Blue Marlin is a critical web infrastructure for advertising based monetization. It is a cloud platform that adds intelligence to a plain Ad System.
Home Page: https://incubator.apache.org/
License: Apache License 2.0
For previous announcements, please go to our twitter: https://twitter.com/fw_marlin
Blue Marlin events, please go to https://github.com/apache/incubator-bluemarlin/wiki/Editing-Blue-Marlin-Events
Blue Marlin meeting notes, please go to https://github.com/apache/incubator-bluemarlin/wiki/Blue-Marlin-Meeting-Notes
This is a test.
Training fails at the beginning itself if line 40 in model.py is not commented. If we comment line 40 in model.py then train.py runs successfully.
In below code taken from model.py, if we comment -> user_emb_w which is taken from line 40 of model.py, then training is successful.
hidden_units = 128
user_emb_w = tf.get_variable("user_emb_w", [user_count, hidden_units])
item_emb_w = tf.get_variable("item_emb_w", [item_count, hidden_units // 2])
item_b = tf.get_variable("item_b", [item_count],
initializer=tf.constant_initializer(0.0))
cate_emb_w = tf.get_variable("cate_emb_w", [cate_count, hidden_units // 2])
cate_list = tf.convert_to_tensor(cate_list, dtype=tf.int64)
`
Below is the error displayed.
2022-02-21 17:42:43.558189: W tensorflow/core/framework/op_kernel.cc:1651] OP_REQUIRES failed at random_op.cc:76 : Resource exhausted: OOM when allocating tensor with shape[94315979,128] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
Traceback (most recent call last):
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1365, in _do_call
return fn(*args)
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1350, in _run_fn
target_list, run_metadata)
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1443, in _call_tf_sessionrun
run_metadata)
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[94315979,128] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
[[{{node user_emb_w/Initializer/random_uniform/RandomUniform}}]]
Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "lookalike_model/trainer/train.py", line 179, in
sess.run(tf.global_variables_initializer())
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 956, in run
run_metadata_ptr)
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1180, in _run
feed_dict_tensor, options, run_metadata)
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1359, in _do_run
run_metadata)
File "/usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/client/session.py", line 1384, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[94315979,128] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
[[node user_emb_w/Initializer/random_uniform/RandomUniform (defined at usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py:1748) ]]
Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.
Original stack trace for 'user_emb_w/Initializer/random_uniform/RandomUniform':
File "/algorithm/lookalike_model/trainer/train.py", line 178, in
model = Model(user_count, item_count, cate_count, cate_list, predict_batch_size, predict_ads_num)
File "/algorithm/lookalike_model/trainer/model.py", line 40, in init
user_emb_w = tf.get_variable("user_emb_w", [user_count, hidden_units])
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 1500, in get_variable
aggregation=aggregation)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 1243, in get_variable
aggregation=aggregation)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 567, in get_variable
aggregation=aggregation)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 519, in _true_getter
aggregation=aggregation)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 933, in _get_single_variable
aggregation=aggregation)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 258, in call
return cls._variable_v1_call(*args, **kwargs)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 219, in _variable_v1_call
shape=shape)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 197, in
previous_getter = lambda **kwargs: default_variable_creator(None, **kwargs)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 2519, in default_variable_creator
shape=shape)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 262, in call
return super(VariableMetaclass, cls).call(*args, **kwargs)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 1688, in init
shape=shape)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variables.py", line 1818, in _init_from_args
initial_value(), name="initial_value", dtype=dtype)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/variable_scope.py", line 905, in
partition_info=partition_info)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/init_ops.py", line 533, in call
shape, -limit, limit, dtype, seed=self.seed)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/random_ops.py", line 245, in random_uniform
rnd = gen_random_ops.random_uniform(shape, dtype, seed=seed1, seed2=seed2)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/ops/gen_random_ops.py", line 822, in random_uniform
name=name)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/framework/op_def_library.py", line 794, in _apply_op_helper
op_def=op_def)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/util/deprecation.py", line 507, in new_func
return func(*args, **kwargs)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py", line 3357, in create_op
attrs, op_def, compute_device)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py", line 3426, in _create_op_internal
op_def=op_def)
File "usr/local/python3/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py", line 1748, in init
self._traceback = tf_stack.extract_stack()
For the latest code that you have released for lookalike, we are facing a problem in the last step of data pipeline i.e. tf record generation.
script: https://github.com/apache/incubator-bluemarlin/blob/main/Model/lookalike-model/lookalike_model/pipeline/main_tfrecord_generator.py
We are currently running the pipeline for 120 Million AIDs and we get the following error on the highlighted line.
Kindly help us in running the pipeline for big data.
DLPredictor predicts traffic based on profile and geolocation attributes. We like to expand the system to accept interests and behavioral attributes as well (TBR project). We like to know how real-world queries might be.
Hello Jimmy,
We would like to understand how you did sampling in the lookalike_build_dataset.py script.
It would be great if you can share an informal description of this sampling method, so that we can reproduce it at our end.
Thanks.
we'll have an open discussion for the potential of request based prediction for dlpreditor
time: 9:30am EST 2/8/2022
location: zoom meeting
zoom id: 693 070 5942
The current Factdata is based on the Impression table, not the Request table. To have more accurate results, we need to use the Request table which contains fewer human interrupted data.
Hello,
As discussed in previous meetings, the total number of records in Trainready table is same as the total number of distinct users(aids). We have confirmed it on our side.
It would be helpful if you once check your data and verify if you have same number of records as the number of distinct users or not in the final Trainready table.
Thanks and Regards
Describe the bug
A clear and concise description of what the bug is.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
A clear and concise description of what you expected to happen.
Screenshots
If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
Smartphone (please complete the following information):
Additional context
Add any other context about the problem here.
Testing
This is a illustration of the seq2seq model with attention for DLpredictor.
This is the relationship of functions defined in model.py
process:
results:
| user_id | kw1 | kw2 | kw3 | kw4 | kw5 | kw6 | kw7 | kw8 | kw9 | kw10 | kw11 | kw12 | kw13 | kw14 | kw15 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 0.000 | 0.000 | 0.000 | 0.130 | 0.000 | 0.399 | 0.000 | 0.000 | 0.000 | 0.612 | 0.000 | 0.000 | 0.301 | 0.458 | 0.000 |
| 5 | 0.000 | 0.000 | 0.078 | 0.000 | 0.000 | 0.416 | 0.000 | 0.000 | 0.366 | 0.436 | 0.384 | 0.000 | 0.189 | 0.000 | 0.541 |
| 8 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.563 | 0.000 | 0.000 | 0.649 | 0.678 | 0.000 | 0.000 | 0.000 | 0.600 | 0.000 |
| 10 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.279 | 0.000 | 0.125 | 0.000 | 0.223 | 0.000 |
| 11 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.354 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.162 | 0.275 | 0.000 | 0.000 |
| 15 | 0.000 | 0.000 | 0.099 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.509 | 0.000 | 0.000 | 0.249 | 0.000 | 0.000 |
| 22 | 0.000 | 0.000 | 0.152 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.515 | 0.000 | 0.000 | 0.000 | 0.423 | 0.000 |
| 30 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.474 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 34 | 0.000 | 0.000 | 0.000 | 0.000 | 0.299 | 0.000 | 0.000 | 0.000 | 0.000 | 0.410 | 0.000 | 0.149 | 0.000 | 0.383 | 0.000 |
| 35 | 0.000 | 0.000 | 0.145 | 0.000 | 0.000 | 0.646 | 0.000 | 0.000 | 0.311 | 0.000 | 0.000 | 0.000 | 0.000 | 0.440 | 0.000 |
| 37 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.423 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 39 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.496 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 41 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.327 | 0.250 | 0.000 | 0.000 | 0.000 | 0.307 | 0.000 | 0.000 | 0.382 | 0.000 |
| 43 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.349 | 0.000 | 0.000 | 0.000 | 0.430 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 47 | 0.000 | 0.000 | 0.094 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.424 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 49 | 0.305 | 0.509 | 0.000 | 0.000 | 0.000 | 0.721 | 0.000 | 0.000 | 0.000 | 0.758 | 0.000 | 0.000 | 0.000 | 0.740 | 0.000 |
| 51 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.415 | 0.000 | 0.000 | 0.128 | 0.000 | 0.000 | 0.000 |
| 52 | 0.000 | 0.000 | 0.134 | 0.000 | 0.000 | 0.336 | 0.000 | 0.000 | 0.000 | 0.446 | 0.000 | 0.090 | 0.000 | 0.415 | 0.000 |
| 53 | 0.106 | 0.000 | 0.000 | 0.000 | 0.406 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| 55 | 0.000 | 0.000 | 0.000 | 0.122 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.371 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| user1 | user2 | user3 | user4 | user5 | user6 | user7 | user8 | user9 | user10 | user11 | user12 | user13 | user14 | user15 | user16 | user17 | user18 | user19 | user20 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| user1 | 1.000 | 0.536 | 0.794 | 0.782 | 0.510 | 0.729 | 0.807 | 0.663 | 0.708 | 0.583 | 0.000 | 0.000 | 0.517 | 0.788 | 0.648 | 0.837 | 0.000 | 0.906 | 0.000 | 0.674 |
| user2 | 0.536 | 1.000 | 0.621 | 0.325 | 0.422 | 0.486 | 0.349 | 0.441 | 0.277 | 0.466 | 0.370 | 0.370 | 0.401 | 0.607 | 0.447 | 0.451 | 0.354 | 0.488 | 0.000 | 0.419 |
| user3 | 0.794 | 0.621 | 1.000 | 0.684 | 0.335 | 0.481 | 0.707 | 0.543 | 0.623 | 0.779 | 0.520 | 0.520 | 0.517 | 0.706 | 0.530 | 0.774 | 0.497 | 0.831 | 0.000 | 0.516 |
| user4 | 0.782 | 0.325 | 0.684 | 1.000 | 0.112 | 0.653 | 0.920 | 0.737 | 0.884 | 0.304 | 0.000 | 0.000 | 0.352 | 0.572 | 0.720 | 0.704 | 0.097 | 0.844 | 0.000 | 0.701 |
| user5 | 0.510 | 0.422 | 0.335 | 0.112 | 1.000 | 0.250 | 0.000 | 0.000 | 0.078 | 0.562 | 0.000 | 0.000 | 0.379 | 0.468 | 0.000 | 0.379 | 0.100 | 0.393 | 0.000 | 0.000 |
| user6 | 0.729 | 0.486 | 0.481 | 0.653 | 0.250 | 1.000 | 0.705 | 0.885 | 0.556 | 0.029 | 0.000 | 0.000 | 0.000 | 0.687 | 0.901 | 0.475 | 0.000 | 0.585 | 0.000 | 0.841 |
| user7 | 0.807 | 0.349 | 0.707 | 0.920 | 0.000 | 0.705 | 1.000 | 0.753 | 0.836 | 0.357 | 0.000 | 0.000 | 0.369 | 0.585 | 0.784 | 0.729 | 0.000 | 0.872 | 0.000 | 0.716 |
| user8 | 0.663 | 0.441 | 0.543 | 0.737 | 0.000 | 0.885 | 0.753 | 1.000 | 0.628 | 0.000 | 0.000 | 0.000 | 0.000 | 0.776 | 0.976 | 0.537 | 0.000 | 0.624 | 0.000 | 0.950 |
| user9 | 0.708 | 0.277 | 0.623 | 0.884 | 0.078 | 0.556 | 0.836 | 0.628 | 1.000 | 0.302 | 0.000 | 0.000 | 0.350 | 0.488 | 0.613 | 0.644 | 0.067 | 0.761 | 0.443 | 0.597 |
| user10 | 0.583 | 0.466 | 0.779 | 0.304 | 0.562 | 0.029 | 0.357 | 0.000 | 0.302 | 1.000 | 0.365 | 0.365 | 0.694 | 0.477 | 0.037 | 0.656 | 0.349 | 0.688 | 0.000 | 0.000 |
| user11 | 0.000 | 0.370 | 0.520 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.365 | 1.000 | 1.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.956 | 0.000 | 0.000 | 0.000 |
| user12 | 0.000 | 0.370 | 0.520 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.365 | 1.000 | 1.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.956 | 0.000 | 0.000 | 0.000 |
| user13 | 0.517 | 0.401 | 0.517 | 0.352 | 0.379 | 0.000 | 0.369 | 0.000 | 0.350 | 0.694 | 0.000 | 0.000 | 1.000 | 0.322 | 0.000 | 0.573 | 0.000 | 0.587 | 0.000 | 0.000 |
| user14 | 0.788 | 0.607 | 0.706 | 0.572 | 0.468 | 0.687 | 0.585 | 0.776 | 0.488 | 0.477 | 0.000 | 0.000 | 0.322 | 1.000 | 0.758 | 0.739 | 0.000 | 0.782 | 0.000 | 0.738 |
| user15 | 0.648 | 0.447 | 0.530 | 0.720 | 0.000 | 0.901 | 0.784 | 0.976 | 0.613 | 0.037 | 0.000 | 0.000 | 0.000 | 0.758 | 1.000 | 0.524 | 0.000 | 0.650 | 0.000 | 0.928 |
| user16 | 0.837 | 0.451 | 0.774 | 0.704 | 0.379 | 0.475 | 0.729 | 0.537 | 0.644 | 0.656 | 0.000 | 0.000 | 0.573 | 0.739 | 0.524 | 1.000 | 0.000 | 0.880 | 0.055 | 0.510 |
| user17 | 0.000 | 0.354 | 0.497 | 0.097 | 0.100 | 0.000 | 0.000 | 0.000 | 0.067 | 0.349 | 0.956 | 0.956 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.037 | 0.000 | 0.000 |
| user18 | 0.906 | 0.488 | 0.831 | 0.844 | 0.393 | 0.585 | 0.872 | 0.624 | 0.761 | 0.688 | 0.000 | 0.000 | 0.587 | 0.782 | 0.650 | 0.880 | 0.037 | 1.000 | 0.000 | 0.593 |
| user19 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.443 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.055 | 0.000 | 0.000 | 1.000 | 0.000 |
| user20 | 0.674 | 0.419 | 0.516 | 0.701 | 0.000 | 0.841 | 0.716 | 0.950 | 0.597 | 0.000 | 0.000 | 0.000 | 0.000 | 0.738 | 0.928 | 0.510 | 0.000 | 0.593 | 0.000 | 1.000 |
Please add your trained model files for dl-predictor to
Model/predictor-dl-model/experiments
Dears,
we have a requirement to use DL predictor to predict both ad requests and impressions in parallel and save the predictions of impressions and ad-requests into different output tables and elastic search indexes. Can we discuss this feasibility today ?
is it possible to organize a meeting before Thursday's weekly meeting?
If you have a different process for dlpredictor to process huge log files, please add it to
/Processes/dlpredictor/experiments/
In dlpredictor pipeline/main_cluster, the uckeys with 0 imp in ts (training window) are removed before the calculation of std and mean.
This modification resulted to have fewer uckeys but the overall performance of the model stayed intact.
The new overall error rate at slot-id is 11.1% compared to previous 10.7%
Describe the bug
This is a test.
Training scenario:
Following datasets details include users with minimum one click count with step = 10.
test_dataset_count = 110755727,
train_dataset_count = 517801469,
user_count = 94315979,
item_count = 19
EPOCH = 250
train_batch_size = 20480
test_batch_size = 2048
Current model takes around 12 hours to train 1 epoch if we use all datasets. If we use around 50% datasets by randomly selecting
then also model takes around 7-8 hours to train for 1 epoch.
By this analogy, If we want to train the model for complete 250 epochs on full datasets, then it will take around 125 days.
Currently we are using Tensorflow 1.15, Two GPU are there in training but only one GPU is used.
Target about model.
1. It is required that model should not take more than 24 hours to train.
2. Model should be able to use all the available GPU.
3. Is it possible to further reduce the datasets with regard to size without losing insights.
4. Is it possible to get DIN-Lookalike model and trainer code in Tensorflow 2.0 version.
As per the weekly meeting discussions, the hyper parameter "train_skip_first" value 12 was the reason for difference in predictions using the model you shared and the model we trained. Below is the predicted data and accuracy before and after updating the train_skip_first.
If this value need to be used for train_skip_first, please update the hparams.py file
| Slot id | Actual values | Model trained Using feeder files shared by BM with train_skip_first = 0 | Predictions using model trained after changing train_skip_first = 12 |
|---|---|---|---|
| a47eavw7ex | 618,874,425 | 710379690 | 679778335 |
| 66bcd2720e5011e79bc8fa163e05184e | 141,771,777 | 150189391 | 134106959 |
| x0ej5xhk60kjwq | 140,838,705 | 166090973 | 150899108 |
| l03493p0r3 | 111,445,915 | 95256513 | 97194166 |
| 7b0d7b55ab0c11e68b7900163e3e481d | 105,754,734 | 110975030 | 89637188 |
| b6le0s4qo8 | 101,684,589 | 116538098 | 108032992 |
| e351de37263311e6af7500163e291137 | 77,548,938 | 85192615 | 69053892 |
| a290af82884e11e5bdec00163e291137 | 72,593,126 | 78752741 | 62835164 |
| 68bcd2720e5011e79bc8fa163e05184e | 48,699,653 | 52896931 | 43842830 |
| f1iprgyl13 | 38,603,295 | 40964141 | 33260309 |
| w3wx3nv9ow5i97 | 35,212,394 | 32632932 | 29544796 |
| w9fmyd5r0i | 33,533,582 | 35375640 | 31485616 |
| d971z9825e | 26,376,462 | 29203099 | 20635427 |
| l2d4ec6csv | 21,556,143 | 27453184 | 27377759 |
| z041bf6g4s | 17,440,168 | 25409583 | 6177790 |
| 71bcd2720e5011e79bc8fa163e05184e | 12,459,624 | 10217224 | 8926710 |
| 5cd1c663263511e6af7500163e291137 | 7,096,585 | 6485828 | 5206453 |
| x2fpfbm8rt | 6,390,364 | 6004435 | 4104797 |
| d9jucwkpr3 | 4,475,456 | 6371902 | 4366933 |
| k4werqx13k | 4,464,357 | 4717659 | 3375137 |
| a8syykhszz | 886,345 | 880801 | 790722 |
| j1430itab9wj3b | 864,567 | 698216 | 797136 |
| s4z85pd1h8 | 637,263 | 714682 | 562948 |
| d4d7362e879511e5bdec00163e291137 | 188,922 | 42751 | 35520 |
| 17dd6d8098bf11e5bdec00163e291137 | 168,582 | 60899 | 92402 |
Hello @jimmylao,
I was wondering that whether doing operations on negative and positive sampling would exclude some of the device IDs out from training data ?
If it does then we might get an error at the time of ctr generation for those DIDs which were not present during the training.
Wanted to know your opinion on this.
For 1.2m user records which produce 4.3m training and 1m test datasets, it takes more than 3 days (250 epochs) to train a lookalike model; 20 minutes for each epoch.
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 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.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
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.