tests/testthat/_snaps/luz_callbacks.md

In macmillancontentscience/torchtransformers: Transformer Models in Torch

luz callback works.

Code
  fitted
Output
  A `luz_module_fitted`
  -- Time ------------------------------------------------------------------------
  * Total time: 
Warning <simpleWarning>
  argument is not numeric or logical: returning NA
Output
  * Avg time per training epoch: NA

  -- Results ---------------------------------------------------------------------
  Metrics observed in the last epoch.

  i Training:
  loss: 1.1151
  acc: 0

  -- Model -----------------------------------------------------------------------
  An `nn_module` containing 4,369,666 parameters.

  -- Modules ---------------------------------------------------------------------
  * bert: <BERT_pretrained> #4,369,408 parameters
  * linear: <nn_linear> #258 parameters

Code
  luz_callback_bert_tokenize()
Output
  <bert_tokenize_callback>
    Inherits from: <LuzCallback>
    Public:
      call: function (callback_nm) 
      clone: function (deep = FALSE) 
      initialize: function (submodel_name = NULL, n_tokens = NULL, verbose = TRUE) 
      n_tokens: NULL
      on_fit_begin: function () 
      on_predict_begin: function () 
      set_ctx: function (ctx) 
      submodel_name: NULL
      verbose: TRUE

Code
  luz_callback_bert_tokenize(submodel_name = "bert")
Output
  <bert_tokenize_callback>
    Inherits from: <LuzCallback>
    Public:
      call: function (callback_nm) 
      clone: function (deep = FALSE) 
      initialize: function (submodel_name = NULL, n_tokens = NULL, verbose = TRUE) 
      n_tokens: NULL
      on_fit_begin: function () 
      on_predict_begin: function () 
      set_ctx: function (ctx) 
      submodel_name: bert
      verbose: TRUE

Code
  luz_callback_bert_tokenize(n_tokens = 32L)
Output
  <bert_tokenize_callback>
    Inherits from: <LuzCallback>
    Public:
      call: function (callback_nm) 
      clone: function (deep = FALSE) 
      initialize: function (submodel_name = NULL, n_tokens = NULL, verbose = TRUE) 
      n_tokens: 32
      on_fit_begin: function () 
      on_predict_begin: function () 
      set_ctx: function (ctx) 
      submodel_name: NULL
      verbose: TRUE

Code
  luz_callback_bert_tokenize(verbose = FALSE)
Output
  <bert_tokenize_callback>
    Inherits from: <LuzCallback>
    Public:
      call: function (callback_nm) 
      clone: function (deep = FALSE) 
      initialize: function (submodel_name = NULL, n_tokens = NULL, verbose = TRUE) 
      n_tokens: NULL
      on_fit_begin: function () 
      on_predict_begin: function () 
      set_ctx: function (ctx) 
      submodel_name: NULL
      verbose: FALSE

macmillancontentscience/torchtransformers documentation built on Aug. 6, 2023, 5:35 a.m.