CS

Rong "Mantle" Bao

CS undergraduate | @vidar-team binary enthusiast | writer at @D-3CTF

Ningbo, Zhejiang, China

Joined October 2017

66 Followers

18 Following

AI Insights & Achievements

53/ 100

60 new projects with 17 stars.

31 day streak.

Coding for 9 years.

Profile README

Show-Markdown -Path CSharperMantle\README.md

Profile

I am Rong "Mantle" Bao, a third-year Computer Science student at Hangzhou Dianzi University. Feel free to reach me in 中文/English.

He/him
Blog: https://csmantle.top
Natural languages:
- zh-cmn-Hans-CN (native)
- zh-wuu-Hans-CN (native)
- en (near native)
Developing personal projects

Polyglot, enjoy C/Rust/TypeScript most
Reverse engineering, messing around with OS and ISA. Member of Vidar-Team
Bare-metal, desktop native, single-page WebApps development
Interested in x86-64 assembly, RISC-V/LoongArch ISAs and microarch implementation
Mozilla Level 1 Commit Access, active maintainer of SpiderMonkey’s RV64 JIT backend

My attitudes towards AI

Machine learning problems are analytical and statistical problems.
Neural networks are good at discovering patterns not obvious to traditional approaches.
LLMs are probabilistic token predictors. They often intrinsically hallucinate.
Garbage in, garbage out. Observation in, imitation out.
I do use LLMs to brainstorm ideas.
I do use LLMs to perform repetitive clerical work and typographical proofreading.
I do chat with LLMs to have fun in my spare time.
I do use generative models to craft images for hobby use.
I do not use LLMs to write code for production use.
I do not use LLMs to write casual articles, like blogs.
I do not use LLMs to write academic articles or carry out research in place of myself.
I do not use LLMs as a search engine.
I do not trust LLMs as a sole source of information.
I do not accept responses from LLMs without verification.
I believe the loopholes and waivers in LLM hosting companies' EULA put much personal information at risk.
I believe LLMs are energy hogs that should be used with conservation in mind.
I believe LLMs are becoming more performant as theories and technologies evolve.