TurboLoader
High-performance ML data loading — a C++20 core with SIMD transforms, GPU kernels, and one pip install.
Real recording (tape, script): 9,469 real ImageNet JPEGs decoded → RandomResizedCrop+flip → resized → normalized, ~60k img/s per epoch on an M4 Max laptop.
How it works
The fast path does everything in one fused, GIL-released C++ pass — no worker processes, no per-sample Python, no offline format conversion:
flowchart LR
A["TAR of JPEGs<br/>(local / http / s3 / gs)"] --> B["persistent C++<br/>thread pool"]
B --> C["decode → augment → resize → normalize<br/>SIMD (NEON / AVX2 / AVX-512), fused"]
C --> D[("contiguous batch<br/>N×3×H×W float32")]
D --> E["your training step<br/>(zero-copy to torch)"]
- Fast on CPU: ~55k img/s on-the-fly (2.0×
tf.data, 2.7× PyTorch DataLoader); trains a real ResNet-18 1.05–1.17× faster end-to-end (run-dependent), ~9% above the pure-GPU floor - Fast on GPU: beats NVIDIA DALI on-the-fly (+12%, RTX 3090) and FFCV on pre-processed data (1.6–3.5×); ~757k img/s resident on Apple unified memory
- Pre-processed pipeline (TBL-RAW): decode once, mmap-serve every epoch — 586k img/s raw serve on CPU, and with serve-time RandomResizedCrop + flip (fused SIMD kernel, torchvision-parity sampler) 87k img/s = 2.7× the on-the-fly
train_augpath for the full-augmentation recipe; bit-identical batches on the identity path,float16output, DDP sharding, any hardware; fastest e2e input pipeline we've measured (3.64s epochs vs 3.76 TAR / 3.92 PyTorch, floor 3.39) - Video: hardware decode to training batches — 3.9× the best industry standard on Apple Silicon; CUDA
VideoDatasetLoadertrains a real video classifier 1.16× faster than the PyTorch+PyAV recipe (first e2e video benchmark) - Train-ready: fused
train_aug(torchvision-parity RandomResizedCrop+flip),state_dict()mid-epoch resume, pinned-memory rings, DDP sharding - Also tokens & arrays: memory-mapped
TokenDataLoader(1.9× nanoGPTget_batchto-device, zero-alloc pinned ring,device='cuda'overlapped H2D),ArrayDataLoader, andMapDataLoaderfor any__getitem__dataset - Every number is honest: interleaved medians, real consumption, corrections published — full methodology
Which loader do I use?
flowchart TD
S{{"What are you loading?"}}
S --> IMG["🖼 Images<br/>(TAR of JPEGs)"]
S --> VID["🎬 Video files"]
S --> TOK["🔤 LLM tokens"]
S --> ARR["📊 Arrays / tabular"]
S --> ANY["🐍 Anything with<br/>__getitem__"]
IMG --> Q0{"Many epochs on the<br/>same dataset?"}
Q0 -- "no / one pass" --> Q2{"Where to decode?"}
Q0 -- yes --> Q1{"Fits in GPU /<br/>unified memory?"}
Q1 -- yes --> RES["CudaResidentLoader · NVIDIA<br/>MetalResidentLoader · Apple<br/>(both ingest .tbl)"]
Q1 -- no --> TBL["preprocess_to_tbl once →<br/>DataLoader('data.tbl', train_aug=True)<br/>mmap + serve-time crop/flip"]
Q2 -- "CPU fast path (default)" --> DL["DataLoader(output_format='pytorch',<br/>image_size=N)"]
Q2 -- "NVIDIA GPU" --> CIL["CudaImageLoader(decode='nvimgcodec',<br/>return_indices=True)"]
VID --> QV{"Training on a labeled<br/>video dataset?"}
QV -- yes --> VDS["VideoDatasetLoader · NVIDIA<br/>(dir of class folders → clips)"]
QV -- "stream one file" --> MV["MetalVideoLoader · Apple<br/>CudaVideoLoader · NVIDIA"]
TOK --> TDL["TokenDataLoader<br/>(device='cuda' for GPU batches)"]
ARR --> ADL["ArrayDataLoader<br/>MetalResidentArrays (GPU gathers)"]
ANY --> MAP["MapDataLoader"]
style DL stroke-width:3px
Full decision table + lifetime rules
| You have | Use | Notes |
|---|---|---|
| A TAR of JPEGs, training on any hardware | DataLoader(..., output_format='pytorch', image_size=N) |
The default fast path — auto-fused C++ decode+resize+normalize. Start here. |
| The same, need per-sample dicts (inspection, irregular data) | DataLoader(...) (default output_format='dict') |
Several times slower; not for training loops. |
| Labels | derive from meta['indices'] / sample['filename'] |
Samples carry no label key; align an external label array by index. |
| A dataset that fits in GPU/unified memory, many epochs | CudaResidentLoader (NVIDIA) / MetalResidentLoader (Apple) |
Decode once, ~280k / 433–757k img/s per epoch. return_indices=True for labels. Both ingest .tbl. |
| Many epochs, any hardware (full aug or fixed recipe) | preprocess_to_tbl(..., image_size=192) once → DataLoader('data.tbl', image_size=160, train_aug=True) |
mmap serve, zero decode, ~zero owned RAM; serve-time RandomResizedCrop + flip via a fused SIMD kernel (2.7× the TAR train_aug path); identity path bit-identical to the TAR pipeline; dtype='float16', world_rank/world_size. |
| A pre-processed dataset larger than VRAM (NVIDIA) | CudaStreamLoader |
Fully-C++ streaming, ~140k img/s. |
| On-the-fly GPU decode (NVIDIA) | CudaImageLoader(decode='nvimgcodec', return_indices=True) |
Beats DALI; batches complete OUT of order — align labels via the returned indices. |
| On-the-fly GPU transforms (Apple) | MetalImageLoader (alias of GpuImageLoader) |
Metal decode+transforms. |
| Video files (stream one) | MetalVideoLoader (Apple) / CudaVideoLoader (NVIDIA) |
Hardware decode → training batches; iter_clips() for augmented clips. |
| Labeled video dataset, training (NVIDIA) | VideoDatasetLoader(root_dir) |
root/class_x/*.mp4 → (clips, labels, meta) CUDA batches; threaded decode + ONE fused kernel per clip. |
| LLM token streams (memmap) | TokenDataLoader |
CPU memmap is already optimal (measured); device='cuda' yields ready GPU batches (pinned ring, overlapped H2D). |
| Arrays / embeddings / tabular | ArrayDataLoader; MetalResidentArrays for GPU row gathers |
|
| WebDataset-style TARs | WebDatasetLoader |
Two lifetime rules: (1) loaders yielding zero-copy views (pin_memory=True ring,
Metal/CUDA resident + video loaders) reuse their buffers — consume or copy a batch
before advancing past the documented window; (2) GPU loaders yield
__cuda_array_interface__ objects — adopt with torch.as_tensor(x, device='cuda').
Installation
pip install turboloader # Linux x86_64/aarch64 + macOS arm64 wheels (CPU + Apple Metal)
CUDA loaders: prebuilt cu13 wheel on the latest release, or build from source — see GPU acceleration. Details: installation guide.
Quick Start
import turboloader
loader = turboloader.DataLoader(
'imagenet.tar', # TAR archive of JPEGs
batch_size=128,
image_size=224, # fixed size => one contiguous tensor per batch
output_format='pytorch', # (N, 3, H, W) float32, normalized
transform=turboloader.ImageNetNormalize(),
shuffle=True,
train_aug=True, # fused RandomResizedCrop + flip in C++
)
for images, meta in loader:
# images: numpy (N,3,224,224); torch.from_numpy(images) is zero-copy.
# meta['indices'] aligns external labels to this batch.
...
Labels: samples carry no
labelkey (a TAR is flat). UsePyTorchCompatibleLoaderfor ImageFolder-style(image, label)tuples, or align a label array viameta['indices'].
More: quickstart · per-sample dict API & transforms · tokens, arrays & any Python dataset · interactive notebook
Benchmarks (headlines)
Real data, real consumption, interleaved medians, warmup excluded. Run them
yourself — scripts in benchmarks/, full methodology + honest
caveats (and the corrections we published) in docs/benchmarks.
| Regime | TurboLoader | Best alternative | Hardware |
|---|---|---|---|
| On-the-fly CPU (decode every epoch) | ~55k img/s | tf.data ~27k · PyTorch ~20k |
M4 Max |
| On-the-fly GPU | 28.5k img/s | NVIDIA DALI 25.5k (+12%) | RTX 3090 |
| Pre-processed, fits in VRAM | ~280k img/s | FFCV ~80k (3.5×) | RTX 3090 |
| Pre-processed, streaming > VRAM | ~140k img/s | FFCV ~85k (1.6×) | RTX 3090 |
| Pre-processed, unified memory | 433–757k img/s | numpy resident ~3.7k | M4 Max |
| Pre-processed, CPU mmap (TBL-RAW, any hardware) | 586k img/s raw serve (99k np.sum-consumed w/ prefetch) | float32 RAM cache 137k (101k) at 3.3× the peak RSS + a decode-all startup | M4 Max |
| Full-aug training recipe (RandomResizedCrop + flip) | 87k img/s TBL-RAW serve-time aug (78k consumed) | on-the-fly TAR train_aug 32k (2.7×) |
M4 Max |
| Video → training batches | 2,556 f/s (3.9×) | OpenCV 657 · PyAV 535 · torchcodec 173 | M4 Max |
| End-to-end ResNet-18 training | 1.05–1.17× vs PyTorch recipe | ~9% above the pure-GPU floor | RTX 3090 |
| End-to-end VIDEO training (r3d_18) | 1.16× vs PyTorch+PyAV recipe | both decode-bound (honest) | RTX 3090 |
| LLM tokens → device | 168M tok/s (1.9×) | nanoGPT get_batch 88M |
RTX 3090 |
Honest notes worth knowing before you quote these: FFCV is faster than us
on-the-fly is impossible for it (needs .beton conversion); decord beats our CUDA
video cpu-backend on weak-CPU hosts; MetalTokenGather ties the CPU path (so we
recommend the CPU path); e2e ResNet-18 on Apple MPS is a tie because the GPU is
the bottleneck; CudaPrefetcher measured neutral in our e2e runs (decode, not
H2D, binds them). All in the full write-ups.
Architecture
flowchart TD
subgraph PY["Python (thin orchestration)"]
API["DataLoader · TokenDataLoader · ArrayDataLoader · video/GPU loaders"]
end
subgraph CPP["C++20 core — GIL released"]
MM["memory-mapped TAR / TBL v2 reader"]
POOL["persistent thread pool<br/>per-thread libjpeg-turbo decoders"]
SIMD["SIMD transforms<br/>NEON / AVX2 / AVX-512"]
BUF["fused write into the<br/>output batch buffer"]
MM --> POOL --> SIMD --> BUF
end
subgraph GPU["GPU kernels"]
METAL["Apple Metal<br/>resident · video · transforms"]
CUDA["NVIDIA CUDA + nvImageCodec<br/>resident · stream · video · clips"]
end
API --> MM
BUF --> API
API -.-> METAL
API -.-> CUDA
Deep dive: architecture · GPU acceleration · transform library (24 transforms) · TBL v2 binary format
Documentation
| Getting started | installation · quickstart · notebook · troubleshooting |
| API | API reference · transforms |
| Guides | PyTorch · TensorFlow · distributed (DDP) |
| Examples | ResNet-50 training · Lightning · DDP · GPT on tokens |
| Benchmarks | methodology + full results · video · Metal resident · e2e training |
License & Citation
MIT. If you use TurboLoader in your research:
@software{turboloader,
author = {Jain, Arnav},
title = {TurboLoader: High-Performance ML Data Loading},
year = {2026},
url = {https://github.com/ALJainProjects/TurboLoader}
}
Support: issues ·
discussions ·
PyPI · python scripts/verify_installation.py
1 maintainer
Release files for turboloader 2.38.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| turboloader-2.38.0.tar.gz | 936.4 kB | Details |
Built distributions (wheels)
Total release size: 116.7 MB
Release files / turboloader-2.38.0.tar.gz
| Download URL | turboloader-2.38.0.tar.gz |
|---|---|
| Size | 936.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
11eeefd1d826d2d1e13625479b680134cdfbc06addd98959e275ee2cfb180df6
|
|
BLAKE2b-256 checksum How to use checksums |
85a0764c43d9e4525e59fab7e4d659e962677429fd45af4ec69a1ed2c9c572af
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | turboloader-2.38.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 10.9 MB |
| Tags | CPython 3.14 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
885edcf74cb0d7fa2d737361888333d2e3cc3ffbe1101ed695fff2806cb9f4b2
|
|
BLAKE2b-256 checksum How to use checksums |
0b6838acc5ebe792ee5e09d173ec99c61218385adf8710611ba01ef8ee386024
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
| Download URL | turboloader-2.38.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl |
|---|---|
| Size | 10.5 MB |
| Tags | CPython 3.14 Linux glibc 2.27+ ARM64 Linux glibc 2.28+ ARM64 |
|
SHA-256 checksum How to use checksums |
4b726af7619a2b341b54b84e5dc0ecb0912160ac1fb7b91716464a93e228bec7
|
|
BLAKE2b-256 checksum How to use checksums |
52de40b6b3290cd7a1d597adec75c091f93f5311c1faf35fe6be5123f4a679ef
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp314-cp314-macosx_11_0_x86_64.whl
| Download URL | turboloader-2.38.0-cp314-cp314-macosx_11_0_x86_64.whl |
|---|---|
| Size | 915.3 kB |
| Tags | CPython 3.14 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
564387957bfc651518b0f04a7fd607806392475e6d2e96924202ed40f20a2195
|
|
BLAKE2b-256 checksum How to use checksums |
f2d89cc670e4b0ab5be3c2e2292e398d465e6100be6168d4b171447ebba1906a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp314-cp314-macosx_11_0_arm64.whl
| Download URL | turboloader-2.38.0-cp314-cp314-macosx_11_0_arm64.whl |
|---|---|
| Size | 900.1 kB |
| Tags | CPython 3.14 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
97fbef56356166d64cdaa40d7e456c6faeb001fa6c87c48a2c4c283afda0066f
|
|
BLAKE2b-256 checksum How to use checksums |
a4ca2bc7179b6a1f9dc1831026416efca1d57e57c794969938e30b87e3d6b681
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | turboloader-2.38.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 10.9 MB |
| Tags | CPython 3.13 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
dc0227c05fc6f74a3c0d1b209499615bb88a7ade20fc1547be14d3ae001e2862
|
|
BLAKE2b-256 checksum How to use checksums |
9d080dd0048447d7d0b3103c91aa899010b15cec69630cf469462f9d6a7dc681
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
| Download URL | turboloader-2.38.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl |
|---|---|
| Size | 10.5 MB |
| Tags | CPython 3.13 Linux glibc 2.27+ ARM64 Linux glibc 2.28+ ARM64 |
|
SHA-256 checksum How to use checksums |
b29a544ed7915aee499c7130d4c2d488a15632898a31782bf17643798863f91e
|
|
BLAKE2b-256 checksum How to use checksums |
d45f706e712e419bb00f58a2f703f5d0be363b3e3e32dab716172cca88431bb5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp313-cp313-macosx_11_0_x86_64.whl
| Download URL | turboloader-2.38.0-cp313-cp313-macosx_11_0_x86_64.whl |
|---|---|
| Size | 915.3 kB |
| Tags | CPython 3.13 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
66c86e1998358578d806354ea23780a6952f116aeb322677c8f81561432d89cb
|
|
BLAKE2b-256 checksum How to use checksums |
d570543c6965cffad0c4233adcfcb7f2050d9848529ed56ea2a3f2070e2c0eeb
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp313-cp313-macosx_11_0_arm64.whl
| Download URL | turboloader-2.38.0-cp313-cp313-macosx_11_0_arm64.whl |
|---|---|
| Size | 899.5 kB |
| Tags | CPython 3.13 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
49b7fbb9640aab6e146eee1febd5151cd2c217e6b1a116dde36048bfdf7efb64
|
|
BLAKE2b-256 checksum How to use checksums |
e638f528900305e42e5859f03830e5e81697cb2927be9a8d3b6e3e3deecc782d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | turboloader-2.38.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 10.9 MB |
| Tags | CPython 3.12 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
c9df4c694f77cdceaaa0c0448fa5368ce385c9b9e4dc5bd3c0ad4cae41b733a4
|
|
BLAKE2b-256 checksum How to use checksums |
92a4b4143cbbf44bfbb40cdf447519ca7c6b65f953bbd9e2ac136c7b33d022e4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
| Download URL | turboloader-2.38.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl |
|---|---|
| Size | 10.5 MB |
| Tags | CPython 3.12 Linux glibc 2.27+ ARM64 Linux glibc 2.28+ ARM64 |
|
SHA-256 checksum How to use checksums |
d377f1e87eea52a10704aff31000f77fc93967061504bb8acddeed7eaadd388a
|
|
BLAKE2b-256 checksum How to use checksums |
aab87eb8618854a10a78210635d5404ca4cb956a65029427349baececcca7df6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp312-cp312-macosx_11_0_x86_64.whl
| Download URL | turboloader-2.38.0-cp312-cp312-macosx_11_0_x86_64.whl |
|---|---|
| Size | 915.3 kB |
| Tags | CPython 3.12 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
4b123c6200329d53c4fd56c8ea2e9d90644469659cb452e7301af684511a03f5
|
|
BLAKE2b-256 checksum How to use checksums |
d13b328ead6c35661e465802fcb18df188cbcedbec7f86c2f08a8f6dfe4527f9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp312-cp312-macosx_11_0_arm64.whl
| Download URL | turboloader-2.38.0-cp312-cp312-macosx_11_0_arm64.whl |
|---|---|
| Size | 899.5 kB |
| Tags | CPython 3.12 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
c5b01e35c6c913ce6d81f775dc71033511dcaf6076a9d32dc57798f811387d09
|
|
BLAKE2b-256 checksum How to use checksums |
cd5bb0f2964de5e7598f9579c7c010a5938b1016ab5d96ef1501d5831c384fdb
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | turboloader-2.38.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 10.8 MB |
| Tags | CPython 3.11 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
d9bc6bad763a06a67629c261a09639bef8f87212700fe72461036c33e5e741f3
|
|
BLAKE2b-256 checksum How to use checksums |
bd9b87042e3d287c953040cd3bced5aa8c1e5a1575ebd99e9e19807dafebfcea
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
| Download URL | turboloader-2.38.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl |
|---|---|
| Size | 10.4 MB |
| Tags | CPython 3.11 Linux glibc 2.27+ ARM64 Linux glibc 2.28+ ARM64 |
|
SHA-256 checksum How to use checksums |
b0586a3df09fec7e1d328f76d90d76e70d8dccdd54ef6e5d381132a044dc6f1a
|
|
BLAKE2b-256 checksum How to use checksums |
96ac13ba6bfb86d46258417a5d403d1cdc7621e2a20f07345ccf68f09b66e1bd
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp311-cp311-macosx_11_0_x86_64.whl
| Download URL | turboloader-2.38.0-cp311-cp311-macosx_11_0_x86_64.whl |
|---|---|
| Size | 901.4 kB |
| Tags | CPython 3.11 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
d4dc300639cf22ea89ff226bb000a8e29e6d707198dce4e6a059c9408a3802df
|
|
BLAKE2b-256 checksum How to use checksums |
488763c5c8964f75d3d15afc8a28a10d5c658c414d6f86a44930c79af3b1bcd0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp311-cp311-macosx_11_0_arm64.whl
| Download URL | turboloader-2.38.0-cp311-cp311-macosx_11_0_arm64.whl |
|---|---|
| Size | 894.4 kB |
| Tags | CPython 3.11 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
36f021d12e731d3b4811f23bce0c25d540397cce225e4c9a97a3829facdae60c
|
|
BLAKE2b-256 checksum How to use checksums |
680338ae634b1364831c4bc012ede73603c8e969cce44936b2e592dc0aea0a5f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | turboloader-2.38.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 10.8 MB |
| Tags | CPython 3.10 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
9a9d3edb896627be06b56ce5f8536507130c7ed584dad9050f1442271871c9af
|
|
BLAKE2b-256 checksum How to use checksums |
f37b003d25ec418531897b2e59b4c82a82df95a855afb8ac9b6e4f6af8fa8358
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
| Download URL | turboloader-2.38.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl |
|---|---|
| Size | 10.4 MB |
| Tags | CPython 3.10 Linux glibc 2.27+ ARM64 Linux glibc 2.28+ ARM64 |
|
SHA-256 checksum How to use checksums |
d4a8409f652007fe5c131a42bd8d6612759e2b5b798e4b76670162ea79c78dd0
|
|
BLAKE2b-256 checksum How to use checksums |
ab311873580e52e10e4ec30ef83b77e263fa73590a511834181362cd067fc9f3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp310-cp310-macosx_11_0_x86_64.whl
| Download URL | turboloader-2.38.0-cp310-cp310-macosx_11_0_x86_64.whl |
|---|---|
| Size | 901.0 kB |
| Tags | CPython 3.10 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
ebb427398ee6af67fff5381a5b3a33681c72bc8b6aecd144d2ba571d625643e4
|
|
BLAKE2b-256 checksum How to use checksums |
11ed67e508c126cc7b53c34b6d90e164dbce96676fb3331ff7952873a9099e80
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency logRelease files / turboloader-2.38.0-cp310-cp310-macosx_11_0_arm64.whl
| Download URL | turboloader-2.38.0-cp310-cp310-macosx_11_0_arm64.whl |
|---|---|
| Size | 893.0 kB |
| Tags | CPython 3.10 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
0ded5602f02461a089ed566a110101955491a5dec7135c3a5878eb20e4c42b3b
|
|
BLAKE2b-256 checksum How to use checksums |
a0a11e8324b75e00abf375144de85b1deeae7b71b184f12b2573e32e0bc701d6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Sep 25, 2026.
Transparency log
