Upload 2 files
Browse files
optimizer/How-to-Use-EmoNAVI(ENG).txt
CHANGED
|
@@ -26,4 +26,34 @@ If you find that your training isn't progressing beyond a certain point, try an
|
|
| 26 |
|
| 27 |
Conclusion
|
| 28 |
|
| 29 |
-
We hope this explanation helps you acquire valuable know-how for setting up your training, not just for the EmoNavi series. We believe it will be beneficial to all of you. Thank you for reading to the end.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
Conclusion
|
| 28 |
|
| 29 |
+
We hope this explanation helps you acquire valuable know-how for setting up your training, not just for the EmoNavi series. We believe it will be beneficial to all of you. Thank you for reading to the end.
|
| 30 |
+
|
| 31 |
+
postscript
|
| 32 |
+
|
| 33 |
+
I'd like to explain the learning rate in an easy-to-understand way, so you can truly grasp its concept.
|
| 34 |
+
|
| 35 |
+
You can think of the learning rate like reading speed.
|
| 36 |
+
Imagine this: a high learning rate is like skim reading (or speed reading), while a low learning rate is like perusing (or close reading).
|
| 37 |
+
|
| 38 |
+
The scheduler manages this, much like a learning schedule.
|
| 39 |
+
EmoNAVI has a "shadow" function that encourages the model to review and reflect on its own learning progress.
|
| 40 |
+
With EmoNAVI, you have a choice: you can allow external guidance to determine the learning path and let the model's autonomy supplement it, or you can rely solely on its autonomy.
|
| 41 |
+
|
| 42 |
+
Here are some other analogies:
|
| 43 |
+
High learning rate: Like shooting a photo from a distance, giving you an overview where details are fuzzy.
|
| 44 |
+
Low learning rate: Like shooting up close, capturing accurate details.
|
| 45 |
+
Think of autofocus as being handled by the scheduler and the "shadow" function.
|
| 46 |
+
From another perspective:
|
| 47 |
+
|
| 48 |
+
When aiming for detailed expressions, you can also consider increasing the amount of training data or increasing the number of iterations.
|
| 49 |
+
As the number of iterations increases, detailed features are gradually accumulated.
|
| 50 |
+
|
| 51 |
+
However, color representation largely depends on the performance of the VAE (Variational Autoencoder).
|
| 52 |
+
To accurately reflect colors, the only options are to improve the VAE's performance itself or to use teacher data that correctly reflects colors.
|
| 53 |
+
|
| 54 |
+
Furthermore, the "shadow" function also acts like an autofocus system.
|
| 55 |
+
It's a mechanism that allows the model to review and reflect on its own learning, essentially learning from its own experience.
|
| 56 |
+
This means it captures one feature, learns from it, then identifies another, and the process repeats.
|
| 57 |
+
Consequently, its "focus" (or understanding) continuously evolves and adapts.
|
| 58 |
+
|
| 59 |
+
That concludes the additional explanation. Thank you for reading to the end!
|
optimizer/学習の進め方(日本語).txt
CHANGED
|
@@ -24,4 +24,29 @@ emonavi系はコンスタントで回し続けても過学習にはなりませ
|
|
| 24 |
もし学習が一定のところから進まない、と感じたときは、追加学習で低学習率へ変更してください。そうすると詳細を一気に吸収し始めます。
|
| 25 |
|
| 26 |
<謝意>
|
| 27 |
-
emonavi系に限らず、この説明で学習設定のノウハウの獲得に寄与できれば嬉しいです。皆さまのお役に立てれば幸いです。最後までご覧いただきありがとうございました。
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
もし学習が一定のところから進まない、と感じたときは、追加学習で低学習率へ変更してください。そうすると詳細を一気に吸収し始めます。
|
| 25 |
|
| 26 |
<謝意>
|
| 27 |
+
emonavi系に限らず、この説明で学習設定のノウハウの獲得に寄与できれば嬉しいです。皆さまのお役に立てれば幸いです。最後までご覧いただきありがとうございました。
|
| 28 |
+
|
| 29 |
+
<追記>
|
| 30 |
+
学習率について実感を得られるように、わかりやすく伝えたい、と思っています。
|
| 31 |
+
|
| 32 |
+
学習率とは、読書の速さにも置き換えられると思います、
|
| 33 |
+
学習率高:飛ばし読み(速読)、
|
| 34 |
+
学習率低:熟読(精読)、と想像してください。
|
| 35 |
+
スケジューラーはこれを学習予定として管理します
|
| 36 |
+
|
| 37 |
+
emonavi は shadow の機能で、モデル自身の復習や振り返りを促し学習を進行します
|
| 38 |
+
進み方を外部に決めさせ自律で補うか、自律のみに任せるか、になります。
|
| 39 |
+
|
| 40 |
+
他にも、以下のように例えることも可能です、
|
| 41 |
+
学習率高:遠くからの撮影=概要(細部はあいまい)
|
| 42 |
+
学習率低:寄りで撮影=詳細(細部を正確に)
|
| 43 |
+
オートフォーカス:スケジューラー、shadow、
|
| 44 |
+
|
| 45 |
+
別視点からも説明しますと、細部表現を獲得したい場合は教師データを増やす、ことでも可能です。
|
| 46 |
+
繰り返し数が増加することで細部の特徴も少しづつ蓄積される、となります。
|
| 47 |
+
ただし、色についてはVAEの性能に依拠する部分が多く、これを正しく反映できる教師データか、VAEの性能の向上しかありません。
|
| 48 |
+
|
| 49 |
+
それとですね、shadow はオートフォーカスでもありますが、これは学習の振り返り、復習をするもので、自分自身の経験に学ぶ仕組みです、
|
| 50 |
+
ですから、特徴を捉えて学んで、別の特徴を見つけて、、を繰り返す、その結果としてフォーカスもピントも変化し続けるようになります。
|
| 51 |
+
|
| 52 |
+
以上となります。追記も最後までご覧頂いてありがとうございました。
|