issues
search
konumaru
/
commonLit_readability_prize
https://www.kaggle.com/c/commonlitreadabilityprize
0
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Target に関してDiscussionに散らばっている知見をまとめる。
#19
konumaru
opened
3 years ago
0
DataCollatorForLanguageModeling を試す
#18
konumaru
closed
3 years ago
4
pytorch の RandomSampler を使った BootstrapSampling の実装
#17
konumaru
closed
3 years ago
0
Stochastic Weight Averaging の検証
#16
konumaru
closed
3 years ago
3
2層目の予測にSentence2Doc の特徴用を追加する
#15
konumaru
opened
3 years ago
0
損失をMSEに加えて、Robertaが返すlogitも加算して学習した結果と比較する
#14
konumaru
closed
3 years ago
0
Differential Learning Rateを試す
#13
konumaru
closed
3 years ago
1
Multi Sample Dropoutを試す
#12
konumaru
closed
3 years ago
0
`init_weights` の設定
#11
konumaru
closed
3 years ago
0
Stacking の notebook を fork して２層目でアンサンブルを試す
#10
konumaru
closed
3 years ago
1
tokenizerの出力を `**inputs` で入力できるように変更する
#9
konumaru
closed
3 years ago
0
BatchSize=8, 16, 32 で精度比較を行う（ ← 収束の安定性が改善されるらしい）
#8
konumaru
closed
3 years ago
1
Experiment History
#7
konumaru
opened
3 years ago
2
Roberta と RobertaForSequenceClassification の違いについて調べる
#6
konumaru
closed
3 years ago
4
損失関数にはMSE, 評価関数にはRMSEを利用する
#5
konumaru
closed
3 years ago
0
Pretrained-model で予測してみる
#4
konumaru
closed
3 years ago
1
論文を読む
#3
konumaru
closed
3 years ago
2
実際のサービスを触る
#2
konumaru
closed
3 years ago
4
Upload Images
#1
konumaru
opened
3 years ago
4