Struct TensorboardRecorder

pub struct TensorboardRecorder<E, R>
where
    E: Env,
    R: ReplayBufferBase,
{ /* private fields */ }

Write records to TFRecord.

Implementations

impl<E, R> TensorboardRecorder<E, R>

where E: Env, R: ReplayBufferBase,

pub fn new( log_dir: impl AsRef<Path>, model_dir: impl AsRef<Path>, check_unsupported_value: bool, ) -> Self

Construct a TensorboardRecorder.

• log_dir - Directory in which TFRecords will be stored.
• model_dir - Directory in which the trained model will be saved.
• check_unsupported_value - If true, check unsupported record value in the write() method.

Trait Implementations

impl<E, R> Recorder<E, R> for TensorboardRecorder<E, R>

where E: Env, R: ReplayBufferBase,

fn write(&mut self, record: Record)

Writes a given Record into a TFRecord.

This method handles RecordValue::Scalar and RecordValue::DateTime in the Record. Other variants will be ignored.

fn save_model(&self, base: &Path, agent: &Box<dyn Agent<E, R>>) -> Result<()>

Saves the model parameters in the local file system.

fn load_model( &self, base: &Path, agent: &mut Box<dyn Agent<E, R>>, ) -> Result<()>

Loads the model parameters from the local file system.

fn store(&mut self, record: Record)

Stores a record for later processing or aggregation.

fn flush(&mut self, step: i64)

Writes aggregated values from stored records.