<![CDATA[
# <p class="text-h">Homework 6 - Malicious Comments Identification</p>
### <p class="text-hh">Announcements</p>
### 5/3
* train_x.csv has duplicate sentences from id 119018-119999
#### 5/2
* HW6 sample code release!
* Strong baseline release!
#### 4/25
* HW6 release!
* HW6 raw data in ceiba!
<hr>
### <p class="text-hh">Links</p>
* 作業投影片 <a href="https://docs.google.com/presentation/d/1KJAYYM-n_DqOSWEwrPh52yVO1rzkJDBc8WgjLVWpB_E/edit?usp=sharing" target="_blank"><i class="fa fa-book"></i></a>
* Kaggle 連結 <a href="https://www.kaggle.com/c/ml2019spring-hw6/" target="_blank"><i class="fa fa-trophy"></i></a>
* Report 模板 <a href="https://docs.google.com/document/d/1bWaZfjG8g8fPe2VxzuWV8IuMRcmGztPUD7fR19iB7Cs/edit?usp=sharing" target="_blank"><i class="fa fa-file-text"></i></a>
* 遲交表單 <a href="https://docs.google.com/forms/d/e/1FAIpQLSdsspthuruApgcizvnoUW2OdBCEhJ24_LLYjXAw1gpYyQMggw/viewform" target="_blank"><i class="fa fa-pencil-square-o"></i></a>
* Sample code <a href="https://hackmd.io/s/S1YZNj4iE" target="_blank"><i class="fa fa-book"></i></a>
* Facebook Discussion <a href="https://www.facebook.com/groups/314613059175222/permalink/351075982195596/" target="_blank"><i class="fa fa-facebook-square"></i></a>
<!--* 小老師表單 <a href="https://goo.gl/feHp6o" target="_blank"><i class="fa fa-pencil-square-o"></i></a>-->
<hr>
### <p class="text-hh">Deadlines</p>
* Simple Bonus Deadline: 05/02/2019 11:59:59 (GMT+8)
* Kaggle Deadline: 05/09/2019 11:59:59 (GMT+8)
* Github Deadline: 05/10/2019 23:59:59 (GMT+8)
<hr>
### <p class="text-hh">Assignment Regulation</p>
* ALL code must be written in python3.6
* For `hw6_train.sh` and `hw6_test.sh` :
* ALL python standard library is permitted (e.g. sys, csv, time)
* numpy >=1.14
* pandas >= 0.24.1
* PyTorch 1.0.1, TensorFlow 1.12.0, Keras == 2.2.4
* jieba 0.39
* gensim 3.7.1 ( 只可使用 word2vec api !! )
* emoji 0.5.1
<hr>
### <p class="text-hh">FAQ</p>
<p>Q1：請問kaggle的組隊人數上限？</p>
<p>A1：hw6為個人作業，不用在kaggle上進行組隊。</p>
<p>Q2：reproduce 規則？</p>
<p>A2：
1. reproduce時間限制10分鐘不包含下載時間
<br>2. reproduce誤差為±0.2%
<br>3. reproduce標準為kaggle勾選的二擇一即可
<br>4. baseline分數以kaggle為準
</p>
<hr>
]]>