This release is a pre-release and may not be stable for production use.
GTBoost
Gradient boosting for tabular data. Rust core, Python/sklearn API. Native NaN and categorical handling.
Alpha software — APIs may change.
Install
pip install gtboost
From source:
git clone https://github.com/gtboost-org/gtboost.git
cd gtboost
pip install maturin
maturin develop --release
Regression — California housing
from gtboost import GTBoostRegressor
from sklearn.datasets import fetch_california_housing
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error
X, y = fetch_california_housing(return_X_y=True)
Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.2, random_state=0)
model = GTBoostRegressor(n_estimators=1000, learning_rate=0.05, max_depth=6, seed=0)
model.fit(Xtr, ytr, eval_set=[(Xte, yte)], early_stopping_rounds=50)
print(mean_squared_error(yte, model.predict(Xte)) ** 0.5)
Classification — diabetes (binarized)
import numpy as np
from gtboost import GTBoostClassifier
from sklearn.datasets import load_diabetes
from sklearn.model_selection import train_test_split
X, y = load_diabetes(return_X_y=True)
y = (y > np.median(y)).astype(int)
Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.2, random_state=0)
clf = GTBoostClassifier(n_estimators=500, learning_rate=0.05, max_depth=3, seed=0)
clf.fit(Xtr, ytr, eval_set=[(Xte, yte)], early_stopping_rounds=50)
proba = clf.predict_proba(Xte)[:, 1]
Main features
All are constructor kwargs (defaults shown). Descriptions are mechanical, not benchmark claims.
| kwarg | default | what it does |
|---|---|---|
honest |
False |
fit split structure and leaf values on disjoint halves of the data |
stein_leaves |
False |
positive-part shrinkage of each tree's leaf vector toward the global step |
vertical_init |
False |
fit per-feature linear effects before the trees |
monotone_constraints |
[] |
per-feature monotonicity: 1 up, -1 down, 0 none |
interval_splits |
False |
bounded split candidates a <= x <= b |
sparse_oblique_splits |
False |
2-feature diagonal split candidates |
grow_policy |
"depthwise" |
depthwise / leafwise / oblivious / adaptive |
huber_delta |
0.0 |
Huber loss for regression (0.0 = squared error) |
n_refine |
0 |
post-fit leaf-value refit passes |
apx |
True |
average predictions across boosting-trajectory checkpoints |
cat_features |
None |
mark categorical columns; pandas category/object/string auto-detected |
Noisy, small data (diabetes) — structure-honest fit with leaf shrinkage and a linear prior:
reg = GTBoostRegressor(honest=True, stein_leaves=True, vertical_init=True, seed=0)
reg.fit(Xtr, ytr)
Monotone constraint (California, target non-decreasing in feature 0, MedInc):
reg = GTBoostRegressor(monotone_constraints=[1, 0, 0, 0, 0, 0, 0, 0], seed=0)
reg.fit(Xtr, ytr)
Categorical features
Pandas category/object/string columns are detected automatically. Pass cat_features to override. Fit and predict on DataFrames with the same columns so categories map consistently.
clf = GTBoostClassifier(cat_features=["city", "segment"], seed=0)
clf.fit(df_train, y_train)
proba = clf.predict_proba(df_test)
Tuning
from gtboost.tuner import tune_gtboost
result = tune_gtboost(Xtr, ytr, task="regression", n_trials=30, n_folds=3)
model = GTBoostRegressor(**result.best_params).fit(Xtr, ytr)
Refine
Post-hoc joint leaf re-solve (regression, binary, multiclass): re-fits all leaf
values against the final residual, shrunk toward the greedy values, behind an
evidence gate. See REFINE_2.md.
Scope
Stable: GTBoostRegressor, GTBoostClassifier, Dataset, native NaN and
categorical handling, interval splits. A native gtboost.train / Dataset API
also exists.
Experimental: research knobs on GTBoostModel — see GTBOOST_FEATURES.md.
1 maintainer
Metadata
1 maintainer
Release files for gtboost 0.1.0a2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| gtboost-0.1.0a2.tar.gz | 370.3 kB | With provenance | Details for gtboost-0.1.0a2.tar.gz |
Built distributions (wheels)
Total release size: 14.1 MB
Release files / gtboost-0.1.0a2.tar.gz
| Download URL | gtboost-0.1.0a2.tar.gz |
|---|---|
| Size | 370.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
1ccaf8e6eb2ac3d378368eb06dd4ac420b22acf33872cd1f257826959c755518
|
|
BLAKE2b-256 checksum How to use checksums |
67fab826fca1201e337b44df8a0b1795b09089cf0c44ec352bd5ef35079ea8b4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp312-cp312-win_amd64.whl
| Download URL | gtboost-0.1.0a2-cp312-cp312-win_amd64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.12 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
2335a22b769876d14de658d83d44c3efcaaf73583719e1d8cd5e50ba7d96a0ea
|
|
BLAKE2b-256 checksum How to use checksums |
08b8323f2edc34a4865c1e8068e28dfc5f29172512435144156b0c31b800554c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | gtboost-0.1.0a2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.12 Linux glibc 2.17+ x86-64 |
|
SHA-256 checksum How to use checksums |
3c10353ff66d1260040f2698ff59461f6e990068332badb053967043512dab4e
|
|
BLAKE2b-256 checksum How to use checksums |
fbfa023fc5befea4ec1c56e737b48682c991f6adf2d4a2e0589576aa83ea0e67
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp312-cp312-macosx_11_0_arm64.whl
| Download URL | gtboost-0.1.0a2-cp312-cp312-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.4 MB |
| Tags | CPython 3.12 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
a2ac0163367a76298b5d8e52ebebabf06b607fbb89d9bfbdeaeadb798a768067
|
|
BLAKE2b-256 checksum How to use checksums |
2eae99923fa8bf58ae35210c270f0cf0467629667f823458809f973be8a92cab
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp311-cp311-win_amd64.whl
| Download URL | gtboost-0.1.0a2-cp311-cp311-win_amd64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.11 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
c7c04b15a5f45340b16347fc4bee18d7fa9de1d9e25dfc819d45dc12d5887f72
|
|
BLAKE2b-256 checksum How to use checksums |
855b8fd31ae027c18c75b2847786c368833494f944f748ba96ed3f387c1ea5fa
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | gtboost-0.1.0a2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.11 Linux glibc 2.17+ x86-64 |
|
SHA-256 checksum How to use checksums |
f5c868841d907329990358857df068b7e5438aea07d69b0d115c664796ee25c1
|
|
BLAKE2b-256 checksum How to use checksums |
5ff6197172c29b3a196d8ae24e622921b3b400bc9792b4039c35019fcab5c5c7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp311-cp311-macosx_11_0_arm64.whl
| Download URL | gtboost-0.1.0a2-cp311-cp311-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.4 MB |
| Tags | CPython 3.11 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
1d451563b19d5349346ad5fd75eb8f766f4e679fcd57c3d1ad6272153bcf4a3d
|
|
BLAKE2b-256 checksum How to use checksums |
bd0f8979787c6ca07db546ab678a951cb330df78f078f97d3960ebf3f58de907
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp310-cp310-win_amd64.whl
| Download URL | gtboost-0.1.0a2-cp310-cp310-win_amd64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.10 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
c8a6b4825798d3055397ab9c2ebd468b35485ab3a97c42703ae1738cd671f0ea
|
|
BLAKE2b-256 checksum How to use checksums |
2215c9d48261b6c2707fbf80bf72382ab13dbc8f8039fc2ac93e349b9d739764
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | gtboost-0.1.0a2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.10 Linux glibc 2.17+ x86-64 |
|
SHA-256 checksum How to use checksums |
2609e81408bdb76de97d9e41932584b22ac831ecb90aded42a177a1c5dce34f0
|
|
BLAKE2b-256 checksum How to use checksums |
82dfc358b275baa06cecb5e32da03961ff0e03639705d22ad6474c713bb56fc4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logRelease files / gtboost-0.1.0a2-cp310-cp310-macosx_11_0_arm64.whl
| Download URL | gtboost-0.1.0a2-cp310-cp310-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.4 MB |
| Tags | CPython 3.10 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
62383fa9c296582e48fb4e6b6194a42a4d55f642aab63e0fba45f6d2ff4306f2
|
|
BLAKE2b-256 checksum How to use checksums |
af745859cf13e35d2cc7ca4fba2886250be576f584476067a87e0e086dc041a4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logSecurity for gtboost 0.1.0a2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 10 files in this release have been uploaded with provenance metadata.
Attestation for gtboost-0.1.0a2.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 13, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
