MachineLearningLM: Synthetic ML Tasks from Structural Causal Models

A pretraining corpus for MachineLearningLM, a framework to equip large language models with in-context machine learning capabilities. The dataset consists of ML tasks synthesized from millions of structural causal models, spanning various shot counts up to 1,024. It was created by author 'eshmoideas' and last updated on June 15, 2026.

Use Cases

• Pretrain LLMs for in-context learning based on the described corpus of synthesized ML tasks.
• Benchmark LLM performance on standard ML tasks using the described examples spanning up to 1,024 shots.
• Study the effect of example count on in-context learning based on the dataset's span of various shot counts.

Strengths

• Synthesized from millions of structural causal models, suggesting a large generation base.
• Designed to span various shot counts up to 1,024, enabling study of in-context learning scaling.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface user 'eshmoideas'.

Collection Method

Synthesized from structural causal models.

Freshness

Last updated 2026-06-15 11:36:39; freshness should be verified.