|
|
@@ -1,6 +1,7 @@ |
|
|
|
using System; |
|
|
|
using System.Collections; |
|
|
|
using System.Collections.Generic; |
|
|
|
using System.Diagnostics; |
|
|
|
using System.IO; |
|
|
|
using System.Linq; |
|
|
|
using System.Text; |
|
|
@@ -52,7 +53,7 @@ namespace TensorFlowNET.Examples.CnnTextClassification |
|
|
|
protected virtual bool RunWithImportedGraph(Session sess, Graph graph) |
|
|
|
{ |
|
|
|
Console.WriteLine("Building dataset..."); |
|
|
|
var (x, y, alphabet_size) = DataHelpers.build_char_dataset("train", model_name, CHAR_MAX_LEN, DataLimit); |
|
|
|
var (x, y, alphabet_size) = DataHelpers.build_char_dataset("train", model_name, CHAR_MAX_LEN, DataLimit=null); |
|
|
|
Console.WriteLine("\tDONE"); |
|
|
|
|
|
|
|
var (train_x, valid_x, train_y, valid_y) = train_test_split(x, y, test_size: 0.15f); |
|
|
@@ -76,12 +77,13 @@ namespace TensorFlowNET.Examples.CnnTextClassification |
|
|
|
Tensor optimizer = graph.get_operation_by_name("loss/optimizer"); |
|
|
|
Tensor global_step = graph.get_operation_by_name("global_step"); |
|
|
|
Tensor accuracy = graph.get_operation_by_name("accuracy/accuracy"); |
|
|
|
|
|
|
|
var stopwatch = Stopwatch.StartNew(); |
|
|
|
int i = 0; |
|
|
|
foreach (var (x_batch, y_batch) in train_batches) |
|
|
|
foreach (var (x_batch, y_batch, total) in train_batches) |
|
|
|
{ |
|
|
|
i++; |
|
|
|
Console.WriteLine("Training on batch " + i); |
|
|
|
var estimate = TimeSpan.FromSeconds((stopwatch.Elapsed.TotalSeconds / i) * total); |
|
|
|
Console.WriteLine($"Training on batch {i}/{total}. Estimated training time: {estimate}"); |
|
|
|
var train_feed_dict = new Hashtable |
|
|
|
{ |
|
|
|
[model_x] = x_batch, |
|
|
@@ -90,8 +92,7 @@ namespace TensorFlowNET.Examples.CnnTextClassification |
|
|
|
};
|
|
|
|
// original python:
|
|
|
|
//_, step, loss = sess.run([model.optimizer, model.global_step, model.loss], feed_dict = train_feed_dict) |
|
|
|
var result = sess.run(new Tensor[] { optimizer, global_step, loss }, train_feed_dict);
|
|
|
|
// exception here, loss value seems like a float[]
|
|
|
|
var result = sess.run(new ITensorOrOperation[] { optimizer, global_step, loss }, train_feed_dict);
|
|
|
|
//loss_value = result[2]; |
|
|
|
var step = result[1];
|
|
|
|
if (step % 10 == 0) |
|
|
@@ -102,7 +103,7 @@ namespace TensorFlowNET.Examples.CnnTextClassification |
|
|
|
// # Test accuracy with validation data for each epoch.
|
|
|
|
var valid_batches = batch_iter(valid_x, valid_y, BATCH_SIZE, 1); |
|
|
|
var (sum_accuracy, cnt) = (0, 0); |
|
|
|
foreach (var (valid_x_batch, valid_y_batch) in valid_batches) |
|
|
|
foreach (var (valid_x_batch, valid_y_batch, total_validation_batches) in valid_batches) |
|
|
|
{ |
|
|
|
// valid_feed_dict = { |
|
|
|
// model.x: valid_x_batch, |
|
|
@@ -170,16 +171,19 @@ namespace TensorFlowNET.Examples.CnnTextClassification |
|
|
|
return (train_x, valid_x, train_y, valid_y); |
|
|
|
} |
|
|
|
|
|
|
|
private IEnumerable<(NDArray, NDArray)> batch_iter(NDArray inputs, NDArray outputs, int batch_size, int num_epochs) |
|
|
|
private IEnumerable<(NDArray, NDArray, int)> batch_iter(NDArray inputs, NDArray outputs, int batch_size, int num_epochs) |
|
|
|
{ |
|
|
|
var num_batches_per_epoch = (len(inputs) - 1); // batch_size + 1 |
|
|
|
var num_batches_per_epoch = (len(inputs) - 1) / batch_size; |
|
|
|
var total_batches = num_batches_per_epoch * num_epochs; |
|
|
|
foreach (var epoch in range(num_epochs)) |
|
|
|
{ |
|
|
|
foreach (var batch_num in range(num_batches_per_epoch)) |
|
|
|
{ |
|
|
|
var start_index = batch_num * batch_size; |
|
|
|
var end_index = Math.Min((batch_num + 1) * batch_size, len(inputs)); |
|
|
|
yield return (inputs[new Slice(start_index, end_index)], outputs[new Slice(start_index,end_index)]); |
|
|
|
if (end_index <= start_index) |
|
|
|
break; |
|
|
|
yield return (inputs[new Slice(start_index, end_index)], outputs[new Slice(start_index,end_index)], total_batches); |
|
|
|
} |
|
|
|
} |
|
|
|
} |
|
|
|