Hacker News: stan_kirdey

DeepSeek-v3.2

stan_kirdey — Mon, 01 Dec 2025 14:40:26 +0000

Article URL: https://cas-bridge.xethub.hf.co/xet-bridge-us/692cfec93b25b81d09307b94/2d0aa38511b9df084d12a00fe04a96595496af772cb766c516c4e6aee1e21246?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Content-Sha256=UNSIGNED-PAYLOAD&X-Amz-Credential=cas%2F20251201%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20251201T142652Z&X-Amz-Expires=3600&X-Amz-Signature=434ce0f081e5725cb34243380f4c51ebbbc74e1bcc349760c5c702166a2a3bb0&X-Amz-SignedHeaders=host&X-Xet-Cas-Uid=public&response-content-disposition=inline%3B+filename*%3DUTF-8%27%27paper.pdf%3B+filename%3D%22paper.pdf%22%3B&response-content-type=application%2Fpdf&x-id=GetObject&Expires=1764602812&Policy=eyJTdGF0ZW1lbnQiOlt7IkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTc2NDYwMjgxMn19LCJSZXNvdXJjZSI6Imh0dHBzOi8vY2FzLWJyaWRnZS54ZXRodWIuaGYuY28veGV0LWJyaWRnZS11cy82OTJjZmVjOTNiMjViODFkMDkzMDdiOTQvMmQwYWEzODUxMWI5ZGYwODRkMTJhMDBmZTA0YTk2NTk1NDk2YWY3NzJjYjc2NmM1MTZjNGU2YWVlMWUyMTI0NioifV19&Signature=Nq%7EnPts2dmKo9bxaPjzSf4G%7EDXbw7T4B6t69nIYJm4kRRxguQ3CBBPvjRGxhtfu4m2rbAp-kF8ycR755fuiM2rSPh0c287A3r5C5YJ0gq2MpoZJGerzmFGB74uBP6Y1-AE9nkrlbv-uoLgFCoEsc2TBagnD88QD3JkqBjjbrmOslKvKtIDzJvQecloQ--CwMQFoDk9uKAjfKhH2vDJVtiqI-jRJUrZyLZ8y6aqSmtWCWwPiEZ-dfMWW-gkZBrcxryY7XZGsoQ2n55mYSQT2xcgexJpjeIuwg9VOdISE6sHWyTa2txWRYbx0xcWNPg-FBsbrxXsCFmlQnXRX7l2GyuA__&Key-Pair-Id=K2L8F4GPSG1IFC

Comments URL: https://news.ycombinator.com/item?id=46107969

Points: 3

# Comments: 0

New comment by stan_kirdey in "Fara-7B: An efficient agentic model for computer use"

stan_kirdey — Wed, 26 Nov 2025 22:59:10 +0000

* fine tuned Qwen-7B

New comment by stan_kirdey in "Sora 2"

stan_kirdey — Tue, 30 Sep 2025 18:25:42 +0000

That could totally power next generation of green-screen techs. Generative actors may not find favorable response in the audiences; but SFX, decor, extras, environments that react to actors' actions - amazing potential.

New comment by stan_kirdey in "Enough AI copilots, we need AI HUDs"

stan_kirdey — Mon, 28 Jul 2025 02:17:30 +0000

needed to find a kids’ orthodontist. made a tiny voice agent: feed it numbers, it calls, asks about price/availability/insurance, logs the gist.

it kind of worked. the magic was the smallest UI around it:

- timeline of dials + retries

- "call me back" flags

- when it tried, who picked up

- short summaries with links to the raw transcript

once i could see the behavior, it stopped feeling spooky and started feeling useful.

so yeah, copilots are cool, but i want HUDs: quiet most of the time, glanceable, easy to interrupt, receipts for every action.

New comment by stan_kirdey in "Performance and telemetry analysis of Trae IDE, ByteDance's VSCode fork"

stan_kirdey — Sun, 27 Jul 2025 22:58:28 +0000

props to OP for the screenshots and payloads—that’s how you do it. If any IDE wants trust, they know the recipe - make telemetry optin by default and provide a real kill switch.

New comment by stan_kirdey in "Windsurf employee #2: I was given a payout of only 1% what my shares where worth"

stan_kirdey — Fri, 25 Jul 2025 19:54:08 +0000

Engineers: always negotiate for higher base salaries. In the vast majority of cases—especially during acquihires—your equity will be worth little or nothing. Founders and VCs still get paid; employees rarely do.

Don't just accept promises. Ask for the 409A valuation, liquidation preferences, and pay bands. If a company won’t provide transparency, that’s your signal.

Equity is a lottery ticket. Salary is money in the bank.

New comment by stan_kirdey in "Cozy video games can quell stress and anxiety"

stan_kirdey — Sat, 19 Apr 2025 08:26:27 +0000

exactly for this made vortle.com no login, no internet needed, no data collected, no score - just you and the game to de-stress and calm down

Show HN: Jslon – high performance GUI viewer of large JSONL, TSV and CSV files

stan_kirdey — Fri, 04 Apr 2025 16:28:53 +0000

Article URL: https://skirdey-inflection.github.io/jslon/

Comments URL: https://news.ycombinator.com/item?id=43584720

Points: 10

# Comments: 0

New comment by stan_kirdey in "GPT-4.5"

stan_kirdey — Fri, 28 Feb 2025 04:37:07 +0000

For most tasks, GPT-4o/o3-mini are already great, and cheaper.

What is the real-world use case where GPT-4.5? Anyone actually seeing a game-changing difference? Please share your prompts!

New comment by stan_kirdey in "Show HN: Transform your codebase into a single Markdown doc for feeding into AI"

stan_kirdey — Fri, 14 Feb 2025 18:44:24 +0000

Nice! Built something similar in Rust that supports local and remote repos: https://crates.io/crates/r2md

Show HN: R2md – quickly export entire codebase to a single Markdown/pdf

stan_kirdey — Thu, 30 Jan 2025 19:28:35 +0000

super simple tool, what sets it apart is straight forward cli interface without too much noise

Comments URL: https://news.ycombinator.com/item?id=42881243

Points: 1

# Comments: 0

New comment by stan_kirdey in "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via RL"

stan_kirdey — Wed, 29 Jan 2025 03:01:40 +0000

made this super easy to use tool https://github.com/skirdey-inflection/r2md

New comment by stan_kirdey in "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via RL"

stan_kirdey — Sun, 26 Jan 2025 09:29:29 +0000

I've been comparing R1 to O1 and O1-pro, mostly in coding, refactoring and understanding of open source code.

I can say that R1 is on par with O1. But not as deep and capable as O1-pro. R1 is also a lot more useful than Sonnete. I actually haven't used Sonnete in awhile.

R1 is also comparable to the Gemini Flash Thinking 2.0 model, but in coding I feel like R1 gives me code that works without too much tweaking.

I often give entire open-source project's codebase (or big part of code) to all of them and ask the same question - like add a plugin, or fix xyz, etc. O1-pro is still a clear and expensive winner. But if I were to choose the second best, I would say R1.

New comment by stan_kirdey in "Willow, Our Quantum Chip"

stan_kirdey — Mon, 09 Dec 2024 21:38:36 +0000

Is anyone fine-tuning llama to write Q#? I feel LLMs can be a helpful tool in learning how to code quantum systems.

Show HN: LLM driven OS to execute network security exploration agents

stan_kirdey — Sun, 10 Nov 2024 02:02:52 +0000

I made this thing where LLM works like a tiny OS kernel - it plans tasks and runs different agents. Right now it's doing security stuff (poking at networks, checking SSL, testing APIs) with WebGoat as a playground.

The ultimate goal is to ask it "find vulnerabilities in XYZ" and see it orchestrate its own tools, backtrack, try different approaches, etc.

The cool part is the LLM doesn't just blast out a bunch of tokens - it plans, reasons and executes tasks step by step during inference.

Currently playing with security agents but you could plug in anything really. Would love getting you to play with it and maybe add an agent or two.

Comments URL: https://news.ycombinator.com/item?id=42098120

Points: 2

# Comments: 0

Pi on Intel's Gaudi3 for Enterprises – Inflection AI and Intel Partner Up

stan_kirdey — Mon, 07 Oct 2024 17:17:51 +0000

Article URL: https://www.intel.com/content/www/us/en/newsroom/news/inflection-ai-intel-launch-enterprise-ai-system.html

Comments URL: https://news.ycombinator.com/item?id=41768342

Points: 1

# Comments: 0

New comment by stan_kirdey in "Two new Gemini models, reduced 1.5 Pro pricing, increased rate limits, and more"

stan_kirdey — Tue, 24 Sep 2024 17:44:22 +0000

Google should just offer llama3 405b, maybe slightly fine tuned. Geminis are unusable.

New comment by stan_kirdey in "NASA announces Boeing Starliner crew will return on SpaceX Crew-9"

stan_kirdey — Sat, 24 Aug 2024 17:53:47 +0000

does the crew get paid overtime if they were to stay until Feb?

New comment by stan_kirdey in "Gemini Flash"

stan_kirdey — Wed, 15 May 2024 02:53:38 +0000

I've been diligently trying to use Gemini 1.5 Pro, and it is not even on the level of Llama3-70B. I really hope Gemini improves, even if it gets reduced context length.

New comment by stan_kirdey in "Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced"

stan_kirdey — Thu, 08 Feb 2024 20:08:39 +0000

I've been testing Ultra model today and comparing it to open source Mixtral MOE on HF chat. Gemini Ultra lost in every instance to the free open source model, including code generation. I think the tasks I do with the help of LLMs are common, and Gemini Ultra is unusable at the moment.

Gemini refuses to answer or perform even on the simple prompts.

I hope Google team can make it better, but at the moment, for my light coding and text analysis use cases it is not worth $19.99