Top all-time · #1573
Answer.AI
Company blog · answer.ai/posts · on HN as jph00aiprogramming
Answer.AI is the company blog at answer.ai, writing about ai and programming. Since 2023, 45 different Hacker News readers have linked it 71 times, most often “SB-1047 will stifle open-source AI and decrease safety”. It is still writing: the last post was on 2026-08-18, and you can follow it by feed.
Practical AI R&D
Most-linked posts?
- SB-1047 will stifle open-source AI and decrease safety7 people · 7 mentions · first linked 2024-04-28
- Show HN: WebGPU Puzzles - Learn GPU Programming in Your Browser4 people · 4 mentions · first linked 2024-09-13
- Fine tune a 70B language model at home4 people · 4 mentions · first linked 2024-03-07
- Faith and Fate: Transformers as fuzzy pattern matchers3 people · 3 mentions · first linked 2024-07-26
- A little pooling goes a long way for multi-vector representations3 people · 3 mentions · first linked 2024-06-27
- A Replacement for Bert: Introducing ModernBERT3 people · 3 mentions · first linked 2024-12-24
- /llms.txt–a proposal to provide information to help LLMs use websites2 people · 3 mentions · first linked 2024-09-03
- Claudette, a new friend that makes Claude 3.5 Sonnet even nicer2 people · 2 mentions · first linked 2024-06-21
- MonsterUI: Python library for building front end UIs quickly in FastHTML apps2 people · 2 mentions · first linked 2025-02-13
- Efficient finetuning of Llama 3 with FSDP QDoRA2 people · 2 mentions · first linked 2024-04-23
What Hacker News says?
The article actually sneaks in a footnote that answers this ( https://www.answer.ai/posts/2024-03-06-fsdp-qlora.html#fn1 ): "Throughout this article “training” can refer to either pre-training, or fine-tuning". (Generally, we've told students at fast.ai since 2017 that they should almost never be starting from random weights -- most of the time it's best to start with a pretrained model and fine-tune that, even if i…
Comment linking /posts/2024-03-06-fsdp-qlora.html
on “Fine tune a 70B language model at home”Submitted Fine tune a 70B language model at home as a story · 909 points
HN discussionSome laws are just bad . When the API-mediated/closed-weights companies agree with the open-weight/operator-aligned community that a law is bad, it’s probably got to be pretty awful. That said, though my mind might be playing tricks on me, I seem to recall the big labs being in favor at one time. There are a number of related threads linked, but I’ll personally highlight Jeremy Howard’s open letter as IMHO the best-…
Comment linking /posts/2024-04-29-sb1047.html
on “Gavin Newsom vetoes SB 1047”Submitted Thoughts on a month with Devin as a story · 285 points
HN discussionThe web interface gives you access to the Artifacts feature where it can build SPAs and render them in the browser. For terminal access I like using my own https://llm.datasette.io/ tool with the https://github.com/simonw/llm-claude-3 plugin For Python library access I recommend checking out Claudette: https://www.answer.ai/posts/2024-06-21-claudette.html
Comment linking /posts/2024-06-21-claudette.html
on “Diff-pdf: tool to visually compare two PDFs”Submitted Gpu.cpp: A lightweight library for portable low-level GPU computation as a story · 242 points
HN discussionHere is a port without the visualizer: https://twitter.com/srush_nlp/status/1719376959572980094 Here is an amazing in-browser implementation in WebGPU https://www.answer.ai/posts/2024-09-12-gpupuzzles.html
Comment linking /posts/2024-09-12-gpupuzzles.html
on “GPU Puzzles”Submitted SB-1047 will stifle open-source AI and decrease safety as a story · 225 points
HN discussionYes, correct, and that fetching operation is a non-blocking operation, once we dequantize the weights we discard it before moving to the next layer. Technically, you can do it for the weights as well. But that wouldn't work in many situations. For example, when training with FSDP: the quantized weights stay on the device but you can still offload the meta-data ( https://www.answer.ai/posts/2024-03-06-fsdp-qlora.html…
Comment linking /posts/2024-03-06-fsdp-qlora.html
on “Towards 1-bit Machine Learning Models”Submitted MonsterUI: Python library for building front end UIs quickly in FastHTML apps as a story · 166 points
HN discussionI get this eerie feeling she has no idea what a bot even is in a technical sense and most likely is being puppeteered by weasels like State Senator Scott Wiener having introduced gems like the AI Models Act [1][2][3]. If the bill passes wouldn't then she have to be the first to disclose she's a bot? [1] https://www.answer.ai/posts/2024-04-29-sb1047.html (SB-1047 will stifle open-source AI and decrease safety) [2] ht…
Comment linking /posts/2024-04-29-sb1047.html
on “California bill would require bots to disclose that they are bots”Submitted Lessons from history's greatest R&D labs as a story · 153 points
HN discussionchef's kiss This guy and the person that quit the bullshit industrial complex 6 months ago should get together and launch a startup. Better yet, we should all go and join Jeremy Howard's answer.ai pro bono. Besides being miraculously headed by a guy who is Not An Asshole, it incidentally also had the most refreshing launch post (in the warm and fuzzy way) this side of the AI bubble. The launch post concluded with th…
Comment linking /posts/2023-12-12-launch.html
on “A Eulogy for DevOps”Submitted Solveit – A course and platform for solving problems with code as a story · 111 points
HN discussionWe just published an article introducing gpu.cpp, what it's for, and how it works: https://www.answer.ai/posts/2024-07-11--gpu-cpp.html
Comment linking /posts/2024-07-11--gpu-cpp.html
on “Gpu.cpp: A lightweight library for portable low-level GPU computation”Submitted So where are all the AI apps? as a story · 82 points
HN discussionFor an idea of what's possible you might be interested in this story that was just on HN where they fine tune a quantized 70b model in 48GB VRAM: https://www.answer.ai/posts/2024-03-06-fsdp-qlora.html
Comment linking /posts/2024-03-06-fsdp-qlora.html
on “Show HN: LlamaGym – fine-tune LLM agents with online reinforcement learning”Submitted What policy makers need to know about AI as a story · 81 points
HN discussionFrom the post describing llms.txt ( https://www.answer.ai/posts/2024-09-03-llmstxt.html ): "The problem this solves is that today, constructing the right context for LLMs based on a website is ambiguous — do you: 1. Crawl the sitemap and include every page, trying to automatically format into an LLM-friendly form? 2. Selectively include external links in addition to the sitemap? 3. For specific domains like software…
Comment linking /posts/2024-09-03-llmstxt.html
on “Llms.txt”Submitted Train a 70b language model at home (2024) as a story · 71 points
HN discussionShellsage has provided this functionality for quite a while. I've been using it for months, and it's been a game-changer for me. It was created by one of my colleagues, Nathan Cooper. https://www.answer.ai/posts/2024-12-05-introducing-shell-sag...
Comment linking /posts/2024-12-05-introducing-shell-sage.html
on “TmuxAI: AI-Powered, Non-Intrusive Terminal Assistant”Submitted Claudette, a new friend that makes Claude 3.5 Sonnet even nicer as a story · 22 points
HN discussionThe other recent improvement suggested for LoRA is DoRA: https://magazine.sebastianraschka.com/p/lora-and-dora-from-s... . It really does seem to strongly outperform LoRA - see also https://www.answer.ai/posts/2024-04-26-fsdp-qdora-llama3.htm...
Comment linking /posts/2024-04-26-fsdp-qdora-llama3.html
on “LoRA+: Efficient Low Rank Adaptation of Large Models”Submitted Show HN: WebGPU Puzzles - Learn GPU Programming in Your Browser as a story · 14 points
HN discussionI would like to throw our project in the ring that solves this problem: https://github.com/tjmlabs/ColiVara 1. Uses half-vecs, so you cut down everything by half with no recall loss 2. Uses token pooling with hierarchial clustering at 3, so, you further cut down things by 2/3rd with <1% loss 3. Everything is on Postgres and pgvector, so you can do all the Postgres stuff and decrease corpus size by document metadata…
Comment linking /posts/colbert-pooling.html
on “VectorChord: Store 400k Vectors for $1 in PostgreSQL”Submitted Efficient finetuning of Llama 3 with FSDP QDoRA as a story · 13 points
HN discussionI've written a submission to the authors of this bill, and made it publicly available here: https://www.answer.ai/posts/2024-04-29-sb1047.html The EFF have also prepared a submission: https://www.context.fund/policy/2024-03-26SB1047EFFSIA.pdf A key issue with the bill is that it criminalises creating a model that someone else uses to cause harm. But of course, it's impossible to control what someone else does with y…
Comment linking /posts/2024-04-29-sb1047.html
on “Call-to-Action on SB 1047 – Frontier Artificial Intelligence Models Act”Submitted How to Solve It with Code as a story · 11 points
HN discussionhttps://llmstxt.org/ https://www.answer.ai/posts/2024-09-03-llmstxt.html
Comment linking /posts/2024-09-03-llmstxt.html
on “Cloudflare.com's Robots.txt”Submitted Cold Compress: A Hackable Toolkit for KV Cache Compression as a story · 8 points
HN discussion
