mirror of
https://github.com/nod-ai/SHARK-Studio.git
synced 2026-01-09 13:57:54 -05:00
134 lines
4.3 KiB
Python
134 lines
4.3 KiB
Python
from iree import runtime as ireert
|
|
from iree.tf.support import module_utils
|
|
from iree.compiler import tf as tfc
|
|
from iree.compiler import compile_str
|
|
from absl import app
|
|
import time
|
|
|
|
import numpy as np
|
|
import os
|
|
import tensorflow as tf
|
|
|
|
from official.nlp.modeling import layers
|
|
from official.nlp.modeling import networks
|
|
from official.nlp.modeling.models import bert_classifier
|
|
|
|
vocab_size = 100
|
|
NUM_CLASSES = 5
|
|
SEQUENCE_LENGTH = 512
|
|
BATCH_SIZE = 1
|
|
# Create a set of 2-dimensional inputs
|
|
bert_input = [
|
|
tf.TensorSpec(shape=[BATCH_SIZE, SEQUENCE_LENGTH], dtype=tf.int32),
|
|
tf.TensorSpec(shape=[BATCH_SIZE, SEQUENCE_LENGTH], dtype=tf.int32),
|
|
tf.TensorSpec(shape=[BATCH_SIZE, SEQUENCE_LENGTH], dtype=tf.int32),
|
|
]
|
|
|
|
|
|
class BertModule(tf.Module):
|
|
def __init__(self):
|
|
super(BertModule, self).__init__()
|
|
dict_outputs = False
|
|
test_network = networks.BertEncoder(
|
|
vocab_size=vocab_size,
|
|
num_layers=24,
|
|
hidden_size=1024,
|
|
num_attention_heads=16,
|
|
dict_outputs=dict_outputs,
|
|
)
|
|
|
|
# Create a BERT trainer with the created network.
|
|
bert_trainer_model = bert_classifier.BertClassifier(
|
|
test_network, num_classes=NUM_CLASSES
|
|
)
|
|
bert_trainer_model.summary()
|
|
|
|
# Invoke the trainer model on the inputs. This causes the layer to be built.
|
|
self.m = bert_trainer_model
|
|
self.m.predict = lambda x: self.m.call(x, training=False)
|
|
self.predict = tf.function(input_signature=[bert_input])(
|
|
self.m.predict
|
|
)
|
|
self.m.learn = lambda x, y: self.m.call(x, training=False)
|
|
self.loss = tf.keras.losses.SparseCategoricalCrossentropy()
|
|
self.optimizer = tf.keras.optimizers.SGD(learning_rate=1e-2)
|
|
|
|
@tf.function(
|
|
input_signature=[
|
|
bert_input, # inputs
|
|
tf.TensorSpec(shape=[BATCH_SIZE], dtype=tf.int32), # labels
|
|
],
|
|
jit_compile=True,
|
|
)
|
|
def learn(self, inputs, labels):
|
|
with tf.GradientTape() as tape:
|
|
# Capture the gradients from forward prop...
|
|
probs = self.m(inputs, training=True)
|
|
loss = self.loss(labels, probs)
|
|
|
|
# ...and use them to update the model's weights.
|
|
variables = self.m.trainable_variables
|
|
gradients = tape.gradient(loss, variables)
|
|
self.optimizer.apply_gradients(zip(gradients, variables))
|
|
return loss
|
|
|
|
|
|
if __name__ == "__main__":
|
|
# BertModule()
|
|
# Compile the model using IREE
|
|
compiler_module = tfc.compile_module(
|
|
BertModule(), exported_names=["learn"], import_only=True
|
|
)
|
|
|
|
# Compile the model using IREE
|
|
backend = "dylib-llvm-aot"
|
|
args = [
|
|
"--iree-llvmcpu-target-cpu-features=host",
|
|
"--iree-mhlo-demote-i64-to-i32=false",
|
|
"--iree-stream-resource-index-bits=64",
|
|
"--iree-vm-target-index-bits=64",
|
|
]
|
|
backend_config = "dylib"
|
|
# backend = "cuda"
|
|
# backend_config = "cuda"
|
|
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-enable-fusion-with-reduction-ops"]
|
|
flatbuffer_blob = compile_str(
|
|
compiler_module,
|
|
target_backends=[backend],
|
|
extra_args=args,
|
|
input_type="auto",
|
|
)
|
|
# flatbuffer_blob = compile_str(compiler_module, target_backends=["dylib-llvm-aot"])
|
|
|
|
# Save module as MLIR file in a directory
|
|
vm_module = ireert.VmModule.from_flatbuffer(flatbuffer_blob)
|
|
tracer = ireert.Tracer(os.getcwd())
|
|
config = ireert.Config("dylib", tracer)
|
|
ctx = ireert.SystemContext(config=config)
|
|
ctx.add_vm_module(vm_module)
|
|
BertCompiled = ctx.modules.module
|
|
predict_sample_input = [
|
|
np.random.randint(5, size=(BATCH_SIZE, SEQUENCE_LENGTH)),
|
|
np.random.randint(5, size=(BATCH_SIZE, SEQUENCE_LENGTH)),
|
|
np.random.randint(5, size=(BATCH_SIZE, SEQUENCE_LENGTH)),
|
|
]
|
|
learn_sample_input = [
|
|
predict_sample_input,
|
|
np.random.randint(5, size=(BATCH_SIZE)),
|
|
]
|
|
warmup = 5
|
|
total_iter = 10
|
|
num_iter = total_iter - warmup
|
|
for i in range(10):
|
|
if i == warmup - 1:
|
|
start = time.time()
|
|
print(
|
|
BertCompiled.learn(
|
|
predict_sample_input, np.random.randint(5, size=(BATCH_SIZE))
|
|
)
|
|
)
|
|
end = time.time()
|
|
total_time = end - start
|
|
print("time: " + str(total_time))
|
|
print("time/iter: " + str(total_time / num_iter))
|