Optimizer (p.torch.0.10.doc.torch.Torch.Optimizer)

Sourcetype t

Sourcemodule Clip_grad : sig ... end

val adam : 
  ?beta1:float ->
  ?beta2:float ->
  ?weight_decay:float ->
  Var_store.t ->
  learning_rate:float ->
  t

Source

val rmsprop : 
  ?alpha:float ->
  ?eps:float ->
  ?weight_decay:float ->
  ?momentum:float ->
  ?centered:bool ->
  Var_store.t ->
  learning_rate:float ->
  t

Source

val sgd : 
  ?momentum:float ->
  ?dampening:float ->
  ?weight_decay:float ->
  ?nesterov:bool ->
  Var_store.t ->
  learning_rate:float ->
  t

Sourceval step : ?clip_grad:Clip_grad.t -> t -> unit

Sourceval zero_grad : t -> unit

Sourceval backward_step : ?clip_grad:Clip_grad.t -> t -> loss:Tensor.t -> unit

Sourceval set_learning_rate : t -> learning_rate:float -> unit

Sourcemodule Linear_interpolation : sig ... end