Struct TrainerArgs 

pub struct TrainerArgs {
    pub env_dir: PathBuf,
    pub save_model_path: PathBuf,
    pub obs_dim: usize,
    pub act_dim: usize,
    pub buffer_size: usize,
}

Shared filesystem and shape arguments for every trainer constructor in this module.

These values are forwarded into the underlying algorithm implementations (replay sizing, logging paths, and similar runtime configuration).

Fields

• env_dir: working directory for environment-related assets the algorithm may expect.
• save_model_path: base path or directory used when persisting checkpoints (see AlgorithmTrait::save on the concrete algorithm).
• obs_dim, act_dim: observation and action space sizes used when wiring kernels and buffers.
• buffer_size: replay / trajectory buffer capacity for independent trainers; multi-agent trainers use the same field for their buffers.

Examples

use std::path::PathBuf;
use relayrl_algorithms::TrainerArgs;

let args = TrainerArgs {
    env_dir: PathBuf::from("./env"),
    save_model_path: PathBuf::from("./checkpoints"),
    obs_dim: 64,
    act_dim: 8,
    buffer_size: 1_000_000,
};

Fields

env_dir: PathBuf

Directory the algorithm treats as the environment root.

save_model_path: PathBuf

Where to persist models or session output (algorithm-specific).

obs_dim: usize

Observation dimensionality expected by the policy / value stack.

act_dim: usize

Action dimensionality (or discrete action count, depending on your kernel).

buffer_size: usize

Experience buffer capacity in transitions or slots, depending on the algorithm.

Trait Implementations

impl Clone for TrainerArgs

fn clone(&self) -> TrainerArgs

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TrainerArgs

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.