Skip to content

Commit 35e0c8c

Browse files
committed
Images Poc structure + improve transition nb
1 parent b3f3154 commit 35e0c8c

22 files changed

+75
-89
lines changed
-137 KB
Binary file not shown.
205 KB
Loading
207 KB
Loading
206 KB
Loading
205 KB
Loading

talks/pyData/img/poc_dev.jpg

-62.5 KB
Binary file not shown.

talks/pyData/img/poc_dev1.png

22.5 KB
Loading

talks/pyData/img/poc_dev2.png

26.6 KB
Loading

talks/pyData/img/poc_dev3.png

28.8 KB
Loading

talks/pyData/img/poc_dev4.png

31.1 KB
Loading

talks/pyData/img/poc_ml.jpg

-67.5 KB
Binary file not shown.

talks/pyData/img/poc_ml1.png

34.5 KB
Loading

talks/pyData/img/poc_ml2.png

37.8 KB
Loading

talks/pyData/img/poc_ml3.png

39.8 KB
Loading

talks/pyData/img/poc_ml4.png

40.7 KB
Loading

talks/pyData/img/poc_worst.jpg

-70.9 KB
Binary file not shown.

talks/pyData/img/poc_worst.png

58.5 KB
Loading

talks/pyData/img/poc_worst1.png

51.4 KB
Loading

talks/pyData/img/poc_worst2.png

53.1 KB
Loading

talks/pyData/img/poc_worst3.png

57.8 KB
Loading

talks/pyData/notebooks.html

+33-75
Original file line numberDiff line numberDiff line change
@@ -20,18 +20,26 @@ <h4>... are terrible! </h4>
2020
</div>
2121
</div>
2222
<aside class="notes">
23+
@SDG Great part
2324
They are even used by Netflix!
25+
26+
@SBI terrible part
27+
- before ML though it was just a tool for demo
2428
</aside>
2529
</section>
2630
<section>
2731
<p>Copy-pasting </p>
2832
<aside class="notes">
33+
@SDG
2934
End of the joke
3035
</aside>
3136
</section>
3237

3338
<section>
3439
<p>Converting Jupyter Notebooks to <span class="emph"> scripts </span></p>
40+
<aside class="notes">
41+
@SDG or @SBI
42+
</aside>
3543
</section>
3644

3745
<section>
@@ -53,10 +61,10 @@ <h4>... are terrible! </h4>
5361
</ul>
5462
</div>
5563
<div class="fragment"><span class="emph">No effect cells</span></div>
56-
5764
</div>
5865
</div>
5966
<aside class="notes">
67+
@SDG
6068
This is a notebook
6169
- IO params
6270
- a mode
@@ -71,55 +79,24 @@ <h5><span class="emph">nbconvert</span></h5>
7179
<img src="./img/nb_convert/script.png"/>
7280

7381
<aside class="notes">
74-
This is a notebook
75-
- IO params
76-
- a mode
77-
- some no effect cells
82+
@SBI
7883
</aside>
7984
</section>
8085
<!-- ------------------------- NB convert BEGIN-->
81-
<section data-transition="slide fade-out">
82-
<div class="two-halves">
83-
<div class="big-half">
84-
<img src="./img/nb_convert/formated_script.png"/>
85-
</div>
86-
</div>
87-
88-
<aside class="notes">
89-
Same script but formatted
90-
- not configurable
91-
</aside>
92-
</section>
93-
<section data-transition="slide fade-out">
94-
<div class="two-halves">
95-
<div class="big-half">
96-
<img src="./img/nb_convert/formated_script_not_conf.png"/>
97-
</div>
98-
<ul class="small-half">
99-
<li class="fragment"><span class="emph">Not configurable</span></li>
100-
</ul>
101-
</div>
102-
103-
<aside class="notes">
104-
Same script but formatted
105-
- not configurable
106-
</aside>
107-
</section>
108-
<section data-transition="slide fade-out">
86+
<section>
10987
<div class="two-halves">
11088
<div class="big-half">
111-
<img src="./img/nb_convert/formated_script_no_effect.png"/>
89+
<img class="fragment current-visible" data-fragment-index="2" src="./img/nb_convert/formated_script_not_conf.png">
90+
<img class="fragment current-visible" data-fragment-index="3" src="./img/nb_convert/formated_script_no_effect.png">
11291
</div>
11392
<ul class="small-half">
114-
<li><span class="emph">Not configurable</span></li>
115-
<li class="fragment"><span class="emph">Remaining no effect cells</span></li>
93+
<li class="fragment" data-fragment-index="2" >Not configurable</li>
94+
<li class="fragment" data-fragment-index="3" >Remaining no effect cells</li>
11695
</ul>
11796
</div>
11897

11998
<aside class="notes">
120-
Same script but formatted
121-
- not effect cells
122-
-
99+
@SBI
123100
</aside>
124101
</section>
125102
<!-- ------------------------- NB convert END-->
@@ -146,12 +123,15 @@ <h5><span class="emph">MLV-tools: ipynb_to_python</span></h5><br/>
146123
<section>
147124
<div class="two-halves">
148125
<div class="big-half">
149-
<img src="./img/mlv_convert/script.png"/>
126+
<img class="fragment current-visible" data-fragment-index="1" src="./img/mlv_convert/script1.png">
127+
<img class="fragment current-visible" data-fragment-index="2" src="./img/mlv_convert/script2.png">
128+
<img class="fragment current-visible" data-fragment-index="3" src="./img/mlv_convert/script3.png">
129+
<img class="fragment current-visible" data-fragment-index="4" src="./img/mlv_convert/script4.png">
150130
</div>
151-
<ul class="small-half">
152-
<li class="fragment"><span class="emph">Configurable</span></li>
153-
<li class="fragment"><span class="emph">No more no effect cells</span></li>
154-
<li class="fragment"><span class="emph">Executable as a command</span></li>
131+
<ul class="small-half" data-fragment-index="1">
132+
<li class="fragment" data-fragment-index="2">Function with parameters</li>
133+
<li class="fragment" data-fragment-index="3">No more no effect cells</li>
134+
<li class="fragment" data-fragment-index="4">Executable as a command</li>
155135
</ul>
156136
</div>
157137

@@ -162,37 +142,15 @@ <h5><span class="emph">MLV-tools: ipynb_to_python</span></h5><br/>
162142
</aside>
163143
</section>
164144
<section>
165-
<p><pre><code>
166-
/tmp$ ./extract_dataset.py -h
167-
usage: extract_dataset.py [-h] --subset SUBSET --data-in DATA_IN --output-path
168-
OUTPUT_PATH
169-
170-
Command for script mlvtools_extract_dataset
171-
172-
optional arguments:
173-
-h, --help show this help message and exit
174-
--subset SUBSET Subset of data to load {'train', 'test'}
175-
--data-in DATA_IN File directory path
176-
--output-path OUTPUT_PATH
177-
Output file path
178-
</code></pre></p>
179-
180-
<aside class="notes">
181-
Same script but formatted
182-
- not effect cells
183-
-
184-
</aside>
185-
</section>
186-
<section>OR</section>
187-
<section>
188-
<img src="./img/mlv_convert/cmd.png">
189-
190-
<aside class="notes">
191-
</aside>
192-
</section>
193-
<section>
194-
<img src="./img/mlv_convert/cmd_param.png">
195-
145+
<pre><code>
146+
"""
147+
:param str subset: Subset of data to load {'train', 'test'}
148+
:param str data_in: File directory path
149+
:param str output_path: Output file path
150+
"""
151+
</code></pre>
152+
<img class="fragment current-visible" data-fragment-index="2" src="./img/mlv_convert/cmd.png">
153+
<img class="fragment current-visible" data-fragment-index="3" src="./img/mlv_convert/cmd_param.png">
196154
<aside class="notes">
197155
</aside>
198156
</section>

talks/pyData/presentation.html

+42-14
Original file line numberDiff line numberDiff line change
@@ -140,31 +140,28 @@ <h4>From ML experiments to production: </h4>
140140
<h3> Versioning and Reproducibility </br> with MLV-tools</h3>
141141
<p><i> <small> Stéphanie Bracaloni and Sarah Diot-Girard</small></i></p>
142142
<aside class="notes">
143-
143+
@SDG
144144
</aside>
145145
</section>
146146
<section>
147147
<h3>About Us</h3>
148148
<div class="two-halves">
149-
<div class="half">
149+
<div class="half" style="text-align: left">
150150
<b>Sarah Diot-Girard</b> </br>
151151
Data Scientist since 2012 </br>
152152
Interested in DataOps and ethics</br>
153153
<img class="icon" src="./img/GitHub-logo.png"/ alt="PeopleDoc logo" style="width: 1em"> @SgdJlbl
154154
</div>
155-
<div class="half">
155+
<div class="half" style="text-align: right">
156156
<b>Stéphanie Bracaloni</b> </br>
157-
Software Engineer [...]</br>
157+
Software Engineer since 2013</br>
158+
Automation and code Quality</br>
158159
<img class="icon" src="./img/GitHub-logo.png"/ alt="PeopleDoc logo" style="width: 1em"> @sbracaloni
159160
</div>
160161
</div>
161162
<img class="plain" src="./img/logo.png"/ alt="PeopleDoc logo" style="width: 4em">
162163
<aside class="notes">
163-
- Sarah : [petit resumé] + Accroche sur les technos => J'utilise des jupyter notebook
164-
et j'ai besoin de ...pouvoir reproduire facilement, garder de la souplesse, ...
165-
- Stephanie: [petit résumé] + Accroche Automatisation, Livaison, Tests
166-
J'ai besoin de... un truc qui se lance facilement, qui se package, qui soit reproductible
167-
sur n'importe quel environnement
164+
@SDG + @SBI
168165
</aside>
169166
</section>
170167
<section>
@@ -178,6 +175,7 @@ <h3>About Us</h3>
178175
<section>
179176
Monday morning, 9am
180177
<aside class="notes">
178+
@SDG
181179
It's Monday morning, 9am.
182180
You're an happy DS. Friday night, you finally made a breakthrough in your research project.
183181
you sent that to your boss.
@@ -186,21 +184,28 @@ <h3>About Us</h3>
186184
<section>
187185
This is a picture of a nice graph with POC results.
188186
<aside class="notes">
187+
@SDG
189188
You sent the results to your boss.
190189
She loves it. She want it in production asap.
191190
</aside>
192191
</section>
193192
<section>
194193
Monday morning, 10am
195194
<aside class="notes">
195+
@SDG (@SBI => "It's me")
196196
It's Monday morning, 10am.
197197
You are asking your SE coworker for help. You showed her your POC and now she looks like she's gonna faint
198198
and she wants to kill you AT THE SAME TIME!
199199
</aside>
200200
</section>
201201
<section>
202-
<img src="./img/poc_worst.jpg"/>
202+
<div>
203+
<img class="fragment current-visible preload" src="./img/poc_worst1.png">
204+
<img class="fragment current-visible" src="./img/poc_worst2.png">
205+
<img class="fragment current-visible" src="./img/poc_worst3.png">
206+
</div>
203207
<aside class="notes">
208+
@SDG
204209
- git repo with format not really compatible with git versioning
205210
- hardcoded stuff (path, user, ...)
206211
- hardcoded hyperparameters
@@ -209,24 +214,46 @@ <h3>About Us</h3>
209214
<section>
210215
Two weeks later
211216
<aside class="notes">
217+
@SDG
212218
Two weeks later.
213219
It's Monday morning, 9am.
214220
You have worked hard to clean up the POC. It's a bit better but...
215221
</aside>
216222
</section>
217223
<section>
218-
<img src="./img/poc_ml.jpg"/>
224+
<div>
225+
<img class="fragment current-visible preload" src="./img/poc_ml1.png">
226+
<img class="fragment current-visible" src="./img/poc_ml2.png">
227+
<img class="fragment current-visible" src="./img/poc_ml3.png">
228+
<img class="fragment current-visible" src="./img/poc_ml4.png">
229+
</div>
219230
<aside class="notes">
231+
@SDG
220232
Two weeks later.
221233
It's Monday morning, 9am.
222234
You have worked hard to clean up the POC. It's a bit better but...
223235
</aside>
224236
</section>
225237
<section>
226-
<img src="./img/poc_dev.jpg"/>
238+
<div>
239+
<img class="fragment current-visible preload" src="./img/poc_dev1.png">
240+
<img class="fragment current-visible" src="./img/poc_dev2.png">
241+
<img class="fragment current-visible" src="./img/poc_dev3.png">
242+
<img class="fragment current-visible" src="./img/poc_dev4.png">
243+
</div>
227244
<aside class="notes">
228-
It's Monday morning, 10am.
229-
Your coworker is trying to explain what you need to change so that she can bring your POC into production.
245+
@SBI
246+
Ok, If I well undstd POC => PROD
247+
So we need to:
248+
- remove Jup Nb
249+
- structure a python module
250+
- add tests
251+
- create a CI
252+
- [Most of all] REMOVE DATA
253+
=> long time we know how to structure CODE PRJ => tests auto and quality checks
254+
and GIT perfect for versioning
255+
256+
@SDG=> Yes, but you reached a point here! It is not just a CODE PRJ
230257
</aside>
231258
</section>
232259
</section>
@@ -236,6 +263,7 @@ <h3>POC vs PROD</h3>
236263
<h4 class="fragment fade-up"> vs Data Scientist </h4>
237264
<h4 class="fragment fade-up"> vs Software Engineer </h4>
238265
<aside class="notes">
266+
@SDG
239267
This is the story of
240268
</aside>
241269
</section>

0 commit comments

Comments
 (0)