Crate brainwires_finetune 

Brainwires Fine-tune

Cloud fine-tuning for the Brainwires Agent Framework.

Supports cloud fine-tuning (OpenAI, Together, Fireworks, Anyscale, Bedrock, Vertex) plus dataset pipelines. Local adapter training (LoRA, QLoRA, DoRA) and training-from-scratch live in the sibling rullama workspace as rullama-finetune and rullama-training.

Re-exports

pub use config::AdapterMethod;

pub use config::AlignmentMethod;

pub use config::LoraConfig;

pub use config::LrScheduler;

pub use config::TrainingHyperparams;

pub use error::TrainingError;

pub use types::DatasetId;

pub use types::TrainingJobId;

pub use types::TrainingJobStatus;

pub use types::TrainingJobSummary;

pub use types::TrainingMetrics;

pub use types::TrainingProgress;

pub use cloud::CloudFineTuneConfig;

pub use cloud::FineTuneProvider;

pub use cloud::FineTuneProviderFactory;

pub use manager::TrainingManager;

Modules

cloud

Cloud fine-tuning providers.

config

Training configuration and hyperparameters.

datasets

Dataset pipelines (absorbed from brainwires-datasets).

error

Training error types.

manager

Training job management.

types

Training job types and status.