From 182551af1dba6a04a8ce737d996641c1d71cae5b Mon Sep 17 00:00:00 2001
From: Tim O'Donnell <timodonnell@gmail.com>
Date: Mon, 15 Jul 2019 17:34:19 -0400
Subject: [PATCH] fix
---
mhcflurry/class1_neural_network.py | 19 +++++++---
.../data_dependent_weights_initialization.py | 2 +-
mhcflurry/train_pan_allele_models_command.py | 36 ++++++++++++++-----
test/expensive_test_pretrain_optimizable.py | 9 ++---
4 files changed, 47 insertions(+), 19 deletions(-)
diff --git a/mhcflurry/class1_neural_network.py b/mhcflurry/class1_neural_network.py
index ec5feb07..40efdbc2 100644
--- a/mhcflurry/class1_neural_network.py
+++ b/mhcflurry/class1_neural_network.py
@@ -569,6 +569,12 @@ class Class1NeuralNetwork(object):
verbose=verbose)
iterator = itertools.chain([first_chunk], iterator)
+ def progress_update(epoch, logs):
+ if verbose:
+ print(
+ "Cumulative training points:",
+ mutable_generator_state['yielded_values'])
+
fit_history = network.fit_generator(
iterator,
steps_per_epoch=steps_per_epoch,
@@ -577,11 +583,14 @@ class Class1NeuralNetwork(object):
workers=1,
validation_data=(validation_x_dict, validation_y_dict),
verbose=verbose,
- callbacks=[keras.callbacks.EarlyStopping(
- monitor="val_loss",
- patience=patience,
- min_delta=min_delta,
- verbose=verbose)]
+ callbacks=[
+ keras.callbacks.EarlyStopping(
+ monitor="val_loss",
+ patience=patience,
+ min_delta=min_delta,
+ verbose=verbose),
+ keras.callbacks.LambdaCallback(on_epoch_end=progress_update),
+ ]
)
for (key, value) in fit_history.history.items():
fit_info[key].extend(value)
diff --git a/mhcflurry/data_dependent_weights_initialization.py b/mhcflurry/data_dependent_weights_initialization.py
index 165c1e6d..c6cfd6c8 100644
--- a/mhcflurry/data_dependent_weights_initialization.py
+++ b/mhcflurry/data_dependent_weights_initialization.py
@@ -35,7 +35,7 @@ import numpy
def svd_orthonormal(shape):
- # Orthonorm init code is from Lasagne
+ # Orthonormal init code is from Lasagne
# https://github.com/Lasagne/Lasagne/blob/master/lasagne/init.py
if len(shape) < 2:
raise RuntimeError("Only shapes of length 2 or more are supported.")
diff --git a/mhcflurry/train_pan_allele_models_command.py b/mhcflurry/train_pan_allele_models_command.py
index 1fcc9c04..ca5a5a75 100644
--- a/mhcflurry/train_pan_allele_models_command.py
+++ b/mhcflurry/train_pan_allele_models_command.py
@@ -475,17 +475,32 @@ def train_model(
print("%s [pid %d]. Hyperparameters:" % (progress_preamble, os.getpid()))
pprint.pprint(hyperparameters)
- if hyperparameters.get("train_data", {}).get("pretrain", False):
- pretrain_patience = hyperparameters["train_data"].get(
- "pretrain_patience", 10)
- pretrain_min_delta = hyperparameters["train_data"].get(
- "pretrain_min_delta", 0.0)
- pretrain_steps_per_epoch = hyperparameters["train_data"].get(
+ train_params = dict(hyperparameters.get("train_data", {}))
+
+ def get_train_param(param, default):
+ if param in train_params:
+ result = train_params.pop(param)
+ if verbose:
+ print("Train param", param, "=", result)
+ else:
+ result = default
+ if verbose:
+ print("Train param", param, "=", result, "[default]")
+ return result
+
+ if get_train_param("pretrain", False):
+ pretrain_patience = get_train_param("pretrain_patience", 10)
+ pretrain_min_delta = get_train_param("pretrain_min_delta", 0.0)
+ pretrain_steps_per_epoch = get_train_param(
"pretrain_steps_per_epoch", 10)
- pretrain_max_epochs = hyperparameters["train_data"].get(
+ pretrain_max_epochs = get_train_param(
"pretrain_max_epochs", 1000)
+ pretrain_peptides_per_step = get_train_param(
+ "pretrain_peptides_per_step", 1024)
+ max_val_loss = get_train_param("pretrain_max_val_loss", None)
- max_val_loss = hyperparameters["train_data"].get("pretrain_max_val_loss")
+ if verbose:
+ print("Unused train params", train_params)
attempt = 0
while True:
@@ -498,7 +513,10 @@ def train_model(
model = Class1NeuralNetwork(**hyperparameters)
assert model.network() is None
generator = pretrain_data_iterator(
- pretrain_data_filename, allele_encoding)
+ pretrain_data_filename,
+ allele_encoding,
+ peptides_per_chunk=pretrain_peptides_per_step)
+
model.fit_generator(
generator,
validation_peptide_encoding=train_peptides,
diff --git a/test/expensive_test_pretrain_optimizable.py b/test/expensive_test_pretrain_optimizable.py
index c2b0e461..027a88bb 100644
--- a/test/expensive_test_pretrain_optimizable.py
+++ b/test/expensive_test_pretrain_optimizable.py
@@ -33,7 +33,7 @@ FOLDS_DF["fold_0"] = True
HYPERPARAMTERS = {
'activation': 'tanh', 'allele_dense_layer_sizes': [],
'batch_normalization': False,
- 'dense_layer_l1_regularization': 9.999999999999999e-11,
+ 'dense_layer_l1_regularization': 0.0,
'dense_layer_l2_regularization': 0.0, 'dropout_probability': 0.5,
'early_stopping': True, 'init': 'glorot_uniform',
'layer_sizes': [1024, 512], 'learning_rate': None,
@@ -50,9 +50,10 @@ HYPERPARAMTERS = {
'random_negative_distribution_smoothing': 0.0,
'random_negative_match_distribution': True, 'random_negative_rate': 0.2,
'train_data': {'pretrain': True,
- 'pretrain_max_epochs': 3,
- 'pretrain_peptides_per_epoch': 1024,
- 'pretrain_steps_per_epoch': 16},
+ 'pretrain_max_epochs': 30,
+ 'pretrain_patience': 5,
+ 'pretrain_peptides_per_step': 32,
+ 'pretrain_steps_per_epoch': 256},
'validation_split': 0.1,
'data_dependent_initialization_method': "lsuv",
}
--
GitLab