blazen-train-tune 0.6.75

AutoML / hyperparameter search for Blazen fine-tuning, layered on blazen-train
Documentation

AutoML / hyperparameter search on top of blazen-train.

This crate is intentionally separate from blazen-train itself so that the heavy training engine (candle, hf-hub, tokenizers) doesn't have to be compiled in to use the searchers — the Evaluator trait is the only coupling point, and callers wire their own training loop into it.

Pieces:

  • [space] — SearchSpace + Distribution (Categorical / IntUniform / Uniform / LogUniform / Discrete). The space defines what hyperparams exist and what each one's prior looks like.
  • [trial] — Trial record + status state machine.
  • [searcher] — Searcher trait + three real implementations: RandomSearch, GridSearch, and TpeSearch (Tree-structured Parzen Estimator, Bergstra et al. 2011, Algorithm 1).
  • [journal] — JSONL append-only TrialJournal for crash recovery.
  • [runner] — Runner glues a SearchSpace + Searcher + Evaluator
    • TrialJournal into a budgeted (max-trials / time-budget) loop, optionally fan-out across N tokio workers.

See examples/lora_sft_search.rs for an end-to-end LoRA SFT search.

ML acronyms (LoRA, TPE, KDE, SFT, ...) saturate the docs; backticking each one would hurt readability without adding clarity.