, kind, row of docs.jsonl
docs.jsonl one row per document: source, reference (repository, arXiv ID, PG-19 ID, FineWeb-Edu ID), licence, tokens, sha256 of its tokens
repo_pins.json each repository: commit (the last before 2025-01-01), commit date, licence, file count
notices.tar.gz each repository's LICENSE, COPYING, NOTICE and COPYRIGHT files, as found at the pinned commit
manifest.json sources, revisions, mix, filters, statistics, hashes
dev/ the task's development set (repositories and arXiv papers from May to September 2026): token files, character offsets, texts, hard-token masks, index.json (licence, commit or arXiv ID, scoring length), MANIFEST.json (sha256 of every file) and notices/ (each repository's licence files)

Sources and licences

Each part keeps its own licence. This dataset as a whole is offered under those licences and no other.

Source Share Licence
FineWeb-Edu (HuggingFaceFW/fineweb-edu at 87f09149ef47) 35% ODC-By 1.0; credit to Hugging Face (FineWeb-Edu) and Common Crawl
GitHub repositories (repo_pins.json) 30% per repository: MIT, Apache-2.0, BSD-2-Clause, BSD-3-Clause or ISC; licence and notice texts in notices.tar.gz
arXiv papers (common-pile/arxiv_papers at 963fe980c55b) 15% per paper, CC BY or CC0 (listed in docs.jsonl); credit to each paper's authors
PG-19 books (deepmind/pg19 at 4d28bd77e669) 20% Apache-2.0 dataset; the books are in the US public domain

Documents by source and licence:

Source Licence Documents
book Apache-2.0 dataset, public-domain texts 3,823
code apache-2.0 2,020
code bsd-2-clause 87
code bsd-3-clause 287
code isc 23
code mit 4,193
paper Creative Commons - Attribution - https://creativecommons.org/licenses/by/3.0/ 421
paper Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/ 6,218
paper Creative Commons Zero - Public Domain - https://creativecommons.org/publicdomain/zero/1.0/ 266
web ODC-By 1.0 311,531

Changes made to the sources

  • Text was tokenized; line endings became "\n"; the literal text </s>, <s> and <unk> was encoded as plain text.
  • Repositories: code files only (rules in the build's README), source folders first, joined with a # ===== file: <path> ===== line before each file. Licence and notice files were kept separately (notices.tar.gz).
  • Every document was cut at its first 65,536 tokens.
  • FineWeb-Edu pages from Project Gutenberg were removed, and documents that copy the task's evaluation documents were removed.

Takedown

To ask for a document's removal, open a discussion on this dataset's page with the reference from docs.jsonl.

Downloads last month
226