AlphaCode + Codeforces Round #770 (Div. 2)
- masashinakata
- 4013
- 3
- 0
- 0
>競技プログラミングの電王戦はいつ 漠然と5年以内にいけるのではくらいに思ってたけど、電王戦というコンテスト形式に限らなければ、そういうAIはもう作れる時代が既に到来したという事実。 CopilotとCodeNetくっつけれれば出来ちゃうでしょ
2021-07-03 22:03:47ついに「AlphaCode」が登場したらしく、ベーシックインカム、頼んだぞという気持ち twitter.com/verge/status/1…
2022-02-03 01:03:24DeepMind says its new AI coding engine is as good as an average human programmer theverge.com/2022/2/2/22914… pic.twitter.com/jpZpRtpMwO
2022-02-03 01:00:12Google傘下のAI技術、DeepMindを使ったプログラミングAIのAlphaCode、平均的な人間プログラマーの能力に追いつく。囲碁の時もそいだったけれど、ここまで来ると後は速い。 いよいよソフトウェアが自己再生産を始める時代が幕開け。 twitter.com/BBCTech/status…
2022-02-03 01:08:15DeepMind AI rivals average human competitive coder bbc.in/3AU4zmo
2022-02-03 01:03:02Competitive programming with AlphaCode | DeepMind buff.ly/3HmmGE4 ヤバい、競技プログラミングAIが結構なレベルに到達したと…。自分レベルは越えられた可能性…
2022-02-03 01:52:21AlphaCodeを使用した競技プログラミング deepmind.com/blog/article/C…
2022-02-03 02:02:06deepmind.com/blog/article/C… DeepMind の競プロ AI。 GitHub で公開されてるコードと内製の(?)データセットから学習して、コードを大量に生成してから解候補を絞り込むらしい(?) > AlphaCode achieved an estimated rank within the top 54% of participants 54% ってどんなもんだろう?
2022-02-03 02:24:07AlphaCode! DeepMindがやるとヤバそうな雰囲気あるな…… twitter.com/DeepMind/statu…
2022-02-03 05:12:54Introducing #AlphaCode: a system that can compete at average human level in competitive coding competitions like @codeforces. An exciting leap in AI problem-solving capabilities, combining many advances in machine learning! Read more: dpmd.ai/Alpha-Code 1/ pic.twitter.com/Mvhc6Jm7E0
2022-02-03 01:11:03うおぉ、とうとうDeepMindのAI(AlphaCode)が競技プログラミングで成果!批判的思考、論理、アルゴリズム、コーディング、自然言語理解が必要な新しい問題を解き、平均的な参加者に匹敵。あとOpenAIが国際数学オリンピックの2つの問題を解くAIを学習 openai.com/blog/formal-ma… deepmind.com/blog/article/C…
2022-02-03 06:30:30DeepMindの問題解決型AIが競技プログラミングに挑戦、上位54%にランクイン | マイナビニュース DeepMindは2月2日、同社が開発した問題解決型AIによるコード生成システム「AlphaCode... newscollect.jp/article/?id=86…
2022-02-03 07:02:00自分はCodeforcesをほとんどやってないけど、直近10回のコンテストで上位54%ということならまだ負けてはない、かな。AtCoderと比べてレベル感がどうというのはピンとは来ないが deepmind.com/blog/article/C…
2022-02-03 07:16:20I spent 1000s of hours on competitive programming (proof-link: codeforces.com/profile/rizar). This makes me qualified to comment on #AlphaCode by @DeepMind The result is nice, the benchmark will be useful, some ideas are novel. But human level is still light years away. 1/n
2022-02-03 07:57:01The system ranks behind 54.3% participants. Note that many participants are high-school or college students who are just honing their problem-solving skills. Most people reading this could easily train to outperform #AlphaCode, especially if time pressure is removed...
2022-02-03 07:57:02Limited time (e.g. 3 hours to solve 6 problems) is a key difficulty in comp. programming. The baseline human is very constrained in this model-vs-human comparison. For #AlphaCode the pretraining data, the fine-tuning data, the model size, the sampling - all was nearly maxed out.
2022-02-03 07:57:02Importantly, the vast majority of the programs that #AlphaCode generates are wrong (Figure 8). It is the filtering using example tests that allows #AlphaCode to actually solve something. Example tests are part of the input (App. F), yet most sampled programs can't solve them.
2022-02-03 07:57:03Using example tests is a fair game for comp. programming and perhaps for some of real world backend development. But for much of the real-world code (e.g. code that defines front-end behavior) crafting tests is not much easier than coding itself.
2022-02-03 07:57:03The paper emphasizes creative aspects of competitive programming, but from my experience it does involve writing lots of boilerplate code. Many problems involve deployment of standard algorithms: Levenstein-style DP, DFS/BFS graph traversals, max-flow, and so on.
2022-02-03 07:57:03Sec. 6.1 makes a point that #AlphaCode does not exactly copy sequences from training data. That’s a low bar for originality: change a variable name and this is no longer copying. It would be interesting to look at nearest neighbor solutions found using neural representations.
2022-02-03 07:57:04Let me also dilute these critical remarks with a note of appreciation. AlphaCode uses a very cool “clustering” method to marginalize out differently-written but semantically equivalent programs. I think forms of this approach can become a code generation staple.
2022-02-03 07:57:04To sum up: AlphaCode is a great contribution, and AI for coding is a very promising direction with lots of great applications ahead. But this is not AlphaGo in terms of beating humans and not AlphaFold in terms of revolutionizing an entire field of science. We've got work to do.
2022-02-03 07:57:05今度はコーディングする #AlphaCode か。COVIDの間にこんなのを在宅ワークで作っていたチームがいるとは恐るべし。
2022-02-03 08:04:45@DBahdanau @DeepMind I fully agree with you Dima, there is a long way to go. :)
2022-02-03 08:16:40