Hacker News: buppermint

New comment by buppermint in "How concerned should we be about Astra's recurrent architecture?"

buppermint — Fri, 04 Sep 2026 02:28:18 +0000

CoT is both correlated and causal of the model's real computations, it's just imperfect. If you manually add "Let's wrap it up" in the CoT during generation, most LLMs will actually wrap it up (this is a commonly used trick in local LLM circles to get long-winded LLMs to stop reasoning). This wouldn't work if CoT text didn't affect the actual internal model logic.

New comment by buppermint in "Domain-Camouflaged Injection Attacks Evade Detection in Multi-Agent LLM Systems"

buppermint — Fri, 22 May 2026 22:11:48 +0000

The paper title is a bit misleading. The tested detectors and models here are small and rather dated (Llama 3.1 8B and Gemini Flash 2.0 - these are basically in the level of a modern 1B model), and the actual paper says this only shows vulnerability in small model systems.

New comment by buppermint in "Job Postings for Software Engineers Are Rapidly Rising"

buppermint — Sat, 02 May 2026 06:23:36 +0000

Worth seeing the whole chart in perspective:

https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE

New comment by buppermint in "Claude's new constitution"

buppermint — Thu, 22 Jan 2026 03:27:57 +0000

Anthropic has already has lower guardrails for DoD usage: https://www.theverge.com/ai-artificial-intelligence/680465/a...

It's interesting to me that a company that claims to be all about the public good:

- Sells LLMs for military usage + collaborates with Palantir

- Releases by far the least useful research of all the major US and Chinese labs, minus vanity interp projects from their interns

- Is the only major lab in the world that releases zero open weight models

- Actively lobbies to restrict Americans from access to open weight models

- Discloses zero information on safety training despite this supposedly being the whole reason for their existence

New comment by buppermint in "Anthropic made a mistake in cutting off third-party clients"

buppermint — Mon, 12 Jan 2026 18:19:52 +0000

I would disagree on the knowledge sharing. They're the only major AI company that's released zero open weight models. Nor do they share any research regarding safety training, even though that's supposedly the whole reason for their existence.

New comment by buppermint in "GLM-4.7: Advancing the Coding Capability"

buppermint — Mon, 22 Dec 2025 21:37:49 +0000

I've been playing around with this in z-ai and I'm very impressed. For my math/research heavy applications it is up there with GPT-5.2 thinking and Gemini 3 Pro. And its well ahead of K2 thinking and Opus 4.5.

New comment by buppermint in "On the Existence, Impact, and Origin of Hallucination-Associated Neurons in LLMs"

buppermint — Mon, 22 Dec 2025 15:30:00 +0000

From a quick read, this is cool but maybe a little overstated. From Figure 3, completely suppressing these neurons only reduces hallucinations by like ~5% compared to their normal state.

Table 1 is even more odd, H-neurons predicts hallucination ~75% of the time but a similar % of random neurons predict hallucinations ~60% of the time, which doesn't seem like a huge difference to me.

New comment by buppermint in "Google is powering a new US Military AI platform"

buppermint — Wed, 10 Dec 2025 04:08:58 +0000

This is because Grok Code Fast is free via Kilo Code/Cline and has been for months

New comment by buppermint in "Show HN: Erdos – open-source, AI data science IDE"

buppermint — Mon, 27 Oct 2025 19:25:12 +0000

Very cool. Any plans to add support for local models? This has what has prevented us from adopting Positron so far. We have sensitive data and sending to third party APIs is not an option (regardless of their stated retention policies).