Hacker News: a11r

New comment by a11r in "Qwen3.8-Flash-Next"

a11r — Wed, 26 Aug 2026 22:46:34 +0000

Quoting RGFusion from Reddit: LLMs run into an issue where the further you train a model, the more it overwrites facts with generalized concepts. You need the model to be able to do both. Intelligence arises from generalization, but without accurate information the model will hallucinate.

The engram table allows for a low-computational method of fact-recall. You can think of it like a better form of RAG, where the data doesn't take up any of your context window and it's injected deeper into the model's layers, freeing the lower layers to carry out abstraction. This results in better "focus" for the model, both in regards to its intelligence and context recall.

Basically, they've separated the specificity-critical portions of the models memory into a parameter space that doesn't need fast compute (you can run it on system RAM) and allows the model to be trained on higher volumes of data without ruining its knowledge-base.

https://www.reddit.com/r/LocalLLaMA/comments/1vy6smx/comment...

New comment by a11r in "Why your local LLM feels dumber than it is"

a11r — Sun, 23 Aug 2026 00:35:14 +0000

Even a 4-bit quant of Qwen3.8 27b is indistinguishable from Gemini 3.7 flash in our internal tests. With an RTX5090 card and ninfer, you can get ~800 TPS token generation (c=8) and ~140 Tokens per second single stream.

New comment by a11r in "Why your local LLM feels dumber than it is"

a11r — Sun, 23 Aug 2026 00:32:03 +0000

M3 Pro 36GB. I am getting 17 tps with MTPLX.

New comment by a11r in "Interview with Amit Patel, Creator of “Solar Realms Elite” (2013)"

a11r — Mon, 17 Aug 2026 03:35:19 +0000

Amit Patel was employee #7 at Google and built many things including the little geo display in the lobby that would show live search queries as dots color coded by language.

New comment by a11r in "Qwen 3.8 Max Live Now"

a11r — Mon, 03 Aug 2026 02:53:44 +0000

The pricing is 40% of Kimi K3 at the same Model size. It will be interesting to see benchmarks tomorrow and to see how the pricing evolves across this size class.

New comment by a11r in "Kimi-K3 on HuggingFace"

a11r — Mon, 27 Jul 2026 17:38:04 +0000

Seminanlysis is estimating sub $1 cost per MT for ~2Trillion models. The numbers change based on throughput and quant, but it is conceivable that provider costs at scale are low enough that even $2.42 per MT on GLM 5.2 (current best price) is margin positive by a wide margin.

New comment by a11r in "Why Developers Keep Choosing Claude over Every Other AI"

a11r — Thu, 26 Feb 2026 16:55:36 +0000

This resonates with my experience. At Morph we use gemini for well specified point coding tasks, and it does very well across millions of lines of code every day. We also use claude code as an engineering tool for our own codebase and it does better at being adaptive and for working on open ended issues.

New comment by a11r in "Code Storage by the Pierre Computer Company"

a11r — Sat, 14 Feb 2026 19:49:54 +0000

Understood. I am looking for a side-by-side comparison focused on your feature set, not Github's. You answered it partially by calling out your focus areas. Github API reliability has been iffy for us, so it would be good to quantify the difference we can expect to get with you.

New comment by a11r in "Code Storage by the Pierre Computer Company"

a11r — Sat, 14 Feb 2026 18:13:07 +0000

It would be nice to see a side-by-side comparison with Github on pricing and features. We are using github and creating hundreds of repos everyday without any issues (except for the occassional API outages that Github has). Curious to see your take on where Pierre is better.

New comment by a11r in "Improving 15 LLMs at Coding in One Afternoon. Only the Harness Changed"

a11r — Thu, 12 Feb 2026 14:43:11 +0000

This is very nicely done. We have seen the same issue at a higher level of getting separators right when generating multiple files in a single inference call.

New comment by a11r in "Ultrasound is ushering a new era of surgery-free cancer treatment"

a11r — Tue, 14 Oct 2025 14:22:32 +0000

https://house.fandom.com/wiki/Clueless

New comment by a11r in "Create personal illustrated storybooks in the Gemini app"

a11r — Wed, 06 Aug 2025 03:24:49 +0000

Nice ! Curious what your prompt was.

New comment by a11r in "Oklo, the Earth's Two-billion-year-old only Known Natural Nuclear Reactor (2018)"

a11r — Fri, 20 Jun 2025 18:46:06 +0000

These are large coal seams with significant exposure to the atmosphere. See https://en.wikipedia.org/wiki/Jharia_coalfield for an example. That excavator in the picture is not trying to put out the fire, it is just mining coal that happens to be burning. Spray some water, put out the fire and ship it off to customers.

New comment by a11r in "The Illusion of Thinking: Strengths and limitations of reasoning models [pdf]"

a11r — Sun, 08 Jun 2025 16:36:40 +0000

The system prompt in this experiment limits the solution to always spell out the concrete moves verbally. A human solving the Tower of Hanoi gives up around N=4 and goes off to invent a recursive solution instead. Prompted differently, the LLM would solve these puzzles just fine.

Here is my complete review/analysis of the paper: https://www.linkedin.com/pulse/art-abstraction-human-advanta...

edit: fixed typo

New comment by a11r in "Rethinking Diabetes – interview with Gary Taubes"

a11r — Mon, 15 Jan 2024 06:44:26 +0000

Gary Taubes has written multiple scholarly books on the subject. He has won the Science in Society Journalism Award of the National Association of Science Writers three times. I think a training in one scientific field does qualify a person to go spend their time digging into other fields. How well they do is a matter of personal ability and effort and one would have to read their output and judge for oneself. I have read many of Taubes's book and many of the papers cited in his books and have concluded that he is a well qualified expert in the field.

New comment by a11r in "Google Maps has become significantly less usable of late"

a11r — Tue, 26 Sep 2023 14:45:24 +0000

For people with mobility challenges, or those with communication challenges, being able to look at a map of the airport and plan in advance is useful.

New comment by a11r in "Alexander the ‘Accursed’ and Zoroastrianism"

a11r — Wed, 15 Feb 2023 17:36:57 +0000

"Yunnani" medicine comes from the region of Yunnan [ https://en.wikipedia.org/wiki/Yunnan ] and is unrelated to Yunan.

New comment by a11r in "Apple’s ad business set to boom on the back of its own anti-tracking crackdown"

a11r — Mon, 03 Oct 2022 20:19:24 +0000

It depends. Sometimes you don't know that a solution exists for your problem.

New comment by a11r in "The night I met Einstein (1955)"

a11r — Mon, 18 Jul 2022 22:16:16 +0000

I am in the middle of listening to Einstein: His Life and Universe by Walter Isaacson and would highly recommend this biography to anyone interested in more.

New comment by a11r in "California’s demography is at odds with the old California Dream"

a11r — Tue, 03 May 2022 04:03:01 +0000

The State Controller's office publishes detailed reports: https://www.sco.ca.gov/ard_state_acfr.html