ludwig

v0.17.2 Bugfix

This release fixes issues for SREs watching stability and regressions.

Published 12d LLM Frameworks

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

computer-vision data-centric data-science deep machine-learning deeplearning

+11 more

fine-tuning learning llama llama2 llm llm-training machinelearning mistral natural-language natural-language-processing pytorch

Affected surfaces

deps breaking_upgrade

ReleasePort's take

Light signal

editorial:auto 11d

ReleasePort Layer 1 v0.17.2 fixes the LLM fine‑tuning crash caused by incompatible TorchAO and PyTorch versions.

Why it matters: If you use Ludwig's LLM fine‑tuning pipeline, update to v0.17.2 immediately; it resolves a critical incompatibility between torchao and torch that causes crashes.

Summary

AI summary

Fixed LLM fine‑tuning crash caused by incompatible TorchAO and PyTorch versions.

Changes in this release

Type	Severity	Summary	CVE
Feature	Medium	Change AutoML default tabular combiner from tabnet to ft_transformer for better performance. Change AutoML default tabular combiner from tabnet to ft_transformer for better performance. Source: llm_adapter@2026-05-23 Confidence: low	—
Feature	Medium	Add ft_transformer and tabtransformer as AutoML combiner defaults with tuned hyperopt spaces. Add ft_transformer and tabtransformer as AutoML combiner defaults with tuned hyperopt spaces. Source: llm_adapter@2026-05-23 Confidence: low	—
Dependency	Medium	Increase lower bounds for torch, torchaudio, torchvision, transformers, and torchao dependencies. Increase lower bounds for torch, torchaudio, torchvision, transformers, and torchao dependencies. Source: llm_adapter@2026-05-23 Confidence: high	—
Dependency	Medium	Release new Docker images ludwig:0.17.2, ludwig-gpu:0.17.2, ludwig-ray:0.17.2, ludwig-ray-gpu:0.17.2 with torch 2.12.0. Release new Docker images ludwig:0.17.2, ludwig-gpu:0.17.2, ludwig-ray:0.17.2, ludwig-ray-gpu:0.17.2 with torch 2.12.0. Source: llm_adapter@2026-05-23 Confidence: low	—
Bugfix	Medium	Fix LLM fine-tuning crash caused by incompatible torchao and torch versions. Fix LLM fine-tuning crash caused by incompatible torchao and torch versions. Source: llm_adapter@2026-05-23 Confidence: high	—
Bugfix	Medium	Update Docker images to torch 2.12.0, torchvision 0.27.0, torchaudio 2.11.0 eliminating mismatches. Update Docker images to torch 2.12.0, torchvision 0.27.0, torchaudio 2.11.0 eliminating mismatches. Source: llm_adapter@2026-05-23 Confidence: high	—

Full changelog

What's changed

Bug fixes

Fix LLM fine-tuning crash (AttributeError: module 'torch.utils._pytree' has no attribute 'register_constant') caused by torchao>=0.11 being installed against torch<2.11 — tighten all torch-family lower bounds to prevent incompatible combinations (#4170)
Update all Docker images from torch==2.6.0 to torch==2.12.0 / torchvision==0.27.0 / torchaudio==2.11.0, eliminating the mismatch between the Ray base image's bundled torch and the torchao version pulled in by ludwig[llm]

New features

AutoML default tabular combiner changed from tabnet to ft_transformer — better out-of-the-box performance on most tabular datasets
Added ft_transformer and tabtransformer to AutoML combiner defaults with tuned hyperopt parameter spaces

Dependency updates

| Package | Before | After |
|---|---|---|
| torch | >=2.0 | >=2.11 |
| torchaudio | >=2.0 | >=2.11 |
| torchvision | >=0.15 | >=0.26 |
| transformers | >=4.36 | >=5.0 |
| torchao (llm extra) | >=0.9.0 | >=0.17.0 |

Docker images

ludwigai/ludwig:0.17.2 — CPU, torch 2.12.0
ludwigai/ludwig-gpu:0.17.2 — CUDA 12.4, torch 2.12.0+cu124
ludwigai/ludwig-ray:0.17.2 — CPU + Ray 2.54, torch 2.12.0
ludwigai/ludwig-ray-gpu:0.17.2 — CUDA 12.4 + Ray 2.54, torch 2.12.0+cu124

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track ludwig

Get notified when new releases ship.

About ludwig

Low-code framework for building custom LLMs, neural networks, and other AI models

All releases →