Unassigned · 2026-10-02
Autoregressive Drillhole Modelling Under Distribution Shift
Yihao Ding, Daniel Yitian Su, Yiran Zhang, Christopher M. Gonzalez, Wei Liu
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
The authors introduce DrillBench, a public benchmark of 49,671 Western Australian drillholes for next-layer prediction and free-running autoregressive stratigraphic generation under a transfer spectrum (in-distribution, graded spatial shift, and cross-province). They find a transfer boundary: spatial/geochemical conditioning helps locally but degrades under stronger shift, while lithology-sequence autoregressive models transfer more robustly. They propose a backbone-agnostic recipe combining large-scale pretraining with spatial retrieval of neighboring lithology.
Why it matters
Drillhole modelling is formulated as a strictly one-directional generative process (predict the next stratum from the already drilled sequence). DrillBench provides a graded transfer evaluation protocol, enabling study of when location/assay conditioning helps or harms generalisation, and how pretraining and retrieval interact with autoregressive sequence modelling.
Method
- Introduce DrillBench (49,671 drillholes) with two tasks: next-layer prediction from ground-truth history and free-running autoregressive generation from a prefix, evaluated across increasing spatial/geological shift.
- Benchmark classical, geostatistical, and neural models to identify a transfer boundary between conditioning (local gains, sharp degradation) and sequence-only autoregressive transfer (more robust).
- Develop a backbone-agnostic recipe: layer-level autoregressive backbone, large-scale pretraining on WA-DASC-full with auxiliary geochemistry, and spatial retrieval of neighboring lithology at inference without exposing absolute coordinates/assays.
Abstract (from arXiv)
Autoregressive modelling has achieved remarkable success in language and sequence tasks by learning to predict future states from previous observation. Mineral-exploration drillholes provide a natural but largely unexplored setting for this paradigm: as drilling proceeds, lithology is revealed sequentially from shallow to deep, making prediction of deeper strata inherently autoregressive. Existing drillhole modelling, however, is dominated by spatial interpolation and reconstruction, or largely rely on masked modelling, leaving strictly autoregressive prediction largely underexplored. We introduce DrillBench, a benchmark of 49,671 Western Australian drillholes for next-layer prediction and autoregressive stratigraphic generation across a graded transfer spectrum, from local prediction through spatial shift to cross geological province transfer. Benchmarking classical, geostatistical, and neural models reveals a clear \emph{transfer boundary}: spatial and geochemical conditioning provides large local gains but deteriorates sharply under stronger shift, whereas lithology-sequence autoregressive models transfer more robustly. Guided by this finding, we develop a backbone-agnostic recipe combining large-scale pretraining on historical drillholes with spatial retrieval of neighbouring lithology. Retrieval is most effective in weathered cover, when local spatial continuity remains informative, whereas pretraining contributes more strongly in bedrock and under broader geological shift. Together, they retain strong local performance while improving generalisation under spatial and cross-province shift, most markedly on the most distant splits. The benchmark and code are available at https://github.com/yihaoding/drillbench.
Related papers
- GeoState: Learning View-Decoupled Geometry States for 3D World Modeling
- Looking Ahead: Modeling Future Evolution in Autoregressive Video Generation
- EpicWorldModel: Exploration-driven Planning with Latent World Models
- Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
- Glance Forcing: Making Bidirectional Diffusion Autoregressive at a Glance
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-01 · Indexed by Pub-AI: 2026-10-02 UTC
Autoregressive Drillhole Modelling Under Distribution Shift
Preprint posted: 2026-10-01 · Original record ↗
- arxiv
- 2610.01204
Explicit source versions (1)
- v1 · 2026-10-01
- Preprint available
Source date: 2026-10-01 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Autoregressive modelling has achieved remarkable success in language and sequence tasks by learning to predict future states from previous observation. Mineral-exploration drillholes provide a natural but largely unexplored setting for this paradigm: as drilling proceeds, lithology is revealed sequentially from shallow to deep, making prediction of deeper strata inherently autoregressive. Existing drillhole modelling, however, is dominated by spatial interpolation and reconstruction, or largely rely on masked modelling, leaving strictly autoregressive prediction largely underexplored. We introduce DrillBench, a benchmark of 49,671 Western Australian drillholes for next-layer prediction and autoregressive stratigraphic generation across a graded transfer spectrum, from local prediction through spatial shift to cross geological province transfer. Benchmarking classical, geostatistical, and neural models reveals a clear \emph{transfer boundary}: spatial and geochemical conditioning provides large local gains but deteriorates sharply under stronger shift, whereas lithology-sequence autoregressive models transfer more robustly. Guided by this finding, we develop a backbone-agnostic recipe combining large-scale pretraining on historical drillholes with spatial retrieval of neighbouring lithology. Retrieval is most effective in weathered cover, when local spatial continuity remains informative, whereas pretraining contributes more strongly in bedrock and under broader geological shift. Together, they retain strong local performance while improving generalisation under spatial and cross-province shift, most markedly on the most distant splits. The benchmark and code are available at https://github.com/yihaoding/drillbench.
