Is your feature request related to a problem? Please describe. Wi

There is a test case that trains the model: <div class="Box Box-

finetune BERT with custom dataset about keras-bert HOT 2 CLOSED

cyberzhg commented on August 21, 2024

finetune BERT with custom dataset

from keras-bert.

Comments (2)

CyberZHG commented on August 21, 2024

There is a test case that trains the model:

keras-bert/tests/test_bert.py

Lines 26 to 93 in 02c7eb2

 def test_fit(self): 

 current_path = os.path.dirname(os.path.abspath(__file__)) 

 model_path = os.path.join(current_path, 'test_bert_fit.h5') 

 sentence_pairs = [ 

 [['all', 'work', 'and', 'no', 'play'], ['makes', 'jack', 'a', 'dull', 'boy']], 

 [['from', 'the', 'day', 'forth'], ['my', 'arm', 'changed']], 

 [['and', 'a', 'voice', 'echoed'], ['power', 'give', 'me', 'more', 'power']], 

 ] 

 token_dict = get_base_dict() 

 for pairs in sentence_pairs: 

 for token in pairs[0] + pairs[1]: 

 if token not in token_dict: 

 token_dict[token] = len(token_dict) 

 token_list = list(token_dict.keys()) 

 if os.path.exists(model_path): 

 model = keras.models.load_model( 

 model_path, 

 custom_objects=get_custom_objects(), 

 ) 

 else: 

 model = get_model( 

 token_num=len(token_dict), 

 head_num=5, 

 transformer_num=12, 

 embed_dim=25, 

 feed_forward_dim=100, 

 seq_len=20, 

 pos_num=20, 

 dropout_rate=0.05, 

 attention_activation=gelu, 

 lr=1e-3, 

 ) 

 model.summary() 

 def _generator(): 

 while True: 

 yield gen_batch_inputs( 

 sentence_pairs, 

 token_dict, 

 token_list, 

 seq_len=20, 

 mask_rate=0.3, 

 swap_sentence_rate=1.0, 

 ) 

 model.fit_generator( 

 generator=_generator(), 

 steps_per_epoch=1000, 

 epochs=1, 

 validation_data=_generator(), 

 validation_steps=100, 

 callbacks=[ 

 keras.callbacks.ReduceLROnPlateau(monitor='val_MLM_loss', factor=0.5, patience=3), 

 keras.callbacks.EarlyStopping(monitor='val_MLM_loss', patience=5) 

 ], 

 ) 

 # model.save(model_path) 

 for inputs, outputs in _generator(): 

 predicts = model.predict(inputs) 

 outputs = list(map(lambda x: np.squeeze(x, axis=-1), outputs)) 

 predicts = list(map(lambda x: np.argmax(x, axis=-1), predicts)) 

 batch_size, seq_len = inputs[-1].shape 

 for i in range(batch_size): 

 for j in range(seq_len): 

 if inputs[-1][i][j]: 

 self.assertEqual(outputs[0][i][j], predicts[0][i][j]) 

 self.assertTrue(np.allclose(outputs[1], predicts[1])) 

 break

However, I recommend training with the official implementation then load the checkpoint (since the optimizer and the creation of sentence pairs are different).

from keras-bert.

stale commented on August 21, 2024

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

from keras-bert.

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.

Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

TensorFlow

An Open Source Machine Learning Framework for Everyone

Django

The Web framework for perfectionists with deadlines.

Laravel

A PHP framework for web artisans

D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

web

Some thing interesting about web. New door for the world.

server

A server is a program made to process requests and deliver data to clients.

Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

Visualization

Some thing interesting about visualization, use data art

Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.

Microsoft

Open source projects and samples from Microsoft.

Google

Google ❤️ Open Source for everyone.

Alibaba

Alibaba Open Source for everyone

D3

Data-Driven Documents codes.

Tencent

China tencent open source team.

	def test_fit(self):
	current_path = os.path.dirname(os.path.abspath(__file__))
	model_path = os.path.join(current_path, 'test_bert_fit.h5')
	sentence_pairs = [
	[['all', 'work', 'and', 'no', 'play'], ['makes', 'jack', 'a', 'dull', 'boy']],
	[['from', 'the', 'day', 'forth'], ['my', 'arm', 'changed']],
	[['and', 'a', 'voice', 'echoed'], ['power', 'give', 'me', 'more', 'power']],
	]
	token_dict = get_base_dict()
	for pairs in sentence_pairs:
	for token in pairs[0] + pairs[1]:
	if token not in token_dict:
	token_dict[token] = len(token_dict)
	token_list = list(token_dict.keys())
	if os.path.exists(model_path):
	model = keras.models.load_model(
	model_path,
	custom_objects=get_custom_objects(),
	)
	else:
	model = get_model(
	token_num=len(token_dict),
	head_num=5,
	transformer_num=12,
	embed_dim=25,
	feed_forward_dim=100,
	seq_len=20,
	pos_num=20,
	dropout_rate=0.05,
	attention_activation=gelu,
	lr=1e-3,
	)
	model.summary()

	def _generator():
	while True:
	yield gen_batch_inputs(
	sentence_pairs,
	token_dict,
	token_list,
	seq_len=20,
	mask_rate=0.3,
	swap_sentence_rate=1.0,
	)

	model.fit_generator(
	generator=_generator(),
	steps_per_epoch=1000,
	epochs=1,
	validation_data=_generator(),
	validation_steps=100,
	callbacks=[
	keras.callbacks.ReduceLROnPlateau(monitor='val_MLM_loss', factor=0.5, patience=3),
	keras.callbacks.EarlyStopping(monitor='val_MLM_loss', patience=5)
	],
	)
	# model.save(model_path)
	for inputs, outputs in _generator():
	predicts = model.predict(inputs)
	outputs = list(map(lambda x: np.squeeze(x, axis=-1), outputs))
	predicts = list(map(lambda x: np.argmax(x, axis=-1), predicts))
	batch_size, seq_len = inputs[-1].shape
	for i in range(batch_size):
	for j in range(seq_len):
	if inputs[-1][i][j]:
	self.assertEqual(outputs[0][i][j], predicts[0][i][j])
	self.assertTrue(np.allclose(outputs[1], predicts[1]))
	break

Comments (2)

Related Issues (20)

Recommend Projects

Recommend Topics

Recommend Org