Yiding Liu

(former) snowboarding enthusiast 🏂🏻(seriously injured)

research engineer @ tiktok 🇸🇬 vision language models
past: @ baidu search 🇨🇳

[Yiding Liu @Yiding_tanh]

@AixinSG pls enlight us, prof Sun

[Yiding Liu @Yiding_tanh]

still don't understand why people are so hype about jev, can someone enlight me?

[Yiding Liu @Yiding_tanh]

@agihippo Really interesting sharing. Coming from China, it’s hard to imagine someone saying, “I’m not quite sure how, but I somehow ended up at Tsinghua or Peking University.” 🤣🤣🤣

[Yiding Liu @Yiding_tanh]

@agihippo beautiful. every time i go there its always cloudy

[Yiding Liu @Yiding_tanh]

@ltzheng01 what are you doing

[Yiding Liu @Yiding_tanh]

@agihippo haha that's my point. it gives you very promising numbers and good taste. It seems a bit too good to be true

[Junyang Lin @JustinLin610]

https://t.co/mgVhhZVTKg

[Yiding Liu @Yiding_tanh]

@zhang_yuanhan what about you

[Yiding Liu @Yiding_tanh]

@Brian_Bo_Li you are the future goat

[Brian Li @Brian_Bo_Li]

We are building real intelligence into the real world.

Amigogogogo.

[Yiding Liu @Yiding_tanh]

@agihippo This probably generalizes to any form of creation.

[Juri Strumpflohner @juristr]

POV: Senior Agentic Engineer

[Yiding Liu @Yiding_tanh]

🤔🤔

[Naval @naval]

Vibe coding is the new product management.

Training and tuning models is the new coding.

[Yiding Liu @Yiding_tanh]

@BoLi68567011 try marina barrage next time, super chill place

[Yiding Liu @Yiding_tanh]

@agihippo I train at gym with 150+ sgd/session. It’s a totally different than working out by myself. These singaporean little brothers would always push you so hard got me feeling like saitama broke his limiter. It feels great.

[Yiding Liu @Yiding_tanh]

@agihippo what's in your head at this moment

[Yiding Liu @Yiding_tanh]

@serenaa_ge spotify + vscode is lliterally me

[Zhaopeng Tu @tuzhaopeng]

Can AI agents autonomously explore, synthesize, and discover knowledge like researchers? 🤖🔬

Introducing a comprehensive survey on Deep Research (DR) systems, where LLMs evolve from passive text generators into autonomous agents capable of long-horizon reasoning and verifiable knowledge creation.

🗺️ Three-phase roadmap:
1⃣ Agentic Search → Precise evidence acquisition
2⃣ Integrated Research → Multi-source synthesis & reporting
3⃣ Full-stack AI Scientist → Hypothesis generation & discovery

🔧 Four foundational components:
1⃣ Query Planning: Decompose complex questions (parallel, sequential, tree-based).
2⃣ Information Acquisition: Dynamically retrieve from web search, APIs, & multimodal sources.
3⃣ Memory Management: Store, update, and prune context over long horizons.
4⃣ Answer Generation: Synthesize verifiable, cited reports.

🚀 Three optimization paradigms:
1⃣ Workflow Prompting
2⃣ Supervised Fine-Tuning (SFT)
3⃣ End-to-End Agentic Reinforcement Learning (RL)

📊 Key Insight: DR is not just advanced RAG.
Unlike standard RAG, DR enables:
✅ Flexible interaction & tool use beyond static retrieval
✅ Long-horizon planning with autonomous workflows
✅ Reliable, verifiable, and structured outputs

📈 As the field evolves, we are committed to continuously updating this survey to reflect the latest progress!

🧑���💻 Project: https://t.co/B7Ssod4kaV
📃 Paper: https://t.co/qnoxyOklki

[Yiding Liu @Yiding_tanh]

@ZhenghaiX

[Yujia Qin @TsingYoga]

Let's play Genshin step by step