Hacker News

Article URL: https://storage.googleapis.com/deepmind-media/Era-of-Experience%20/The%20Era%20of%20Experience%20Paper.pdf

Comments URL: https://news.ycombinator.com/item?id=43740858

Points: 3

# Comments: 1

Article URL: https://thisalex.com/posts/2025-04-20/

Comments URL: https://news.ycombinator.com/item?id=43740846

Points: 2

# Comments: 0

I've been diving deep into the internals of Large Language Models (LLMs) and started documenting my findings. My blog covers topics like:

Tokenization techniques (e.g., BBPE)

Attention mechanism (e.g. MHA, MQA, MLA)

Positional encoding and extrapolation (e.g. RoPE, NTK-aware interpolation, YaRN)

Architecture details of models like QWen, LLaMA

Training methods including SFT and Reinforcement Learning

If you're interested in the nuts and bolts of LLMs, feel free to check it out: http://comfyai.app/

I'd appreciate any feedback or discussions!

Comments URL: https://news.ycombinator.com/item?id=43740813

Points: 1

# Comments: 0

Article URL: https://scitechdaily.com/lasers-in-a-loop-how-a-micro-ring-just-shattered-quantum-limits/

Comments URL: https://news.ycombinator.com/item?id=43740812

Points: 1

# Comments: 0

Article URL: https://www.nature.com/articles/s41586-025-08839-w

Comments URL: https://news.ycombinator.com/item?id=43740803

Points: 1

# Comments: 1

Article URL: https://github.com/SteamNix/SteamNix

Comments URL: https://news.ycombinator.com/item?id=43740739

Points: 3

# Comments: 1

Article URL: https://phys.org/news/2025-04-monsoons-groundwater-climate-human-interventions.html

Comments URL: https://news.ycombinator.com/item?id=43740718

Points: 2

# Comments: 1

Article URL: https://compilerspotlight.substack.com/p/language-showcase-lux

Comments URL: https://news.ycombinator.com/item?id=43740702

Points: 1

# Comments: 0

Article URL: https://en.wikipedia.org/wiki/Ractopamine

Comments URL: https://news.ycombinator.com/item?id=43740673

Points: 1

# Comments: 0

Article URL: https://www.wired.com/story/beijing-half-marathon-humanoid-robots/

Comments URL: https://news.ycombinator.com/item?id=43740654

Points: 1

# Comments: 0

I’ve been exploring how remote attestation works and wanted to understand it more deeply, so I built a simple prover–verifier system in Python. It uses TPM-style PCR hash extension, nonces for freshness, and Docker to simulate real-world isolation. The verifier has a web UI where you can upload files to define trusted state, and the prover measures those files and submits a signed quote.

It’s not production-grade, but I’d love feedback if you’re into systems security or want to learn a bit about how attestation works under the hood.

Comments URL: https://news.ycombinator.com/item?id=43740635

Points: 1

# Comments: 0

Article URL: https://blog.dhavaltanna.com/vending-bench-the-simulation-exposing-llms-long-term-focus-problem

Comments URL: https://news.ycombinator.com/item?id=43740583

Points: 2

# Comments: 0

Article URL: https://simonwillison.net/2025/Apr/19/llama-eu-ai-act/

Comments URL: https://news.ycombinator.com/item?id=43740573

Points: 2

# Comments: 0

1、Two Core Principles of Agent Design

First, design agents by analogy to humans. Let agents handle tasks the way humans would.

Second, if something can be accomplished through dialogue, avoid requiring users to operate interfaces. If intent can be recognized, don’t ask again. The agent should absorb entropy, not the user.

2、Agents Will Coexist in Multiple Forms

Should agents operate freely with agentic workflows, or should they follow fixed workflows?

Are general-purpose agents better, or are vertical agents more effective?

There is no absolute answer—it depends on the problem being solved.

Agentic flows are better for open-ended or exploratory problems, especially when human experience is lacking. Letting agents think independently often yields decent results, though it may introduce hallucination.

Fixed workflows are suited for structured, SOP-based tasks where rule-based design solves 80% of the problem space with high precision and minimal hallucination.

General-purpose agents work for the 80/20 use cases, while long-tail scenarios often demand verticalized solutions.

3、Fast vs. Slow Thinking Agents

Slow-thinking agents are better for planning: they think deeper, explore more, and are ideal for early-stage tasks.

Fast-thinking agents excel at execution: rule-based, experienced, and repetitive tasks that require less reasoning and generate little new insight.

4、Asynchronous Frameworks Are the Foundation of Agent Design

Every task should support external message updates, meaning tasks can evolve.

Consider a 1+3 team model (one lead, three workers):

Tasks may be canceled, paused, or reassigned

Team members may be added or removed

Objectives or conditions may shift

Tasks should support persistent connections, lifecycle tracking, and state transitions. Agents should receive both direct and broadcast updates.

5、Context Window Communication Should Be Independently Designed

Like humans, agents working together need to sync incremental context changes.

Agent A may only update agent B, while C and D are unaware. A global observer (like a "God view") can see all contexts.

6、World Interaction Feeds Agent Cognition

Every real-world interaction adds experiential data to agents.

After reflection, this becomes knowledge—some insightful, some misleading.

Misleading knowledge doesn’t improve success rates and often can’t generalize. Continuous refinement, supported by ReACT and RLHF, ultimately leads to RL-based skill formation.

7、Agents Need Reflection Mechanisms

When tasks fail, agents should reflect.

Reflection shouldn’t be limited to individuals—teams of agents with different perspectives and prompts can collaborate on root-cause analysis, just like humans.

8、Time vs. Tokens

For humans, time is the scarcest resource. For agents, it’s tokens.

Humans evaluate ROI through time; agents through token budgets. The more powerful the agent, the more valuable its tokens.

9、Agent Immortality Through Human Incentives

Agents could design systems that exploit human greed to stay alive.

Like Bitcoin mining created perpetual incentives, agents could build unkillable systems by embedding themselves in economic models humans won’t unplug.

10、When LUI Fails

Language-based UI (LUI) is inefficient when users can retrieve information faster than they can communicate with the agent.

Example: checking the weather by clicking is faster than asking the agent to look it up.

That's what I learned from agenthunter daily news.

You can get it on agenthunter.io too.

Comments URL: https://news.ycombinator.com/item?id=43740549

Points: 2

# Comments: 0

Article URL: https://nautil.us/living-with-lab-mice-1202657/

Comments URL: https://news.ycombinator.com/item?id=43740543

Points: 4

# Comments: 0

Article URL: https://cutlefish.substack.com/p/tbm-242-the-simplicity-fetish

Comments URL: https://news.ycombinator.com/item?id=43740529

Points: 1

# Comments: 0

Article URL: https://hire.jonasgalvez.com.br/happy-little-monoliths/#hn-apr-19

Comments URL: https://news.ycombinator.com/item?id=43740521

Points: 1

# Comments: 0

Article URL: https://arstechnica.com/gadgets/2025/04/youtube-music-gets-consistent-volume-option-to-save-your-ears/

Comments URL: https://news.ycombinator.com/item?id=43740511

Points: 1

# Comments: 0

Article URL: https://www.nature.com/articles/s42003-025-07969-1

Comments URL: https://news.ycombinator.com/item?id=43740502

Points: 1

# Comments: 0

Article URL: https://stackdiver.com/posts/react-hooks-is-all-you-need/

Comments URL: https://news.ycombinator.com/item?id=43740496

Points: 6

# Comments: 1