Hacker News: verdverm

New comment by verdverm in "The Pulling of Mythos Offline: Why AI KYC Will Fail to Stop Cybercriminals"

verdverm — Sat, 13 Jun 2026 15:44:22 +0000

I'm curious how this will work for (i.e.) company with foreigners who develop agents that run autonomously. Can the prod deployment use Fable?

New comment by verdverm in "RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8"

verdverm — Sat, 13 Jun 2026 15:18:22 +0000

I've been using https://spark-arena.com/leaderboard to glean this kind of information for DGX Spark, a sort of recipe book. The Nvidia forum has people talking about the things you wish to know. I see some on Discord/Reddit/et al, but less cohesive

I've switched from using the spark as a way to run one model as best it can to running several support models for the md kb I'm working on

New comment by verdverm in "U.S. Military Helping Move 7M Bpd Out of Persian Gulf, Wright Says"

verdverm — Sat, 13 Jun 2026 15:05:49 +0000

If you believe the administration... Pentagon and oss observations say otherwise

New comment by verdverm in "Google proposes Open Knowledge Format based on Markdown"

verdverm — Sat, 13 Jun 2026 14:45:55 +0000

A bunch of broken (slopped?) links in the original post, here is the repo

https://github.com/GoogleCloudPlatform/knowledge-catalog

Spec: https://github.com/GoogleCloudPlatform/knowledge-catalog/blo...

New comment by verdverm in "Statement on US government directive to suspend access to Fable 5 and Mythos 5"

verdverm — Sat, 13 Jun 2026 14:40:40 +0000

Boris Cherny has said it many times, you can search YouTube for "coding is solved" to find examples

Or watch Primagen's "I think they are lying to you" with clips in it

New comment by verdverm in "Open source AI must win"

verdverm — Sat, 13 Jun 2026 05:39:36 +0000

Labs can and do open source more than the weights

https://allenai.org/olmo

New comment by verdverm in "Open source AI must win"

verdverm — Sat, 13 Jun 2026 04:37:02 +0000

DGX Spark runs this sized model (I personally like qwen36moe better than gemma4moe) at speeds fast enough for interactive coding sessions. Algorithmic advances like DiffusionGemma ~4x token gen speeds (https://deepmind.google/models/gemma/diffusiongemma/)

New comment by verdverm in "Open source AI must win"

verdverm — Sat, 13 Jun 2026 04:33:39 +0000

case in point: https://spark-arena.com/leaderboard

New comment by verdverm in "Statement on US government directive to suspend access to Fable 5 and Mythos 5"

verdverm — Sat, 13 Jun 2026 04:18:26 +0000

They've also been saying coding is solved while having text flicker in a terminal

New comment by verdverm in "US Government directive to suspend access to Fable 5 and Mythos 5"

verdverm — Sat, 13 Jun 2026 02:15:19 +0000

They'll have to remove sections like this from their AI Action Plan

> We need to ensure America has leading open models founded on American values. Open- source and open-weight models could become global standards in some areas of business and in academic research worldwide. For that reason, they also have geostrategic value. While the decision of whether and how to release an open or closed model is fundamentally up to the developer, the Federal government should create a supportive environment for open models.

New comment by verdverm in "Statement on US government directive to suspend access to Fable 5 and Mythos 5"

verdverm — Sat, 13 Jun 2026 01:52:17 +0000

The biggest open models are also Chinese

New comment by verdverm in "Google is finally killing uBlock Origin in Chrome for good"

verdverm — Thu, 11 Jun 2026 20:13:10 +0000

Ai will help here, I maintain several forks for software I use to add a feature or two

New comment by verdverm in "OpenCode Data"

verdverm — Thu, 11 Jun 2026 20:01:41 +0000

I've been using a lot of DeepSeek through OpenCode Go, it's the best price/perf right now imho.

Now looking for vendors that have the certs for work. Supposedly OC has something in the works, but tbd when that will happen and if it will satisfy secops.

New comment by verdverm in "The Great American AI Act (draft) [pdf]"

verdverm — Thu, 11 Jun 2026 19:42:51 +0000

https://www.fisherphillips.com/en/insights/insights/congress...

Facets discussed in more detail here: https://www.youtube.com/watch?v=uv_IyFimJiU

The Great American AI Act (draft) [pdf]

verdverm — Thu, 11 Jun 2026 19:42:51 +0000

Article URL: https://obernolte.house.gov/sites/evo-subsites/obernolte.house.gov/files/evo-media-document/the-great-american-ai-act-discussion-draft-website-compressed-compressed.pdf

Comments URL: https://news.ycombinator.com/item?id=48495422

Points: 8

# Comments: 1

New comment by verdverm in "Ask HN: How do you get into a flow state when using AI to code?"

verdverm — Thu, 11 Jun 2026 17:21:22 +0000

Keeping an IDEAS.md or SCRATCH.md, which I work on while reviewing changes / code / docs, and it is working on other things.

I tend to focus on on project at a time with multiple agents, rather than agents on multiple project, and then time slice myself across projects

New comment by verdverm in "Ask HN: Is there a metric for AI code quality?"

verdverm — Thu, 11 Jun 2026 14:01:13 +0000

Why would a metric for code quality be different depending on how the code got to to a file? In other words, if there was a good measure, would it not exist already for us? How do we measure the quality of our own code?

New comment by verdverm in "Claude Opus is more performant on OpenCode than Claude Code"

verdverm — Thu, 11 Jun 2026 13:58:45 +0000

One is more intentional in what gets added, one is proud of the speed they ship at

quality vs quantity is an age old trade off

New comment by verdverm in "Bluesky will launch Reddit style communities"

verdverm — Thu, 11 Jun 2026 13:54:21 +0000

I think Bluesky is struggling with user numbers and cannibalizing the ecosystem, leaving little room for other teams to have an opportunity. Leadership over there is the primary reason I left atproto

New comment by verdverm in "Notes on DeepSeek"

verdverm — Thu, 11 Jun 2026 05:28:20 +0000

AllenAi seems to the most open lab to me, sharing not just models and even training data, but also the training code and methodologies.

https://allenai.org/olmo