Index

A

accumulate_grads (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) action_log_std (rl8.models.DefaultContinuousModel attribute) (rl8.models.DefaultContinuousRecurrentModel attribute) action_mean (rl8.models.DefaultContinuousModel attribute) (rl8.models.DefaultContinuousRecurrentModel attribute) action_spec (rl8.env.Env attribute) (rl8.env.GenericEnv attribute) (rl8.models.GenericModel attribute) (rl8.models.GenericModelBase attribute) (rl8.models.GenericRecurrentModel attribute) (rl8.policies.GenericPolicyBase property) Algorithm (class in rl8.algorithms)

algorithm (rl8.trainers.GenericTrainerBase attribute) And (class in rl8.conditions) append() (rl8.nn.modules.skip.SequentialSkipConnection method) apply_all() (rl8.views.PaddedRollingWindow static method) (rl8.views.RollingWindow static method) (rl8.views.View static method) (rl8.views.ViewRequirement method) apply_last() (rl8.views.PaddedRollingWindow static method) (rl8.views.RollingWindow static method) (rl8.views.View static method) (rl8.views.ViewRequirement method) apply_view_requirements() (rl8.models.Model method) attention (rl8.nn.modules.attention.CrossAttention attribute) (rl8.nn.modules.attention.SelfAttention attribute)

B

binary_mask_to_float_mask() (in module rl8.nn.functional) bounds (rl8.env.DummyEnv attribute) buffer (rl8.algorithms.GenericAlgorithmBase attribute)

buffer_spec (rl8.algorithms.GenericAlgorithmBase attribute) build() (rl8.trainers.config.TrainConfig method) (rl8.trainers.TrainConfig method)

C

Categorical (class in rl8.distributions) clip_param (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) coeff (rl8.schedulers.EntropyScheduler attribute) (rl8.schedulers.LRScheduler attribute) collect() (rl8.algorithms.Algorithm method) (rl8.algorithms.GenericAlgorithmBase method) (rl8.algorithms.RecurrentAlgorithm method)

Condition (class in rl8.conditions) conditions (rl8.conditions.And attribute) config (rl8.env.Env attribute) (rl8.models.GenericModelBase attribute) ConstantScheduler (class in rl8.schedulers) ContinuousDummyEnv (class in rl8.env) CrossAttention (class in rl8.nn.modules.attention)

D

default_dist_cls() (rl8.distributions.Distribution static method) default_model_cls() (rl8.models.Model static method) (rl8.models.RecurrentModel static method) DefaultContinuousModel (class in rl8.models) DefaultContinuousRecurrentModel (class in rl8.models) DefaultDiscreteModel (class in rl8.models) DefaultDiscreteRecurrentModel (class in rl8.models) deterministic_sample() (rl8.distributions.Distribution method) (rl8.distributions.SquashedNormal method) (rl8.distributions.TorchDistributionWrapper method) device (rl8.env.Env attribute) (rl8.models.GenericModelBase property) (rl8.policies.GenericPolicyBase property) (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

DiscreteDummyEnv (class in rl8.env) dist (rl8.distributions.TorchDistributionWrapper attribute) Distribution (class in rl8.distributions) distribution_cls (rl8.policies.GenericPolicyBase attribute) (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) drop_size (rl8.models.Model property) (rl8.views.ViewRequirement property) drop_size() (rl8.views.PaddedRollingWindow static method) (rl8.views.RollingWindow static method) (rl8.views.View static method) dropout (rl8.nn.modules.embeddings.PositionalEmbedding attribute) dual_clip_param (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) DummyEnv (class in rl8.env)

E

enable_amp (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) entropy() (rl8.distributions.Distribution method) (rl8.distributions.SquashedNormal method) (rl8.distributions.TorchDistributionWrapper method) entropy_coeff (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) entropy_scheduler (rl8.algorithms.GenericAlgorithmBase attribute)

EntropyScheduler (class in rl8.schedulers) Env (class in rl8.env) env (rl8.algorithms.GenericAlgorithmBase attribute) env_cls (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) env_config (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) EnvFactory (class in rl8.env) eval() (rl8.trainers.GenericTrainerBase method)

F

feature_head (rl8.models.DefaultDiscreteRecurrentModel attribute) feature_model (rl8.models.DefaultDiscreteModel attribute) features (rl8.distributions.Distribution attribute) float_mask_to_binary_mask() (in module rl8.nn.functional) forward() (rl8.models.DefaultContinuousModel method) (rl8.models.DefaultContinuousRecurrentModel method) (rl8.models.DefaultDiscreteModel method) (rl8.models.DefaultDiscreteRecurrentModel method) (rl8.models.Model method) (rl8.models.RecurrentModel method) (rl8.nn.modules.activations.SquaredReLU method) (rl8.nn.modules.attention.CrossAttention method) (rl8.nn.modules.attention.PointerNetwork method) (rl8.nn.modules.attention.SelfAttention method) (rl8.nn.modules.attention.SelfAttentionStack method) (rl8.nn.modules.embeddings.PositionalEmbedding method) (rl8.nn.modules.module.Module method) (rl8.nn.modules.perceiver.PerceiverIOLayer method) (rl8.nn.modules.perceiver.PerceiverLayer method) (rl8.nn.modules.skip.SequentialSkipConnection method)

from_file() (rl8.trainers.config.TrainConfig class method) (rl8.trainers.TrainConfig class method)

G

gae_lambda (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) gamma (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) generalized_advantage_estimate() (in module rl8.nn.functional) GenericAlgorithmBase (class in rl8.algorithms) GenericEnv (class in rl8.env)

GenericModel (class in rl8.models) GenericModelBase (class in rl8.models) GenericPolicyBase (class in rl8.policies) GenericRecurrentModel (class in rl8.models) GenericTrainerBase (class in rl8.trainers) get_activation() (in module rl8.nn.modules.activations) grad_scaler (rl8.algorithms.GenericAlgorithmBase attribute)

H

HitsLowerBound (class in rl8.conditions) HitsUpperBound (class in rl8.conditions) horizon (rl8.env.Env attribute) (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

horizons_per_env_reset (rl8.algorithms.GenericAlgorithmBase property) (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) hparams (rl8.algorithms.GenericAlgorithmBase attribute)

I

in_features (rl8.nn.modules.skip.SequentialSkipConnection property) init_states() (rl8.models.RecurrentModel method) (rl8.policies.RecurrentPolicy method)

InterpScheduler (class in rl8.schedulers)

K

key (rl8.conditions.HitsLowerBound attribute) (rl8.conditions.HitsUpperBound attribute) (rl8.conditions.Plateaus attribute) (rl8.conditions.StopsDecreasing attribute) (rl8.conditions.StopsIncreasing attribute)

kind (rl8.nn.modules.skip.SequentialSkipConnection attribute) kv_norm (rl8.nn.modules.attention.CrossAttention attribute)

L

latent_model (rl8.models.DefaultContinuousModel attribute) load_context() (rl8.policies.MLflowPolicyModel method) (rl8.policies.MLflowRecurrentPolicyModel method) logp() (rl8.distributions.Distribution method) (rl8.distributions.SquashedNormal method) (rl8.distributions.TorchDistributionWrapper method) losses (rl8.conditions.Plateaus attribute) (rl8.conditions.StopsDecreasing attribute) (rl8.conditions.StopsIncreasing attribute)

lower_bound (rl8.conditions.HitsLowerBound attribute) lr_scheduler (rl8.algorithms.GenericAlgorithmBase attribute) LRScheduler (class in rl8.schedulers) lstm (rl8.models.DefaultContinuousRecurrentModel attribute) (rl8.models.DefaultDiscreteRecurrentModel attribute)

M

mask_from_lengths() (in module rl8.nn.functional) masked_avg() (in module rl8.nn.functional) masked_categorical_sample() (in module rl8.nn.functional) masked_log_softmax() (in module rl8.nn.functional) masked_max() (in module rl8.nn.functional) max_ (rl8.conditions.StopsIncreasing attribute) max_grad_norm (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) max_horizon (rl8.env.Env attribute) (rl8.env.EnvFactory attribute) max_num_envs (rl8.env.Env attribute) (rl8.env.EnvFactory attribute) memory_stats() (rl8.algorithms.GenericAlgorithmBase method) method (rl8.views.ViewRequirement attribute) min_ (rl8.conditions.StopsDecreasing attribute) MLflowPolicyModel (class in rl8.policies) MLflowRecurrentPolicyModel (class in rl8.policies) MLP (class in rl8.nn.modules.mlp) Model (class in rl8.models) model (rl8.distributions.Distribution attribute) (rl8.policies.GenericPolicyBase attribute) model_cls (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) model_config (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

ModelFactory (class in rl8.models) module rl8 rl8.algorithms rl8.conditions rl8.distributions rl8.env rl8.models rl8.nn rl8.nn.functional rl8.nn.modules rl8.nn.modules.activations rl8.nn.modules.attention rl8.nn.modules.embeddings rl8.nn.modules.mlp rl8.nn.modules.module rl8.nn.modules.perceiver rl8.nn.modules.skip rl8.policies rl8.schedulers rl8.trainers rl8.trainers.config rl8.views Module (class in rl8.nn.modules.module)

N

Normal (class in rl8.distributions) normalize_advantages (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) normalize_rewards (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

num_envs (rl8.env.Env attribute) (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) num_sgd_iters (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

O

observation_spec (rl8.env.Env attribute) (rl8.env.GenericEnv attribute) (rl8.models.GenericModel attribute) (rl8.models.GenericModelBase attribute) (rl8.models.GenericRecurrentModel attribute) (rl8.policies.GenericPolicyBase property) old_value (rl8.conditions.Plateaus attribute)

optimizer (rl8.algorithms.GenericAlgorithmBase attribute) (rl8.schedulers.LRScheduler attribute) optimizer_cls (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) optimizer_config (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) out_features (rl8.nn.modules.skip.SequentialSkipConnection property)

P

pad_last_sequence() (in module rl8.views) pad_whole_sequence() (in module rl8.views) PaddedRollingWindow (class in rl8.views) params (rl8.algorithms.GenericAlgorithmBase property) patience (rl8.conditions.Plateaus attribute) (rl8.conditions.StopsDecreasing attribute) (rl8.conditions.StopsIncreasing attribute) pe (rl8.nn.modules.embeddings.PositionalEmbedding attribute) PerceiverIOLayer (class in rl8.nn.modules.perceiver)

PerceiverLayer (class in rl8.nn.modules.perceiver) Plateaus (class in rl8.conditions) PointerNetwork (class in rl8.nn.modules.attention) Policy (class in rl8.policies) policy (rl8.algorithms.GenericAlgorithmBase attribute) PositionalEmbedding (class in rl8.nn.modules.embeddings) ppo_losses() (in module rl8.nn.functional) predict() (rl8.policies.MLflowPolicyModel method) (rl8.policies.MLflowRecurrentPolicyModel method)

Q

q_norm (rl8.nn.modules.attention.CrossAttention attribute)

R

recurrent (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) RecurrentAlgorithm (class in rl8.algorithms) RecurrentModel (class in rl8.models) RecurrentModelFactory (class in rl8.models) RecurrentPolicy (class in rl8.policies) RecurrentTrainer (class in rl8.trainers) reset() (rl8.env.DummyEnv method) (rl8.env.Env method) rl8 module rl8.algorithms module rl8.conditions module rl8.distributions module rl8.env module rl8.models module rl8.nn module rl8.nn.functional module rl8.nn.modules module

rl8.nn.modules.activations module rl8.nn.modules.attention module rl8.nn.modules.embeddings module rl8.nn.modules.mlp module rl8.nn.modules.module module rl8.nn.modules.perceiver module rl8.nn.modules.skip module rl8.policies module rl8.schedulers module rl8.trainers module rl8.trainers.config module rl8.views module rolling_window() (in module rl8.views) RollingWindow (class in rl8.views) rtol (rl8.conditions.Plateaus attribute) run() (rl8.trainers.GenericTrainerBase method)

S

sample() (rl8.distributions.Distribution method) (rl8.distributions.SquashedNormal method) (rl8.distributions.TorchDistributionWrapper method) (rl8.policies.Policy method) (rl8.policies.RecurrentPolicy method) save() (rl8.policies.GenericPolicyBase method) (rl8.policies.Policy method) (rl8.policies.RecurrentPolicy method) schedule (rl8.schedulers.StepScheduler attribute) Scheduler (class in rl8.schedulers) scheduler (rl8.schedulers.EntropyScheduler attribute) (rl8.schedulers.LRScheduler attribute) SelfAttention (class in rl8.nn.modules.attention) SelfAttentionStack (class in rl8.nn.modules.attention) seq_len (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) seqs_per_state_reset (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) SequentialSkipConnection (class in rl8.nn.modules.skip) sgd_minibatch_size (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) shift (rl8.views.ViewRequirement attribute) shuffle_minibatches (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) skip_connection (rl8.nn.modules.attention.CrossAttention attribute) (rl8.nn.modules.attention.SelfAttention attribute)

skip_connection() (in module rl8.nn.functional) SquaredReLU (class in rl8.nn.modules.activations) SquashedNormal (class in rl8.distributions) state (rl8.algorithms.GenericAlgorithmBase attribute) (rl8.env.DummyEnv attribute) (rl8.trainers.GenericTrainerBase attribute) state_spec (rl8.models.RecurrentModel attribute) (rl8.policies.RecurrentPolicy property) step() (rl8.algorithms.Algorithm method) (rl8.algorithms.GenericAlgorithmBase method) (rl8.algorithms.RecurrentAlgorithm method) (rl8.env.ContinuousDummyEnv method) (rl8.env.DiscreteDummyEnv method) (rl8.env.Env method) (rl8.schedulers.ConstantScheduler method) (rl8.schedulers.EntropyScheduler method) (rl8.schedulers.InterpScheduler method) (rl8.schedulers.LRScheduler method) (rl8.schedulers.Scheduler method) (rl8.schedulers.StepScheduler method) (rl8.trainers.GenericTrainerBase method) StepScheduler (class in rl8.schedulers) StopsDecreasing (class in rl8.conditions) StopsIncreasing (class in rl8.conditions)

T

target_kl_div (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) to() (rl8.models.DefaultContinuousModel method) (rl8.models.DefaultContinuousRecurrentModel method) (rl8.models.DefaultDiscreteModel method) (rl8.models.DefaultDiscreteRecurrentModel method) (rl8.models.GenericModelBase method) (rl8.models.Model method) (rl8.models.RecurrentModel method) (rl8.policies.GenericPolicyBase method)

TorchDistributionWrapper (class in rl8.distributions) TrainConfig (class in rl8.trainers) (class in rl8.trainers.config) Trainer (class in rl8.trainers)

U

upper_bound (rl8.conditions.HitsUpperBound attribute)

V

validate() (rl8.algorithms.Algorithm method) (rl8.algorithms.RecurrentAlgorithm method) validate_view_requirements() (rl8.models.Model method) value (rl8.schedulers.ConstantScheduler attribute) value_function() (rl8.models.DefaultContinuousModel method) (rl8.models.DefaultContinuousRecurrentModel method) (rl8.models.DefaultDiscreteModel method) (rl8.models.DefaultDiscreteRecurrentModel method) (rl8.models.GenericModelBase method) vf_clip_param (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute)

vf_coeff (rl8.trainers.config.TrainConfig attribute) (rl8.trainers.TrainConfig attribute) vf_head (rl8.models.DefaultDiscreteRecurrentModel attribute) vf_model (rl8.models.DefaultContinuousModel attribute) (rl8.models.DefaultContinuousRecurrentModel attribute) (rl8.models.DefaultDiscreteModel attribute) View (class in rl8.views) view_requirements (rl8.models.Model attribute) ViewRequirement (class in rl8.views) VT (rl8.nn.modules.attention.PointerNetwork attribute)

W

W1 (rl8.nn.modules.attention.PointerNetwork attribute)

W2 (rl8.nn.modules.attention.PointerNetwork attribute)

X

x (rl8.schedulers.InterpScheduler attribute)

x_norm (rl8.nn.modules.attention.SelfAttention attribute)

Y

y (rl8.schedulers.InterpScheduler attribute)