smy20011

Nice to see a AI coding tool that not (complete) vibe coded! Well done.

Are we reinventing RPC again? Calling CLI program with JSON sounds like RPC call. The schema feels likes something LSP can provided for such function.

Maybe asking agent to write/execute code that wraps CLI is a better solution.

I think the good thing about it is that if you are given good specification, you are likely to get good result. Writing a C compiler is not something new, but it will be great for all the porting projects.

I miss entering flow state when coding. When vibe coding, you are in constant interruption and only think very shallow. I never see anyone enter flow state when vibe coding.

I think It's not because AI working on "misaligned" goals. The user never specify the goal clearly enough for AI system to work.

However, I think producing detailed enough specification requires same or even larger amount of work than writing code. We write rough specification and clarify these during the process of coding. I think there are minimal effort required to produce these specification, AI will not help you speed up these effort.

According to https://cdn.realfood.gov/Daily%20Serving%20Sizes.pdf, their recommendations do not meet their calories goal. Eg, for 2000 calories, you can eat 4 egg, 3 cup of milk, 4 slice of bread, 2 apple and 3 tbsp of oil per day.

Total calories will be 1,608 kcal/day.

It's a very depressing diet menu.

Just don't play the game have winner and loser. Play the game that both side can win.

> Often, it will drop you precisely at that golden moment where shit almost works, and development means tweaking code and immediately seeing things work better. That dopamine hit is why I code.

Only if you are familiar with the project/code. If not, you were throw into a foreign codebase and have no idea how to tweak it.

Yeah, fully local LLM+TTS setup.

Use jupyter notebook to fetch the stock and weather info and feed that into a local LLM and convert that to speech using opensource TTS.

https://github.com/smy20011/MorningRadio

Local server.

https://github.com/smy20011/MorningRadio

I use tailscale to build my personal podcast that include local weather and stocks I interested in. Running the whole pipeline on a steamdeck and use tailscale to securely delivery the generated podcast to my phone.

Curious why large open source project (like linux) have no manager for all the political stuff but works well but every private company have a huge management chain for less complex stuff.

Not that simple, It could cause resource curse [1] for developers. Why optimize algorithm when you have nearly infinity resources? For deepseek, their constrains is one of the reason they achieve breakthrough. One of their contribution, fp8 training, is to find a way to train models with GPUs that limit fp32 performance due to export control.

[1]: https://www.investopedia.com/terms/r/resource-curse.asp#:~:t...

Human do write code that scalable with compute.

The performance is always raw performance * software efficiency. You can use shitty software and waste all these FLOPs.

Did they? Deepseek spent about 17 months achieving SOTA results with a significantly smaller budget. While xAI's model isn't a substantial leap beyond Deepseek R1, it utilizes 100 times more compute.

If you had $3 billion, xAI would choose to invest $2.5 billion in GPUs and $0.5 billion in talent. Deepseek, would invest $1 billion in GPUs and $2 billion in talent.

I would argue that the latter approach (Deepseek's) is more scalable. It's extremely difficult to increase compute by 100 times, but with sufficient investment in talent, achieving a 10x increase in compute is more feasible.

Looks like the beginning of the downfall of America.

Do you have an example of such account? Most of the thing I saw is engagement bait.

It seems O3 following trend of Chess engine that you can cut your search depth depends on state.

It's good for games with clear signal of success (Win/Lose for Chess, tests for programming). One of the blocker for AGI is we don't have clear evaluation for most of our tasks and we cannot verify them fast enough.

Mind share the list? Asking for a friend.

Then why BlueSky is #1 free apps while twitter is not?

The top developer is the one who committed five PRs that changed Godot to Redot, but they had fewer than ten PRs the last time I checked.

A pretty good UX improvement for ChatGPT. I think they will fork VScode and make an AI editor later. Although It's confusing for me the OAI seems spend more time on end user product instead of working on AGI.

Numbers are impressive, not sure why it's down in AH.

Cisco in 2000: Gross margin in fiscal 2000 was 64.4%, compared with 65.0% in fiscal 1999. The following table shows the standard

[flagged]

It must be a big offer

Some cheapest BYD cars sells in China cost 14k USD. It's cheaper than a lot of EVs in US even considering tariffs.

Thanks but I would like to have a localLM version. I don't want my email sent to any third party for analysis.

I don't quite understand. Most of LLM inference is MatMul and Softmax. It's not rocket science to implement matrix multiplication in any GPU. We don't have very complex custom kernel for the LLM inference. If AMD can provide fast GPU with cheaper price, a lot of people don't mind to add support for them. GGML shows you can create a working LLM inference solution with one man's work.

Really cool!