The new learning module in tensorflow.contrib.slim looks very promising :
https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/slim/python/slim/learning.py
I'm trying to figure out how can I reproduce the CIFAR 10 multi-gpu example (or the ImageNet example) using this new module on a configuration where I have only a single worker node but with several GPU on it.
I've had some success using https://github.com/tensorflow/models/tree/master/slim/deployment
When creating the config object you would set num_clones = [num_gpus]
For example,
config = model_deploy.DeploymentConfig(num_clones=2)
Check out the example in model_deploy.py file.
Related
I have customized NER pipeline with following procedure
doc = nlp("I am going to Vallila. I am going to Sörnäinen.")
for ent in doc.ents:
print(ent.text, ent.label_)
LABEL = 'DISTRICT'
TRAIN_DATA = [
(
'We need to deliver it to Vallila', {
'entities': [(25, 32, 'DISTRICT')]
}),
(
'We need to deliver it to somewhere', {
'entities': []
}),
]
ner = nlp.get_pipe("ner")
ner.add_label(LABEL)
nlp.disable_pipes("tagger")
nlp.disable_pipes("parser")
nlp.disable_pipes("attribute_ruler")
nlp.disable_pipes("lemmatizer")
nlp.disable_pipes("tok2vec")
optimizer = nlp.get_pipe("ner").create_optimizer()
import random
from spacy.training import Example
for i in range(25):
random.shuffle(TRAIN_DATA)
for text, annotation in TRAIN_DATA:
example = Example.from_dict(nlp.make_doc(text), annotation)
nlp.update([example], sgd=optimizer)
I tried to save that customized NER to disk and load it again with following code
ner.to_disk('/home/feru/ner')
import spacy
from spacy.pipeline import EntityRecognizer
nlp = spacy.load("en_core_web_lg", disable=['ner'])
ner = EntityRecognizer(nlp.vocab)
ner.from_disk('/home/feru/ner')
nlp.add_pipe(ner)
I got however following error:
---> 10 ner = EntityRecognizer(nlp.vocab)
11 ner.from_disk('/home/feru/ner')
12 nlp.add_pipe(ner)
~/.local/lib/python3.8/site-packages/spacy/pipeline/ner.pyx in
spacy.pipeline.ner.EntityRecognizer.init()
TypeError: init() takes at least 2 positional arguments (1 given)
This method to save and load custom component from disk seems to be from some erly SpaCy version. What's the second argument EntityRecognizer needs?
The general process you are following of serializing a single component and reloading it is not the recommended way to do this in spaCy. You can do it - it has to be done internally, of course - but you generally want to save and load pipelines using high-level wrappers. In this case this means that you would save like this:
nlp.to_disk("my_model") # NOT ner.to_disk
And then load it with spacy.load("my_model").
You can find more detail about this in the saving and loading docs. Since it seems you're just getting started with spaCy, you might want to go through the course too. It covers the new config-based training in v3, which is much easier than using your own custom training loop like in your code sample.
If you want to mix and match components from different pipelines, you still will generally want to save entire pipelines, and you can then combine components from them using the "sourcing" feature.
I need to run a custom GluonCV object detection module on Android.
I already fine-tuned the model (ssd_512_mobilenet1.0_custom) on a custom dataset, I tried running inference with it (loading the .params file produced during the training) and everything works perfectly on my computer. Now, I need to export this to Android.
I was referring to this answer to figure out the procedure, there are 3 suggested options:
You can use ONNX to convert models to other runtimes, for example [...] NNAPI for Android
You can use TVM
You can use SageMaker Neo + DLR runtime [...]
Regarding the first one, I converted my model to ONNX.
However, in order to use it with NNAPI, it is necessary to convert it to daq. In the repository, they provide a precomplied AppImage of onnx2daq to make the conversion, but the script returns an error. I checked the issues section, and they report that "It actually fails for all onnx object detection models".
Then, I gave a try to DLR, since it's suggested to be the easiest way.
As I understand, in order to use my custom model with DLR, I would first need to compile it with TVM (which also covers the second point mentioned in the linked post). In the repo, they provide a Docker image with some conversion scripts for different frameworks.
I modified the 'compile_gluoncv.py' script, and now I have:
#!/usr/bin/env python3
from tvm import relay
import mxnet as mx
from mxnet.gluon.model_zoo.vision import get_model
from tvm_compiler_utils import tvm_compile
shape_dict = {'data': (1, 3, 300, 300)}
dtype='float32'
ctx = [mx.cpu(0)]
classes_custom = ["CML_mug"]
block = get_model('ssd_512_mobilenet1.0_custom', classes=classes_custom, pretrained_base=False, ctx=ctx)
block.load_parameters("ep_035.params", ctx=ctx) ### this is the file produced by training on the custom dataset
for arch in ["arm64-v8a", "armeabi-v7a", "x86_64", "x86"]:
sym, params = relay.frontend.from_mxnet(block, shape=shape_dict, dtype=dtype)
func = sym["main"]
func = relay.Function(func.params, relay.nn.softmax(func.body), None, func.type_params, func.attrs)
tvm_compile(func, params, arch, dlr_model_name)
However, when I run the script it returns the error:
ValueError: Model ssd_512_mobilenet1.0_custom is not supported. Available options are
alexnet
densenet121
densenet161
densenet169
densenet201
inceptionv3
mobilenet0.25
mobilenet0.5
mobilenet0.75
mobilenet1.0
mobilenetv2_0.25
mobilenetv2_0.5
mobilenetv2_0.75
mobilenetv2_1.0
resnet101_v1
resnet101_v2
resnet152_v1
resnet152_v2
resnet18_v1
resnet18_v2
resnet34_v1
resnet34_v2
resnet50_v1
resnet50_v2
squeezenet1.0
squeezenet1.1
vgg11
vgg11_bn
vgg13
vgg13_bn
vgg16
vgg16_bn
vgg19
vgg19_bn
Am I doing something wrong? Is this thing even possible?
As a side note, after this I'd need to deploy on Android a pose detection model (simple_pose_resnet18_v1b) and an activity recognition one (i3d_nl10_resnet101_v1_kinetics400) as well.
You actually can run GluonCV model directly on Android with Deep Java Library (DJL)
What you need to do is:
hyridize your GluonCV model and save as MXNet model
Build MXNet engine for android, MXNET already support Android build
Include MXNet shared library into your android project
Use DJL in your android project, you can follow this DJL Android demo for PyTorch
The error message is self-explanatory - there is no model "ssd_512_mobilenet1.0_custom" supported by mxnet.gluon.model_zoo.vision.get_model. You are confusing GluonCV's get_model with MXNet Gluon's get_model.
Replace
block = get_model('ssd_512_mobilenet1.0_custom',
classes=classes_custom, pretrained_base=False, ctx=ctx)
with
import gluoncv
block = gluoncv.model_zoo.get_model('ssd_512_mobilenet1.0_custom',
classes=classes_custom, pretrained_base=False, ctx=ctx)
I have an export.pkl model which has been trained on a cuda machine. I want to use it on a macbook:
from fastai.text import load_learner
from utils import get_corpus
learner = load_learner('./models')
corpus = get_corpus()
res = [ str(learner.predict(c)[0]) for c in corpus ]
I get the following error:
...
File "/Users/gautiergilabert/Envs/cc/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 146, in forward
"them on device: {}".format(self.src_device_obj, t.device))
RuntimeError: module must have its parameters and buffers on device cuda:0 (device_ids[0]) but found one of them on device: cpu
I have two questions:
I found the raise in my export.pkl:
for t in chain(self.module.parameters(), self.module.buffers()):
if t.device != self.src_device_obj:
raise RuntimeError("module must have its parameters and buffers "
"on device {} (device_ids[0]) but found one of "
"them on device: {}".format(self.src_device_obj, t.device))
It is said about the module in the docstring: module to be parallelized. I don't really understand what it is. My macbook ?
Apart my macbook, I would like to run the model on a cpu
Is there a way to make this export.pkl model works on a cpu ?
Is there a way to make another export.pkl on cuda and make it available on a cpu ?
Thanks
One way is to load the learner by specifying the model with empty dataset and loading the model weights afterwards. For resnet image classifier something like this should work:
from fastai.vision import *
# path where the model is saved under path/models/model-name
path = "model_path"
tfms = get_transforms()
data = ImageDataBunch.single_from_classes(".", classes=["class1", "class2"], ds_tfms=tfms)
learner = cnn_learner(data, models.resnet34, metrics=accuracy)
# loads model from model_path/models/model_name.pth
learner.load("model_name")
image = open_image("test.jpg")
pred_class, pred_idx, outputs = learner.predict(image)
I am trying to work with the quite recently published tensorflow_dataset API to train a Keras model on the Open Images Dataset. The dataset is about 570 GB in size. I downloaded the data with the following code:
import tensorflow_datasets as tfds
import tensorflow as tf
open_images_dataset = tfds.image.OpenImagesV4()
open_images_dataset.download_and_prepare(download_dir="/notebooks/dataset/")
After the download was complete, the connection to my jupyter notebook somehow interrupted but the extraction seemed to be finished as well, at least all downloaded files had a counterpart in the "extracted" folder. However, I am not able to access the downloaded data now:
tfds.load(name="open_images_v4", data_dir="/notebooks/open_images_dataset/extracted/", download=False)
This only gives the following error:
AssertionError: Dataset open_images_v4: could not find data in /notebooks/open_images_dataset/extracted/. Please make sure to call dataset_builder.download_and_prepare(), or pass download=True to tfds.load() before trying to access the tf.data.Dataset object.
When I call the function download_and_prepare() it only downloads the whole dataset again.
Am I missing something here?
Edit:
After the download the folder under "extracted" has 18 .tar.gz files.
This is with tensorflow-datasets 1.0.1 and tensorflow 2.0.
The folder hierarchy should be like this:
/notebooks/open_images_dataset/extracted/open_images_v4/0.1.0
All the datasets have a version. Then the data could be loaded like this.
ds = tf.load('open_images_v4', data_dir='/notebooks/open_images_dataset/extracted', download=False)
I didn't have open_images_v4 data. I put cifar10 data into a folder named open_images_v4 to check what folder structure tensorflow_datasets was expecting.
The solution to this was to also use the "data_dir" parameter when initializing the dataset:
builder = tfds.image.OpenImagesV4(data_dir="/raid/openimages/dataset")
builder.download_and_prepare(download_dir="/raid/openimages/dataset")
This way the dataset is donwloaded and extracted in the same directory. Before, it was (for me unnoticeably) extracting to the default directory, which is under /home/.../. That's what caused the error, as there wasn't enough space left under my home directory.
After the extraction, the folder structure is exactly as Manoj-Mohan described.
Above solution haven't worked for me.
builder = tfds.builder(name='folder_name', data_dir=data_dir)
builder.download_and_prepare(download_dir="/home/...")
ds = builder.as_dataset()
After following this tutorial on summaries and TensorBoard, I've been able to successfully save and look at data with TensorBoard. Is it possible to open this data with something other than TensorBoard?
By the way, my application is to do off-policy learning. I'm currently saving each state-action-reward tuple using SummaryWriter. I know I could manually store/train on this data, but I thought it'd be nice to use TensorFlow's built in logging features to store/load this data.
As of March 2017, the EventAccumulator tool has been moved from Tensorflow core to the Tensorboard Backend. You can still use it to extract data from Tensorboard log files as follows:
from tensorboard.backend.event_processing.event_accumulator import EventAccumulator
event_acc = EventAccumulator('/path/to/summary/folder')
event_acc.Reload()
# Show all tags in the log file
print(event_acc.Tags())
# E. g. get wall clock, number of steps and value for a scalar 'Accuracy'
w_times, step_nums, vals = zip(*event_acc.Scalars('Accuracy'))
Easy, the data can actually be exported to a .csv file within TensorBoard under the Events tab, which can e.g. be loaded in a Pandas dataframe in Python. Make sure you check the Data download links box.
For a more automated approach, check out the TensorBoard readme:
If you'd like to export data to visualize elsewhere (e.g. iPython
Notebook), that's possible too. You can directly depend on the
underlying classes that TensorBoard uses for loading data:
python/summary/event_accumulator.py (for loading data from a single
run) or python/summary/event_multiplexer.py (for loading data from
multiple runs, and keeping it organized). These classes load groups of
event files, discard data that was "orphaned" by TensorFlow crashes,
and organize the data by tag.
As another option, there is a script
(tensorboard/scripts/serialize_tensorboard.py) which will load a
logdir just like TensorBoard does, but write all of the data out to
disk as json instead of starting a server. This script is setup to
make "fake TensorBoard backends" for testing, so it is a bit rough
around the edges.
I think the data are encoded protobufs RecordReader format. To get serialized strings out of files you can use py_record_reader or build a graph with TFRecordReader op, and to deserialize those strings to protobuf use Event schema. If you get a working example, please update this q, since we seem to be missing documentation on this.
I did something along these lines for a previous project. As mentioned by others, the main ingredient is tensorflows event accumulator
from tensorflow.python.summary import event_accumulator as ea
acc = ea.EventAccumulator("folder/containing/summaries/")
acc.Reload()
# Print tags of contained entities, use these names to retrieve entities as below
print(acc.Tags())
# E. g. get all values and steps of a scalar called 'l2_loss'
xy_l2_loss = [(s.step, s.value) for s in acc.Scalars('l2_loss')]
# Retrieve images, e. g. first labeled as 'generator'
img = acc.Images('generator/image/0')
with open('img_{}.png'.format(img.step), 'wb') as f:
f.write(img.encoded_image_string)
You can also use the tf.train.summaryiterator: To extract events in a ./logs-Folder where only classic scalars lr, acc, loss, val_acc and val_loss are present you can use this GIST: tensorboard_to_csv.py
Chris Cundy's answer works well when you have less than 10000 data points in your tfevent file. However, when you have a large file with over 10000 data points, Tensorboard will automatically sampling them and only gives you at most 10000 points. It is a quite annoying underlying behavior as it is not well-documented. See https://github.com/tensorflow/tensorboard/blob/master/tensorboard/backend/event_processing/event_accumulator.py#L186.
To get around it and get all data points, a bit hacky way is to:
from tensorboard.backend.event_processing.event_accumulator import EventAccumulator
class FalseDict(object):
def __getitem__(self,key):
return 0
def __contains__(self, key):
return True
event_acc = EventAccumulator('path/to/your/tfevents',size_guidance=FalseDict())
It looks like for tb version >=2.3 you can streamline the process of converting your tb events to a pandas dataframe using tensorboard.data.experimental.ExperimentFromDev().
It requires you to upload your logs to TensorBoard.dev, though, which is public. There are plans to expand the capability to locally stored logs in the future.
https://www.tensorflow.org/tensorboard/dataframe_api
You can also use the EventFileLoader to iterate through a tensorboard file
from tensorboard.backend.event_processing.event_file_loader import EventFileLoader
for event in EventFileLoader('path/to/events.out.tfevents.xxx').Load():
print(event)
Surprisingly, the python package tb_parse has not been mentioned yet.
From documentation:
Installation:
pip install tensorflow # or tensorflow-cpu pip install -U tbparse # requires Python >= 3.7
Note: If you don't want to install TensorFlow, see Installing without TensorFlow.
We suggest using an additional virtual environment for parsing and plotting the tensorboard events. So no worries if your training code uses Python 3.6 or older versions.
Reading one or more event files with tbparse only requires 5 lines of code:
from tbparse import SummaryReader
log_dir = "<PATH_TO_EVENT_FILE_OR_DIRECTORY>"
reader = SummaryReader(log_dir)
df = reader.scalars
print(df)