Cheng Qian (@qiancheng1231) 's Twitter Profile
Cheng Qian

@qiancheng1231

UIUC PhD @uiuc_nlp advised by @hengjinlp | Prev THU Undergrad @TsinghuaNLP advised by @zibuyu9 | Incoming intern @salesforce | #LLM #Agent

ID: 1610501809151381505

linkhttp://qiancheng0.github.io calendar_today04-01-2023 05:02:23

53 Tweet

393 Followers

460 Following

Hongru Wang (@wangcarrey) 's Twitter Profile Photo

Write a blog to share my recent thoughts about knowledge boundaries & tool use & language agent. This is the first time to propose three laws of knowledge boundaries!๐Ÿ”ฅ candle-walker-56d.notion.site/NAACL-2025-Oraโ€ฆ Chinese Version: mp.weixin.qq.com/s/XzjiLUFAr1Ycโ€ฆ

Emre Can Acikgoz (@emrecanacikgoz) 's Twitter Profile Photo

๐ŸŽ‰ Thrilled to share that two of our Agent papers are accepted to ACL 2025! 1๏ธโƒฃ CoALM: arxiv.org/abs/2502.08820 2๏ธโƒฃ SMART: arxiv.org/abs/2502.11435 Looking forward to presenting these and exchanging ideas on Agents at #ACL2025 ACL 2025! ๐Ÿš€ Additionally, I am also excited to be

๐ŸŽ‰ Thrilled to share that two of our Agent papers are accepted to ACL 2025!

1๏ธโƒฃ CoALM: arxiv.org/abs/2502.08820
2๏ธโƒฃ SMART: arxiv.org/abs/2502.11435

Looking forward to presenting these and exchanging ideas on Agents at #ACL2025 <a href="/aclmeeting/">ACL 2025</a>!

๐Ÿš€ Additionally, I am also excited to be
Cheng Qian (@qiancheng1231) 's Twitter Profile Photo

๐Ÿ“ฃ SMARTAgent is accepted to ACL 2025 Findings! Itโ€™s increasingly important to form an agentโ€™s metacognition, which we believe should guide its action and reasoning. We are continuing on this way!! Position paper will be released soon!

Cheng Qian (@qiancheng1231) 's Twitter Profile Photo

๐Ÿ“ฃ EscapeBench is accepted to ACL 2025 Main! Creativity is what many current agent works neglect, but will be extremely important for agent to reach human level intelligence and be applied to solve real world challenges. Another paper continuing this work is also on the way!

Avi Sil (@aviaviavi__) 's Twitter Profile Photo

While building Agents for Enterprise applications, one thing is very important: not to overuse tool-calling with LLMs - that makes your AI agent very expensive. In our new ACL paper, we show a method to mitigate over-use of tools using a SMART way. Read more from post below ๐Ÿ‘‡

Hongru Wang (@wangcarrey) 's Twitter Profile Photo

Mathematical modeling is a key way for our humans to understand how the world runs. If you truly believe in your agent, you should test them on our new benchmark!

Peixuan Han (้Ÿฉๆฒ›็…Š) (@peixuanhakhan) 's Twitter Profile Photo

(1/5) Want to make your LLM a skilled persuader? Check out our latest paper: "ToMAP: Training Opponent-Aware LLM Persuaders with Theory of Mind"! For details: ๐Ÿ“„Arxiv: arxiv.org/pdf/2505.22961 ๐Ÿ› ๏ธGitHub: github.com/ulab-uiuc/ToMAP

(1/5) Want to make your LLM a skilled persuader?

Check out our latest paper: "ToMAP: Training Opponent-Aware LLM Persuaders with Theory of Mind"!

For details:
๐Ÿ“„Arxiv: arxiv.org/pdf/2505.22961
๐Ÿ› ๏ธGitHub: github.com/ulab-uiuc/ToMAP
Hongru Wang (@wangcarrey) 's Twitter Profile Photo

Whatโ€™s is the agent? What is the optimal behavior to achieve the predefined goal? And how to learn that behavior policy? We formally introduce a systematic Theory of Agent (ToA), analogous to the cognitive framework of Theory of Mind (ToM). Where ToM refers to the ability to

Whatโ€™s is the agent? What is the optimal behavior to achieve the predefined goal? And how to learn that behavior policy?

We formally introduce a systematic Theory of Agent (ToA), analogous to the cognitive framework of Theory of Mind (ToM). Where ToM refers to the ability to
Cheng Qian (@qiancheng1231) 's Twitter Profile Photo

Theory of Agent: From reasoning and tool use, we are defining agent from a knowledge and behavior driven perspective. Welcome to check our newest release!! arxiv.org/pdf/2506.00886

Xiusi Chen (@xiusi_chen) 's Twitter Profile Photo

Can LLMs make rational decisions like human experts? ๐Ÿ“–Introducing DecisionFlow: Advancing Large Language Model as Principled Decision Maker We introduce a novel framework that constructs a semantically grounded decision space to evaluate trade-offs in hard decision-making

Can LLMs make rational decisions like human experts?

๐Ÿ“–Introducing DecisionFlow: Advancing Large Language Model as Principled Decision Maker

We introduce a novel framework that constructs a semantically grounded decision space to evaluate trade-offs in hard decision-making
Rui Yang (@ruiyang70669025) 's Twitter Profile Photo

Excited to share that EmbodiedBench was selected for an Oral at ICML 2025! We recently added results for new models (InternVL3, Gemma3, Ovis2) and released a large agent trajectory dataset on ๐Ÿค—: embodiedbench.github.io Try training and evaluating your MLLM for embodied agents!

Excited to share that EmbodiedBench was selected for an Oral at ICML 2025!

We recently added results for new models (InternVL3, Gemma3, Ovis2) and released a large agent trajectory dataset on ๐Ÿค—: embodiedbench.github.io

Try training and evaluating your MLLM for embodied agents!
Manling Li (@manlingli_) 's Twitter Profile Photo

What is key of agent decision making? Is there a decision making boundary? I am always thinking of the potential boundary of correct decision making and the uncertainty of this boundary. The alignment of decision making boundary and tool-use boundary led by @WangCarrey

Yuji Zhang (@yuji_zhang_nlp) 's Twitter Profile Photo

๐Ÿง Letโ€™s teach LLMs to learn smarter, not harder๐Ÿ’ฅ[arxiv.org/pdf/2506.06972] ๐Ÿค–How can LLMs verify complex scientific information efficiently? ๐Ÿš€We propose modular, reusable atomic reasoning skills that reduce LLMsโ€™ cognitive load to verify scientific claims with little data.

๐Ÿง Letโ€™s teach LLMs to learn smarter, not harder๐Ÿ’ฅ[arxiv.org/pdf/2506.06972]
๐Ÿค–How can LLMs verify complex scientific information efficiently?
๐Ÿš€We propose modular, reusable atomic reasoning skills that reduce LLMsโ€™ cognitive load to verify scientific claims with little data.
May Fung (@may_f1_) 's Twitter Profile Photo

๐Ÿง  How can AI evolve from statically ๐˜ต๐˜ฉ๐˜ช๐˜ฏ๐˜ฌ๐˜ช๐˜ฏ๐˜จ ๐˜ข๐˜ฃ๐˜ฐ๐˜ถ๐˜ต ๐˜ช๐˜ฎ๐˜ข๐˜จ๐˜ฆ๐˜ด โ†’ dynamically ๐˜ต๐˜ฉ๐˜ช๐˜ฏ๐˜ฌ๐˜ช๐˜ฏ๐˜จ ๐˜ธ๐˜ช๐˜ต๐˜ฉ ๐˜ช๐˜ฎ๐˜ข๐˜จ๐˜ฆ๐˜ด as cognitive workspaces, similar to the human mental sketchpad? ๐Ÿ” Whatโ€™s the ๐—ฟ๐—ฒ๐˜€๐—ฒ๐—ฎ๐—ฟ๐—ฐ๐—ต ๐—ฟ๐—ผ๐—ฎ๐—ฑ๐—บ๐—ฎ๐—ฝ from tool-use โ†’ programmatic

๐Ÿง  How can AI evolve from statically ๐˜ต๐˜ฉ๐˜ช๐˜ฏ๐˜ฌ๐˜ช๐˜ฏ๐˜จ ๐˜ข๐˜ฃ๐˜ฐ๐˜ถ๐˜ต ๐˜ช๐˜ฎ๐˜ข๐˜จ๐˜ฆ๐˜ด โ†’ dynamically ๐˜ต๐˜ฉ๐˜ช๐˜ฏ๐˜ฌ๐˜ช๐˜ฏ๐˜จ ๐˜ธ๐˜ช๐˜ต๐˜ฉ ๐˜ช๐˜ฎ๐˜ข๐˜จ๐˜ฆ๐˜ด as cognitive workspaces, similar to the human mental sketchpad?
๐Ÿ” Whatโ€™s the ๐—ฟ๐—ฒ๐˜€๐—ฒ๐—ฎ๐—ฟ๐—ฐ๐—ต ๐—ฟ๐—ผ๐—ฎ๐—ฑ๐—บ๐—ฎ๐—ฝ from tool-use โ†’ programmatic