Expand description
Sampling configuration types.
Structs§
- Chunking
Strategy - Controls how long text sections are chunked and weighted.
- Denoiser
Config - Configuration for the OCR denoiser that filters digit-heavy text.
- Sampler
Config - Top-level sampler configuration.
- Text
Recipe - Defines how to build a text sample from a record.
- Triplet
Recipe - Defines a triplet recipe (anchor/positive/negative selection + weighting).
Enums§
- Negative
Strategy - Strategy for picking the negative record in a triplet.
- Selector
- Selector for choosing a section or neighboring record.