Hacker News

Subscribe to Hacker News feed
Hacker News RSS
Updated: 54 min 11 sec ago

Bidirectional Enhancement of Graph and Agent

Mon, 05/05/2025 - 2:26am

Chat2Graph is a graph-native agentic system designed to bridge the power of graph databases with advanced AI capabilities. It establishes a multi-agent system (MAS) directly upon a graph database, and harnesses the inherent strengths of graph data structures, such as relationship modeling and interpretability, to enhance core AI agent capabilities like reasoning, planning, memory, and tool utilization.

In Chat2Graph, we expect to deeply implement the concept of "Graph + AI" and deeply integrate graph computing and AI technology into the design and implementation of agents.

GitHub: https://github.com/TuGraph-family/chat2graph

Comments URL: https://news.ycombinator.com/item?id=43892425

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Google Review Link and QR Code Generator for Business Profiles

Mon, 05/05/2025 - 2:25am

A simple chrome extension to create Google review links and QR codes, with built in link clicks analytics.

Comments URL: https://news.ycombinator.com/item?id=43892420

Points: 1

# Comments: 0

Categories: Hacker News

AI Slop 2027

Mon, 05/05/2025 - 2:20am
Categories: Hacker News

Show HN: Raman-01 – A Pocket Physics Solver LLM

Mon, 05/05/2025 - 2:13am

I built a tiny physics solver LLM that performs surprisingly well on easy-to-medium difficulty physics problems. Most LLMs today still struggle with physics QA (as PhyBench recently highlighted), so I wanted to see how far I could push a small model with careful data and minimal compute.

Model: Qwen3-1.7B

Supervised Finetuning: ~1500 curated examples spanning kinematics, EM, acoustics, and more

RL Fine-tuning: GRPO, 1-shot RLVR style (single example, 70 steps)

Total cost: ~$5 on H100

It started with a cold-start SFT (~3 epochs, loss to 0.3), then I ran RL with accuracy reward that climbed from 0.1 → 0.8.

Goal: Create a lightweight physics solver that’s small enough to deploy anywhere—think of it as a "pocket tutor" for foundational physics.

Still working on evaluations—most benchmarks focus on very hard problems, while I want something that evaluates basic correctness, reasoning, and unit sense on easy/medium problems. If anyone has suggestions, I’d love to hear them.

Comments URL: https://news.ycombinator.com/item?id=43892365

Points: 1

# Comments: 0

Categories: Hacker News

LiveScript

Mon, 05/05/2025 - 2:04am

Article URL: https://livescript.net

Comments URL: https://news.ycombinator.com/item?id=43892316

Points: 1

# Comments: 1

Categories: Hacker News

Llama-Nemotron: Efficient Reasoning Models

Mon, 05/05/2025 - 1:44am

Article URL: https://arxiv.org/abs/2505.00949

Comments URL: https://news.ycombinator.com/item?id=43892224

Points: 1

# Comments: 0

Categories: Hacker News

Ask HN: Use C with Classes for Embedded?

Mon, 05/05/2025 - 1:43am

Hello. I am writing code for embedded systems. Most cpu are ARM or SoftCpu from FPGA vendors. Up to now I use C11. All the CPU are supported by gcc, thank god. I used to write c++ before C++ 11. The good old time. I now think more and more often to switch to C++ on embedded. No exceptions no RTTI no STL or Boost. Just C with Classes for the luxury of C++. What do you think? Is it worth the hazel?

Comments URL: https://news.ycombinator.com/item?id=43892213

Points: 1

# Comments: 2

Categories: Hacker News

Fighting LLMs using LLMs

Mon, 05/05/2025 - 1:40am
Categories: Hacker News

Kent State Shootings

Mon, 05/05/2025 - 1:25am
Categories: Hacker News

Pages