Hacker News Reader

jbentley1on Claude Code for VSCode

Yes there is. IDEs just aren't designed for it. The main screen in an IDE is a single branch at a time, I want to be managing a swarm of agents on multiple branches/worktrees

jbentley1on Claude Code for VSCode

Noted! Thanks

jbentley1on Claude Code for VSCode

I think integration into existing IDEs is the wrong form for agentic coding. The best way to work is managing several Git worktrees with agents running so you aren't stuck waiting 20+ minutes for Claude Code to finish.

I built a UI to manage this, and it is starting to turn into a new type of IDE, based around agent management and review rather than working on one thing at a time.

https://github.com/stravu/crystal

jbentley1on Show HN: Claude Code Usage Monitor – real-time tracker to dodge usage cut-offs

This is great. I built a UI tool to run simultaneous Claude Code sessions (https://github.com/stravu/crystal) but by working on multiple features at once I hit the limits of my max account. Usually close to the reset, but it would be nice to know when it is time to take a break.

jbentley1on Multiple Claude Code Sessions in an Easy UI: Crystal

No, but people have been asking about that. Can you describe how you would want to use it remotely? What is your use case?

jbentley1on Multiple Claude Code Sessions in an Easy UI: Crystal

The line between what AI can write and what it can't is moving and it is moving quickly. My bet with this is that we will want tools built for the new reality rather than repurposing our older tools, even though they won't completely go away.

jbentley1on Multiple Claude Code Sessions in an Easy UI: Crystal

Keep the feature requests coming! And feel free to create PRs. I am getting a lot out of the tool as is but there are so many things that could be added.

jbentley1on Taurine and aging: Is there anything to it?

Read the first couple lines, closed the page, and concluded that I should drink more Red Bull.

jbentley1on The Illusion of Thinking: Strengths and limitations of reasoning models [pdf]

Is Apple failing at AI so they just put all their R&D towards convincing themselves it isn't important?

jbentley1on Deepseek R1-0528

From conversation with someone from Groq, they have a custom compiler and runtime for the models to run on their custom hardware, which is why the selection is poor. For every model type they need to port the architecture to run on their compiler beforehand.

jbentley1on Chomsky on what ChatGPT is good for (2023)

"I think we should probably stop listening to Chomsky"

I've been saying this my whole life, glad it's finally catching on

jbentley1on Claude's system prompt is over 24k tokens with tools

This isn't exactly correct, it is a combination of training and system prompt.

You could train the system prompt into the model. This could be as simple as running the model with the system prompt, then training on those outputs until it had internalized the instructions. The downside is that it will become slightly less powerful, it is expensive, and if you want to change something you have to do it all over again.

This is a little more confusing with Anthropic's naming scheme, so I'm going to describe OpenAI instead. There is GPT-whatever the models, and then there is ChatGPT the user facing product. They want ChatGPT to use the same models as are available via API, but they don't want the API to have all the behavior of ChatGPT. Hence, a system prompt.

If you do use the API you will notice that there is a lot of behavior that is in fact trained in. The propensity to use em dashes, respond in Markdown, give helpful responses, etc.

jbentley1on Visiting Us

I'm guessing somebody was listening to the latest episode of Acquired?

jbentley1on Why R the Critical Value and Emergent Behavior of Large Language Models Fake?

It's fairly easy to observe that LLMs do things that were out of reach for software only a short period ago, and that their abilities are increasing fast. Yet some people try to make trivial arguments like this and try to say AI is somehow 'fake'.

jbentley1on GPT-4.1 in the API

https://fiction.live/stories/Fiction-liveBench-Mar-25-2025/o...

IMO this is the best long context benchmark. Hopefully they will run it for the new models soon. Needle-in-a-haystack is useless at this point. Llama-4 had perfect needle in a haystack results but horrible real-world-performance.

jbentley1on Fine-tune Google's Gemma 3

I am. I have some use cases related to data extraction where using a fine tuned small model outperforms the best-in-class closed source models and at a fraction of the cost.

jbentley1on Mandelbrot Set Explorer

Zoom is jerky and the number of iterations don't automatically scale as I zoom in. Maybe I'm expecting too much for something that runs in the browser but seems like there is room for improvement.

jbentley1on I built a failed startups goods website and named it VcSubsidized.com

From the name I was hoping this would be a list of services being offered at a loss as they burn VC money to get marketshare.

jbentley1on Apple Invites

Unless they add some basic functionality to include Android users, this is the evilest use of their walled garden yet.

jbentley1on Mistral Small 3

It's 24B parameters

jbentley1on Ask HN: Is anyone doing anything cool with tiny language models?

The technical parts are less common and specialized, like understanding the hyperparameters and all that, but I don't think that is the main problem. Most people don't understand how to build a good dataset or how to evaluate their finetune after training. Some parts of this are solid rules like always use a separate validation set, but the task dependent parts are harder to teach. It's a different problem every time.

jbentley1on Ask HN: Is anyone doing anything cool with tiny language models?

Tiny language models can do a lot if they are fine tuned for a specific task, but IMO a few things are holding them back:

1. Getting the speed gains is hard unless you are able to pay for dedicated GPUs. Some services offer LoRA as serverless but you don't get the same performance for various technical reasons.

2. Lack of talent to actually do the finetuning. Regular engineers can do a lot of LLM implementation, but when it comes to actually performing training it is a scarcer skillset. Most small to medium orgs don't have people who can do it well.

3. Distribution. Sharing finetunes is hard. HuggingFace exists, but discoverability is an issue. It is flooded with random models with no documentation and it isn't easy to find a good oen for your task. Plus, with a good finetune you also need the prompt and possibly parsing code to make it work the way it is intended and the bundling hasn't been worked out well.

jbentley1on Zuckerberg approved training Llama on LibGen [pdf]

Things can both please share holders and create value for users of that thing.

jbentley1on Lines of code that will beat A/B testing every time (2012)

Multi-armed bandits make a big assumption that effectiveness is static over time. What can happen is that if they tip traffic slightly towards option B at a time when effectiveness is higher (maybe a sale just started) B will start to overwhelmingly look like a winner and get locked in that state.

You can solve this with propensity scores, but it is more complicated to implement and you need to log every interaction.