Module ml_v1 

Structs

ExperimentInfo

Note: needs to be Copy due to wasmtime constraints.

ExperimentInfoBits

bytemuck-generated type for internal purposes only.

FixedBufferString

A variable length string that can be up to N bytes long.

Metric

PollSimple

Indicates if a future has been completed. Should be an enum, but kept as a struct for symmetry with PollVec

PollSimpleBits

bytemuck-generated type for internal purposes only.

PollVec

Indicates if a future that outputs a Vec<T> has been completed. Should be an enum, but the ffi macros don’t support that

PollVecBits

bytemuck-generated type for internal purposes only.

Enums

EpisodeState

ExperimentStatus

SnapshotFormat

Type used to describe a data object created using either create_data or by the host.

Statics

API

FFI API

Functions

can_connect_to_hive

Ok(true) if we can connect to hive.

download_metrics

Downloads metrics asynchronously

download_snapshot

Retrieves a trained snapshot.

drop_future

Drops a future handle.

experiment_action_count

Number of actions that an experiment outputs

experiment_feature_count

Number of features that an experiment takes as input

list_experiments

Lists all experiments on the machine learning server.

ml__can_connect_to_hive^⚠

Ok(true) if we can connect to hive.

ml__download_metrics^⚠

Downloads metrics asynchronously

ml__download_snapshot^⚠

Retrieves a trained snapshot.

ml__drop_future^⚠

Drops a future handle.

ml__experiment_action_count^⚠

Number of actions that an experiment outputs

ml__experiment_feature_count^⚠

Number of features that an experiment takes as input

ml__list_experiments^⚠

Lists all experiments on the machine learning server.

ml__onnx_to_cervo^⚠

Convert a raw ONNX asset to Cervo format

ml__poll_future_bool^⚠

Polls a future that outputs a boolean

ml__poll_future_training_handle^⚠

ml__poll_future_vec_experiment_info^⚠

ml__poll_future_vec_metric^⚠

ml__poll_future_vec_u8^⚠

ml__push_inference_observation^⚠

Pushes an observation into a queue to be used later.

ml__push_inference_observation_with_metadata^⚠

Pushes an observation with metadata into a queue to be used later.

ml__push_training_observation^⚠

Pushes an observation into a queue to be used later.

ml__push_training_observation_with_metadata^⚠

Pushes an observation with metadata into a queue to be used later.

ml__start_inference^⚠

ml__start_training^⚠

Starts training.

ml__start_training_from_checkpoint^⚠

Starts training from a checkpoint.

ml__stop_inference^⚠

ml__stop_training^⚠

Stops training with a given handle and destroys it.

ml__submit_inference_observations^⚠

Submits observations previously pushed using push_inference_observation. out_actions should be a slice with length num_actions * observation count. out_values should be a slice with length observation count.

ml__submit_training_augmented_observations^⚠

Submits observations previously pushed using push_training_observation as augmented observations.

ml__submit_training_demonstration_observations^⚠

Submits observations previously pushed using push_training_observation as demonstration observations.

ml__submit_training_observations^⚠

Submits observations previously pushed using push_training_observation. out_actions should be a slice with length num_actions * observation count. out_values should be a slice with length observation count.

ml__take_future_bool^⚠

Transfers the data of a completed future to the module side.

ml__take_future_training_handle^⚠

ml__take_future_vec_experiment_info^⚠

ml__take_future_vec_metric^⚠

ml__take_future_vec_u8^⚠

onnx_to_cervo

Convert a raw ONNX asset to Cervo format

poll_future_bool

Polls a future that outputs a boolean

poll_future_training_handle

poll_future_vec_experiment_info

poll_future_vec_metric

poll_future_vec_u8

push_inference_observation

Pushes an observation into a queue to be used later.

push_inference_observation_with_metadata

Pushes an observation with metadata into a queue to be used later.

push_training_observation

Pushes an observation into a queue to be used later.

push_training_observation_with_metadata

Pushes an observation with metadata into a queue to be used later.

start_inference

start_training

Starts training.

start_training_from_checkpoint

Starts training from a checkpoint.

stop_inference

stop_training

Stops training with a given handle and destroys it.

submit_inference_observations

Submits observations previously pushed using push_inference_observation. out_actions should be a slice with length num_actions * observation count. out_values should be a slice with length observation count.

submit_training_augmented_observations

Submits observations previously pushed using push_training_observation as augmented observations.

submit_training_demonstration_observations

Submits observations previously pushed using push_training_observation as demonstration observations.

submit_training_observations

Submits observations previously pushed using push_training_observation. out_actions should be a slice with length num_actions * observation count. out_values should be a slice with length observation count.

take_future_bool

Transfers the data of a completed future to the module side.

take_future_training_handle

take_future_vec_experiment_info

take_future_vec_metric

take_future_vec_u8

Type Aliases

FutureHandle

A handle to a future on the engine-side.

InferenceHandle

TrainingHandle