Skip to content

Commit

Permalink
fix pagerank
Browse files Browse the repository at this point in the history
  • Loading branch information
@huseinzol05
huseinzol05 committed Jul 10, 2020
1 parent 5928f3d commit a7ecfc7
Show file tree
Hide file tree
Showing 7 changed files with 269 additions and 260 deletions.
4 changes: 3 additions & 1 deletion docs/GPU.rst
View file
Original file line number Diff line number Diff line change
Expand Up @@ -16,9 +16,11 @@ After that simply install gpu version,
GPU Version Benefit
--------------------

1. Different models different GPUs.
1. Limit GPU memory.
2. Automatically try to use cugraph for any networkx functions.

**We will add more GPU benefits in the future**.

Different models different GPUs
----------------------------------

Expand Down
155 changes: 77 additions & 78 deletions docs/gpu-environment.rst
View file
Original file line number Diff line number Diff line change
@@ -1,26 +1,27 @@
One model always consumed one unit of gpu. For now we do not support
distributed batch processing to multiple GPUs from one model. But we can
initiate multiple models to multiple GPUs.
.. code:: python
model_emotion -> GPU0
%%time
import malaya
model_sentiment -> GPU1
model_translation -> GPU2
.. parsed-literal::
CPU times: user 5.79 s, sys: 2.45 s, total: 8.24 s
Wall time: 3.63 s
and so on.
.. code:: python
%%time
import malaya
malaya.gpu_available()
.. parsed-literal::
CPU times: user 5.94 s, sys: 2.35 s, total: 8.29 s
Wall time: 4.01 s
True
.. code:: python
Expand All @@ -30,33 +31,31 @@ and so on.
.. parsed-literal::
Tue Jul 7 21:32:37 2020
Fri Jul 10 12:39:26 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.129 Driver Version: 410.129 CUDA Version: 10.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla V100-DGXS... On | 00000000:07:00.0 On | 0 |
| N/A 55C P0 42W / 300W | 0MiB / 32475MiB | 0% Default |
| N/A 43C P0 39W / 300W | 0MiB / 32475MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla V100-DGXS... On | 00000000:08:00.0 Off | 0 |
| N/A 65C P0 250W / 300W | 31452MiB / 32478MiB | 89% Default |
| N/A 45C P0 39W / 300W | 0MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 2 Tesla V100-DGXS... On | 00000000:0E:00.0 Off | 0 |
| N/A 63C P0 270W / 300W | 31452MiB / 32478MiB | 92% Default |
| N/A 44C P0 38W / 300W | 0MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 3 Tesla V100-DGXS... On | 00000000:0F:00.0 Off | 0 |
| N/A 63C P0 252W / 300W | 31452MiB / 32478MiB | 77% Default |
| N/A 44C P0 40W / 300W | 0MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 1 11646 C python3 31431MiB |
| 2 11646 C python3 31431MiB |
| 3 11646 C python3 31431MiB |
| No running processes found |
+-----------------------------------------------------------------------------+
Expand All @@ -65,18 +64,18 @@ Right now all the GPUs in resting mode, no computation happened.
GPU Rules
---------

1. By default, all models will initiate in first GPU, unless override
``gpu`` parameter in any load model API. Example as below.
2. Malaya will not consumed all available GPU memory, but slowly grow
1. Malaya will not consumed all available GPU memory, but slowly grow
based on batch size. This growth only towards positive (use more GPU
memory) dynamically, but will not reduce GPU memory if feed small
batch size.
3. Use ``malaya.clear_session`` to clear session from unused models but
2. Use ``malaya.clear_session`` to clear session from unused models but
this will not free GPU memory.
4. By default Malaya will not set max cap for GPU memory, to put a cap,
3. By default Malaya will not set max cap for GPU memory, to put a cap,
override ``gpu_limit`` parameter in any load model API. ``gpu_limit``
should 0 < ``gpu_limit`` < 1. If ``gpu_limit = 0.3``, it means the
model will not use more than 30% of GPU memory.
4. Even if you installed Malaya CPU version, it will always to load the
models in GPU first, if failed, it will load it in CPU.

.. code:: python
Expand All @@ -89,20 +88,20 @@ GPU Rules
.. code:: python
model = malaya.emotion.transformer(model = 'bert', gpu = '0')
model = malaya.emotion.transformer(model = 'bert', gpu_limit = 0.5)
.. parsed-literal::
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:61: The name tf.gfile.GFile is deprecated. Please use tf.io.gfile.GFile instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:72: The name tf.gfile.GFile is deprecated. Please use tf.io.gfile.GFile instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:62: The name tf.GraphDef is deprecated. Please use tf.compat.v1.GraphDef instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:73: The name tf.GraphDef is deprecated. Please use tf.compat.v1.GraphDef instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:50: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:58: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:51: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:61: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:53: The name tf.InteractiveSession is deprecated. Please use tf.compat.v1.InteractiveSession instead.
WARNING:tensorflow:From /home/husein/malaya/Malaya/malaya/function/__init__.py:63: The name tf.InteractiveSession is deprecated. Please use tf.compat.v1.InteractiveSession instead.
Expand All @@ -117,50 +116,50 @@ GPU Rules
.. parsed-literal::
CPU times: user 1.94 s, sys: 541 ms, total: 2.48 s
Wall time: 2.52 s
CPU times: user 1.8 s, sys: 504 ms, total: 2.3 s
Wall time: 2.3 s
.. parsed-literal::
[{'anger': 0.9998965,
'fear': 1.7692768e-05,
'happy': 1.8747674e-05,
'love': 1.656881e-05,
'sadness': 3.130815e-05,
'surprise': 1.9183277e-05},
{'anger': 7.4469484e-05,
'fear': 0.99977416,
'happy': 6.824215e-05,
'love': 2.773282e-05,
'sadness': 1.9767067e-05,
'surprise': 3.5663204e-05},
{'anger': 0.99963737,
'fear': 3.931449e-05,
'happy': 0.0001562279,
'love': 3.3580774e-05,
'sadness': 0.00011328616,
'surprise': 2.0134145e-05},
{'anger': 3.1319763e-05,
'fear': 1.7286226e-05,
'happy': 2.9899325e-05,
'love': 0.99987257,
'sadness': 2.7867774e-05,
'surprise': 2.096328e-05},
{'anger': 8.965934e-05,
'fear': 1.8196944e-05,
'happy': 2.9275663e-05,
'love': 1.7211949e-05,
'sadness': 0.9998247,
'surprise': 2.0944033e-05},
{'anger': 4.132152e-05,
'fear': 6.202527e-05,
'happy': 3.1012056e-05,
'love': 5.3896296e-05,
'sadness': 6.202101e-05,
'surprise': 0.9997497}]
[{'anger': 0.99989223,
'fear': 1.5843118e-05,
'happy': 1.660186e-05,
'love': 1.9634477e-05,
'sadness': 3.827092e-05,
'surprise': 1.7427232e-05},
{'anger': 4.894743e-05,
'fear': 0.999795,
'happy': 6.764499e-05,
'love': 3.6289443e-05,
'sadness': 1.9702624e-05,
'surprise': 3.2430926e-05},
{'anger': 0.9997905,
'fear': 2.5795038e-05,
'happy': 6.7572015e-05,
'love': 2.6636817e-05,
'sadness': 6.734582e-05,
'surprise': 2.2285754e-05},
{'anger': 2.4449551e-05,
'fear': 2.6033362e-05,
'happy': 3.1518703e-05,
'love': 0.9998758,
'sadness': 1.895303e-05,
'surprise': 2.326243e-05},
{'anger': 8.095824e-05,
'fear': 2.3824483e-05,
'happy': 2.1045413e-05,
'love': 1.6150812e-05,
'sadness': 0.99983835,
'surprise': 1.9708685e-05},
{'anger': 4.470948e-05,
'fear': 0.00010641558,
'happy': 2.9055469e-05,
'love': 4.5270677e-05,
'sadness': 5.7159534e-05,
'surprise': 0.9997173}]
Expand All @@ -171,34 +170,34 @@ GPU Rules
.. parsed-literal::
Tue Jul 7 21:32:57 2020
Fri Jul 10 12:39:56 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.129 Driver Version: 410.129 CUDA Version: 10.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla V100-DGXS... On | 00000000:07:00.0 On | 0 |
| N/A 56C P0 58W / 300W | 1099MiB / 32475MiB | 0% Default |
| N/A 44C P0 54W / 300W | 1099MiB / 32475MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla V100-DGXS... On | 00000000:08:00.0 Off | 0 |
| N/A 64C P0 219W / 300W | 31452MiB / 32478MiB | 99% Default |
| N/A 45C P0 52W / 300W | 418MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 2 Tesla V100-DGXS... On | 00000000:0E:00.0 Off | 0 |
| N/A 62C P0 248W / 300W | 31452MiB / 32478MiB | 99% Default |
| N/A 44C P0 51W / 300W | 418MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 3 Tesla V100-DGXS... On | 00000000:0F:00.0 Off | 0 |
| N/A 62C P0 236W / 300W | 31452MiB / 32478MiB | 76% Default |
| N/A 45C P0 54W / 300W | 418MiB / 32478MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 2536 C /usr/bin/python3 1087MiB |
| 1 11646 C python3 31431MiB |
| 2 11646 C python3 31431MiB |
| 3 11646 C python3 31431MiB |
| 0 35310 C /usr/bin/python3 1087MiB |
| 1 35310 C /usr/bin/python3 407MiB |
| 2 35310 C /usr/bin/python3 407MiB |
| 3 35310 C /usr/bin/python3 407MiB |
+-----------------------------------------------------------------------------+
Expand Down
8 changes: 7 additions & 1 deletion docs/running-on-windows.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,13 @@
"source": [
"## Unable to use any T5 models\n",
"\n",
"T5 depends on tensorflow-text, currently there is no official tensorflow-text binary released for Windows. So no T5 model for Windows users."
"T5 depends on tensorflow-text, currently there is no official tensorflow-text binary released for Windows. So no T5 model for Windows users.\n",
"\n",
"List T5 models,\n",
"\n",
"1. [malaya.summarization.abstractive.t5](https://malaya.readthedocs.io/en/latest/Abstractive.html#load-t5)\n",
"2. [malaya.generator.t5](https://malaya.readthedocs.io/en/latest/Generator.html#load-t5)\n",
"3. [malaya.paraphrase.t5](https://malaya.readthedocs.io/en/latest/Paraphrase.html#load-t5-models)"
]
},
{
Expand Down
6 changes: 6 additions & 0 deletions docs/running-on-windows.rst
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,12 @@ T5 depends on tensorflow-text, currently there is no official
tensorflow-text binary released for Windows. So no T5 model for Windows
users.

List T5 models,

1. `malaya.summarization.abstractive.t5 <https://malaya.readthedocs.io/en/latest/Abstractive.html#load-t5>`__
2. `malaya.generator.t5 <https://malaya.readthedocs.io/en/latest/Generator.html#load-t5>`__
3. `malaya.paraphrase.t5 <https://malaya.readthedocs.io/en/latest/Paraphrase.html#load-t5-models>`__

Lack development on Windows
---------------------------

Expand Down
Loading

0 comments on commit a7ecfc7

Please sign in to comment.