technoabsurdist

hi I work at wafer. yes we ran benchmarks. for example our Kimi K3 is live on open router and in order to host there you have to run accuracy checks. like tau/gpqa.

and then u must pass test regarding thinking, coherency, and tool calling

sounds good feedback taken, thanks beffjezos

hi yes it’s not optimized for single stream it’s optimized for total node throughput

this is exactly our thesis at wafer :) thank you for the support

AMD MI355X uses 1,400W per GPU and NVIDIA B200 uses 1,200W. So AMD uses about 16% more power.

yes it is 213 tok/s single stream (so per user)

hi i work at wafer. no the margins are lower averaging at about ~40%. utilization is one of the highest order bits in determining margins here, yes.

^ We currently just have llama3.1-8b, so we'll be working on adding more models across more hardware options!

We just launched Chip Benchmark, an open-source tool for hardware-centric benchmarking of open-weight LLMs across accelerators like NVIDIA A100/H100/L40S and AMD MI300X. It measures throughput, latency, and time-to-first-token with transparent scripts and an interactive web dashboard—making apples-to-apples comparisons easier.

We're actively welcoming contributions, new hardware support, and benchmark requests.

Repo here: https://github.com/Herdora/chip-benchmark Dashboard: https://herdora.com/benchmark

Feedback and contributions welcome! We made it super easy to add other architectures by including the script we used for benchmarking.

We've been doing lots of GPU kernel profiling and optimization on cloud infrastructure, but without local GPU hardware, that meant constant SSH juggling: upload code, compile remotely, profile kernels, download results, repeat. Or, work entirely on cloud which is expensive, slow, and annoying. We were spending more time managing infrastructure than writing the kernels we wanted to optimize.

So we built Chisel: one command to run profiling commands on any kernel. Zero local GPU hardware required.

Next up we're planning to build a web dashboard for visualizing results, simultaneous profiling across multiple GPU types, and automatic resource cleanup. But please let us know what you would like to see in this project.

Available via PyPI: pip install chisel-cli

Github: https://github.com/Herdora/chisel

We're actively developing and would love community feedback. Feature requests and contributions always welcome!

oh yeah, in my experience anything below ROCm6.x really sucks.

I tried to run qwen2.5-32B on ROCm5.x and it was running at <15tok/s lol.

Have you tried running any sort of LLM inference on your MI25, or what NN workloads are you running?

correct github link: https://github.com/Herdora/chisel

We built Crackd to solve a fundamental problem: helping students understand where they stand technically, and helping them learn how to become the absolute best.

Technical students have no reliable way to know how good they are. Grades don't work because they vary by professor and university. Getting internships can boil down to arbitrary indicators, and online advice is filled with contradictions.

Athletes have it better. A college football player knows exactly where they stand nationally (e.g. https://www.espn.com/college-sports/football/recruiting/play...).

More importantly, they know what the best players do differently: their training methods, their practice routines, the specific skills they're mastering.

CS students are flying blind by comparison. They face a collapsing job market with nothing but twitter threads giving them contradictory advice.

We built Crackd to fix this. Students judge pairs of technical students, and choose the person they believe is more “cracked.” After enough comparisons, the Elo algorithm generates a leaderboard. Humans are terrible at absolute judgments but good at comparisons. You can't reliably say if someone is a 7/10 developer, but you can usually tell who has a better profile based on technical achievement.

What matters about Crackd is letting other students see precisely what the top students do differently. Their projects, internships, skills, school clubs, etc. This is the information that turns rankings from a leaderboard into a roadmap for ambitious students.

For the time being, Crackd is restricted to college students as we refine our platform and build our initial community.

We're attempting to cover as many dimensions of technical skill as possible, covering projects, experiences, and including a miscellaneous section so students can add whatever info they feel showcases their best work. But we'd love your feedback on what other dimensions we should include to create the most comprehensive assessment of technical capability. We would love to hear from you at contact@crackd.io

I'm an undergrad at a T10 college. Walking through our library, I often notice about 30% of students have ChatGPT or Claude open on their screens.

In my circle, I can't name a single person who doesn't heavily use these tools for assignments.

What's fascinating, though, is that the most cracked CS students I know deliberately avoid using these tools for programming work. They understand the value in the struggle of solving technical problems themselves. Another interesting effect: many of these same students admit they now have more time for programming and learning they “care about” because they've automated their humanities, social sciences, and other major requirements using LLMs. They don't care enough about those non-major courses to worry about the learning they're sacrificing.

Congrats on the launch! Big fan of Continue <3

Hey HN! My friend and I are building kojo, the duolingo for coding. It consists of 100% personalized llm-generated curriculum and learning path. The app will be out in the app store in the next couple of weeks. Sign up for waitlist now to get notified when it's out!

can someone explain how this manages to list all 2^122 uuids?

at my last job, I spent an unnecessary amount of time crafting good commit messages. so i created git-autocommit, a minimal CLI tool that generates a git diff, adds and commits changes with a short, to-the-point commit message

Download: `pip install git-autocommit==0.1.0`

Note: Your computer should have at least 16 GB RAM to run the required llama3-8B model

Nice, glad it worked for you.

Splitting paragraphs is a great idea, thank you!

Let me know anything else you would want; I'd love to implement it.

Youtube Video Transcriptions

Submit a link to generate a PDF with AI-generated summary and a full transcription.

Please consider contributing: https://github.com/technoabsurdist/transcript.ai

Thanks Sieve for the credits. Check them out here: https://www.sievedata.com/

Podcast apps have terrible search. Internet searches for good podcasts are flooded with click bait and promoted content. Instead, the approach here is to engage with podcasts by using natural language.

Please let me know what you think.

Github Repo: https://github.com/technoabsurdist/find-good-podcasts

Warning: It's day 1 of this project, so it's going to be buggy

awesome, thank you so much.

Do you guys still offer deferred admissions for juniors who wish to join a batch after graduation?

Vercel page not working yet, just localhost.

I developed this Chrome extension to set opening links in new tabs as the default behavior. Often, I found myself frustrated when intending to save a tab for later, only to have it replace my current one. While aware of the CMD+Click shortcut for opening links in new tabs, I prefer this as the default behavior, which is the reason I created this extension.

Foundational blocks to interact with Youtube API and LLMs to transcribe, summarize, and analyze Youtube videos. The API is designed to be minimal and customizable, allowing for easy integration with other services and APIs.

Foundational blocks to interact with Youtube API and LLMs to transcribe, summarize, and analyze Youtube videos. The API is designed to be extendable and customizable, allowing for easy integration with other services and APIs. Built using Flask, OpenAI's GPT-3.5-turbo, Youtube third-party libraries, and scraping techniques.

Hey hey!

I've been developing this tool for generating AI-powered YB video transcriptions. It currently works very well for videos under 10 minutes, so I'm working on extending it for longer ones.

I'd love some contributions if you're interested in helping out: https://github.com/technoabsurdist/transcript.ai

The biggest current hurdle is that it breaks on longer videos (typically around 10+ minutes).

I'd love some feedback, and even more some contributions.