Lean 4 → MLIR → GPU

The book: Verified Deep Learning with Lean 4 (PDF) — the interactive proof blueprint is the book: every theorem clickable, from the pdiv primitives to the whole-network backward passes.

Lean 4 as a specification language for neural networks. Declare the architecture in Lean, render one StableHLO graph — forward, loss, backward and optimizer fused — from proofs that the backward is the Jacobian-transpose of the forward, hand the graph to a trusted lowerer, train end to end. No Python at run time and no autograd library: the gradients are derived at codegen time, in Lean, and machine-checked over the reals.

Companion code for Verified Deep Learning with Lean 4, forthcoming from Apress (Springer Nature) as the follow-up to Convolutional Neural Networks with Swift for TensorFlow (2021).

Current version: v0.7.0. Release history in CHANGELOG.md.

The tour

Four commands, one per scale, in the order the book meets the nets, then the demos. The numbers are the book's, from the verified XLA path on one RTX 4060 Ti unless the row says otherwise; the two Imagenette side quests the book does not quote (ResNet-50, MobileNetV4) are medians of five seeds. Setup is the book's Getting started, one track per tier; the short form is

curl https://raw.githubusercontent.com/leanprover/elan/master/elan-init.sh -sSf | sh   # Lean 4
python3 -m venv .venv && . .venv/bin/activate && pip install jax-cuda12-pjrt          # the XLA plugin only (jax-rocm7-pjrt on AMD)
gcc -fPIC -O2 -shared ffi/pjrt_ffi.c -ldl -o ffi/libpjrt_ffi.so                        # the shim; needs nothing but libc
lake exe cache get && ./download_mnist.sh && lake run mnist                            # Mathlib oleans, MNIST, tier 1

tier	command	trains	the number	chapter
1	lake run mnist	linear, MLP and CNN on MNIST, 12 epochs each (~1 min)	92.10 · 97.81 · 98.77 %	1 · 2 · 3
2	lake run cifar	the wide 8-conv net on CIFAR-10: SGD / momentum / AdamW × no-BN / BN, 40 epochs at a constant lr (~19 min)	76.3 % — BN + momentum, median of five	4
3	lake run imagenette	seven nets on Imagenette at 224², 80 epochs AdamW, book order (~9 h)	R34 89.50 · R50 89.71 · MNv2 89.25 · MNv4-Conv-M 86.24 · B0 89.96 · ConvNeXt-T 85.07 · ViT-Tiny 68.74 %	5 · 6 · 7 · 8 · 9
4	lake run imagenet	the same nets on ImageNet-1k, 4× 4060 Ti, weeks of wall-clock; bare it prints the plan and every row's estimate, start runs it	R34 74.16 · R50 (RSB-A3) 78.26 · MNv2 71.90 · MNv4-Conv-M 75.48 · B0 77.15 · ConvNeXt-T 81.53 · ViT-Tiny 72.31 %	Track 4

The demos ride on the chapter nets; demos/README.md has the command, the figure and the reasoning for each.

demo	command	the number
segmentation	lake exe unet-brats-r34, then brats-predict	BraTS mIoU 0.742 — a ResNet-34 encoder under a UNet
detection	lake exe yolov1-visdrone-fpn	VisDrone mAP@0.5 0.2363 — ResNet-34 + FPN at 448
diffusion	lake exe mnist-ddpm-train, then mnist-ddpm-sample	the sample grid
language	lake exe tinygpt-shakespeare (also bigram-shakespeare, tinystories)	1.45 nats/char

Everything else in the repository is the lab — apps/baselines/, the ablation and robustness exes, the tests, the Bestiary — the evidence behind these numbers, one level down. lakefile.lean is grouped the same way: the tour first, then the lab by home directory.

The proofs

Every layer's backward is proven to be the Jacobian-transpose of its forward over the exact reals (Mathlib's fderiv), composed up to whole-network VJPs for ResNet-34, MobileNetV2, EfficientNet-B0, ConvNeXt-T and ViT-Tiny, with zero project axioms. For every chapter net the committed train-step render in verified_mlir/ is tied to those proofs at the denotational level: each emitted parameter-update node denotes the certified descent step, and the tiers train on exactly those bytes. What stays trusted is the ℝ→Float32 numerics, the per-op text printing, and the lowerer with its runtime. The book's On Verification appendix is the full argument, gap by gap; LeanMlir/Proofs/README.md is the file-level map.

Check them without a GPU:

lake exe cache get           # Mathlib oleans, ~30 s
lake build ProofsMinimal     # the smallest end-to-end tie, ~1 min
lake build Certs             # every certificate CI checks (the long one)

tests/comparator/run.sh re-runs Lean's kernel typechecker over the headline theorems independently, and tests/comparator/Challenge.lean imports Mathlib and nothing else, so those can be read and checked without reading a line of this project.

Two lowerers, one graph

Training runs through XLA/PJRT, and every number above comes from it. IREE is the second trusted lowerer: it is what the differential oracle (tests/vjp_oracle/) lowers the Lean side through, so that agreement with the JAX reference is evidence from two independent compilers, and each lake run tier has an -iree twin. Building it is historical/IREE_BUILD.md.

Where things are

• LeanMlir/Proofs/ — the proofs, chapter by chapter; verified_mlir/ — the committed renders the tiers train on
• apps/ — one Main per exe, by tier; demos/ — the demos; Bestiary/ — 41 read-only NetSpec catalogue entries, Part 2 of the book
• jax/ — the JAX reference implementations the ImageNet path is ported from, and the oracle's ground truth
• scripts/jobs/ and scripts/supervise.sh — the ImageNet jobs (lake run <job> runs one); runs/ — the logs and READMEs behind the numbers
• historical/ — the two earlier phases (historical/mnist-lean4/, pure Lean 4 with a C BLAS; historical/mlir_poc/, the Python exporters), the survey this README used to be (README_survey.md), and the reference notes: RESULTS.md (per-epoch histories), BENCHMARK.md, CUDA.md, ROCM.md, IREE_BUILD.md
• blueprint/ — the book's source

Citing this work

@software{koonce2026,
  author  = {Brett Koonce},
  title   = {Verified Deep Learning with Lean 4: Formal Backpropagation from MLP to Attention, via MLIR},
  url     = {https://github.com/brettkoonce/lean4-mlir},
  doi     = {10.5281/zenodo.20402133},
  version = {0.7.0},
  year    = {2026},
}