NT (@Mitubant) profile picture

NT

@Mitubant

CS Ph.D. Data Scientist kaggle competitions expert🥇0 🥈3🥉2 street fighter 6 high master melty blood type lumina ex4

Joined November 2012
  • Tweets659
  • Following158
  • Followers129
Xに登録した日を覚えていますか？ #MyXAnniversary
Image from NT (@Mitubant)
1
75
NT retweeted
Venn (@VennYour) profile picture
Venn@VennYourNov 12
Hope you will be back🙏
1
10
766
omg, jensen huang leaved nvidia...
49
Love this! Amazing breakthrough and it is not an LLM (finally!) 😅🔥 Someone just made k-means 10x faster with no new hyperparameters, shipped in FAISS, works on CPU and GPU. A 70-year-old algorithm, rebuilt with modern optimization!
Swann Bessa (@swann_bessa) profile pictureSwann Bessa
@swann_bessaNov 10
Introducing e-means, an algorithm beating k-means in 10x fewer epochs. With order of magnitude wall-clock speedups, e-means is released in FAISS and needs no hyperparameter tuning beyond k. It is robust to various scales and datasets. Still skeptical ? Just try e-means [1/10]🧵
7
60
1
798
49,270
my team's arc-agi-3 writeup, paper and video are out. 5th of 5,070 on the public leaderboard, fingers crossed for the private one on dec 4 writeup: kaggle.com/competitions/arc-… paper: kaggle.com/competitions/arc-… video: youtube.com/watch?v=JLo5Ryml…
[Public LB 5th] Stuffing the Turducken | Kaggle

[Public LB 5th] Stuffing the Turducken | Kaggle

How we kept stuffing the bird—and learned when to stop.

kaggle.com
2
3
26
1,038
Dots can now control many machines in addition to its own little cloud computer that we provide by default.
OpenAI Developers (@OpenAIDevs) profile pictureOpenAI Developers
@OpenAIDevsNov 9
You can now connect multiple local computers to the same dot and choose which one it uses for a task. It can build and test your app on a dedicated machine while you keep coding on your laptop.
1,582
156
115
5,307
887,869
NT retweeted
Dzy (@j4kcdq) profile picture
Dzy
@j4kcdqNov 9
关于GPT-6.2Astra灰度测试的效果总结 1.极大改善了过度防御的问题，速度也有所提升 2.视觉审美大幅提升，整体效果偏写实，3d效果也进步很大 3.非常非常耐用！单对话max推理强度跑了10个小时，仅消耗pro200的10%左右 4.几乎没有bug和二次返工，理解能力变得极强 6.2Astra的整体性能要比opus 5.5强的多，灰测的大概是阉割版本，不知道完整的能不能打得过fable 5.5 具体测试效果可以看我之前发的帖子
33
3
2
120
42,196
our team's solution at ARC3 [Public LB 5th] Stuffing the Turducken on #kaggle kaggle.com/competitions/arc-…
[Public LB 5th] Stuffing the Turducken | Kaggle

[Public LB 5th] Stuffing the Turducken | Kaggle

How we kept stuffing the bird—and learned when to stop.

kaggle.com
1
23
983
Claude Fable 5.2 from @AnthropicAI on ARC-AGI (Verified): - ARC-AGI-3: 97.7%, $18.6K (provider adapter harness), 81.2%, $21.9K (standard harness) - ARC-AGI-2: 93.8%, $0.98/task - ARC-AGI-1: 98.5%, $0.36/task Fable 5.2 sets a new high score on ARC-AGI-3 in the standard harness.
Image from ARC Prize (@arcprize)
42
91
25
1,235
102,748
寒い日は、こうやって出勤したい。
658
8,720
1,525
81,347
9,831,740
MUSEを触り始めました ↓ Check out your personal AI agent, Muse! Redeem my invite code via "Settings" within 48 hours of signing up, and we’ll each receive 1 billion Muse tokens. Invite code: IL6IZC muse.ai/join
Muse — Your Personal AI Agent

Muse — Your Personal AI Agent

Try Muse, your personal AI agent that gets things done. Give Muse a goal or an everyday task and it handles the rest, from finances and health to shopping and the people you care about.

muse.ai
3
2
293
#5 run DeepSeek V4.1 Flash on 4×L4 for ARC-AGI-2 , what an impressive work! #kaggle kaggle.com/competitions/arc-…
[Public 5th] Fit the Whale In: Running DeepSeek V4.1 Flash on 4×L4 for ARC-AGI-2 | Kaggle

[Public 5th] Fit the Whale In: Running DeepSeek V4.1 Flash on 4×L4 for ARC-AGI-2 | Kaggle

kaggle.com
4
275
新世代のAKI対策
5
21
7
587
74,495
こちらがLinkedIn公式の日本語Postになります。アカウントお持ちの方は是非反応いただけると大変嬉しいです！ linkedin.com/posts/sony_tech…
Two members of Sony Group's "Kaggle Club", Kyota Masuyama and Masanori Miyahara, participated in...

Two members of Sony Group's "Kaggle Club", Kyota Masuyama and Masanori Miyahara, participated in...

Two members of Sony Group's "Kaggle Club", Kyota Masuyama and Masanori Miyahara, participated in "Kaggriculture", a competition with a $50,000 prize pool and the most participants in Kaggle's...

linkedin.com
1
7
2
71
3,365
カードをいただきました😁
Image from NT (@Mitubant)
Image from NT (@Mitubant)
13
423
今日は休日だったのでちょっとこれを確認。NVIDIAもオントロジー。油断するとpostgresにクレデンシャルを平文で書くので注意。RDBだけなのでText to SQLだけ。非構造化データとつなげるようなことはまだ出来ないしオープンソースだけどコントリビューションは出来ないみたい。 github.com/NVIDIA/auto-ontol…
GitHub - NVIDIA/auto-ontology: Auto Ontology is a Python framework for building semantic, graph-b...

GitHub - NVIDIA/auto-ontology: Auto Ontology is a Python framework for building semantic, graph-b...

Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data. It enables deterministic query generation, better context retrieval, and more explainable AI for L...

github.com
1
9
99
5,154
ARC2, ARC3の結果は12/4に出る予定です。
Image from NT (@Mitubant)
1
12
581
Arc2、Arc3、お疲れ様でした！
10
311
elithにeliteが集まっている
3
12
704
KagglerがどんどんElithに吸収されていっている！！
3
5
45
3,857
Load more