lukax
Re:Oai-math-revised: a community hub for revising OpenAI's math write-ups
on Saturday October 24, 2026 @02:43PM (#50223916)
And now they can train on everything this community does. And with new models this community won't be needed anymore.
Re:DeepSeek just open-sourced its inference engine
on Tuesday October 20, 2026 @06:14AM (#50163829)
Their inference server/service (wrapper around this engine) could still be written in Rust, this is just the engine.
This is written in Rust: https://github.com/deepseek-ai/deepseek-recipe
This DeepSeek Recipe is also mentioned here: https://zhuanlan.zhihu.com/p/2095560295930508953
Re:Gemini 4 Argon
on Wednesday September 30, 2026 @05:07PM (#49914472)
Yes, through OpenRouter.
Re:HarnessTax: How Much Does the Harness Matter for Coding Agents?
on Thursday September 17, 2026 @12:41PM (#49743305)
OpenCode checks model name and registers the appropriate tools.
const usePatch = model.modelID.includes("gpt-") && !model.modelID.includes("oss") && !model.modelID.includes("gpt-4")
Pi uses its own tools, like Armin wrote in the linked article.
Re:HarnessTax: How Much Does the Harness Matter for Coding Agents?
on Thursday September 17, 2026 @01:27AM (#49736748)
What matters more is that you use the tools that the target model was fine-tuned on.
E.g. for editing files with Claude models you should use Edit(file_path, old_string, new_string, replace_all) but with GPT models you should use apply_patch_call(patch) (where patch is a custom patch string with custom grammar).
It appears newer models are better at narive harness tool calls and worse at custom tools that look similar to default tools.
https://lucumr.pocoo.org/2026/7/4/better-models-worse-tools/
Re:The DeepMind Institute
on Thursday September 17, 2026 @01:10AM (#49736647)
https://news.ycombinator.com/submitted?id=vertigoruntime
This looks like a new account from a PR agency focusing on AI labs.
Re:DeepSeek-v4-flash-vision-exp
on Friday August 21, 2026 @01:28PM (#49391316)
This is only a problem with OpenAI Chat Completions API. With OpenAI Responses API and Anthropic Messages API a tool output can be text, image or file.
Re:DeepSeek v4 Price Increase
on Thursday August 13, 2026 @09:37AM (#49285742)
DeepSeek-V4-Flash
Input (Cache Miss)
Previous: $0.14
New - Off-Peak: $0.22 (approx. 1.57x)
New - Peak: $0.44 (approx. 3.14x)
Input (Cache Hit)
Previous: $0.0028
New - Off-Peak: $0.007 (2.5x)
New - Peak: $0.014 (5x)
Output
Previous: $0.28
New - Off-Peak: $0.66 (approx. 2.35x)
New - Peak: $1.32 (approx. 4.71x)
---
DeepSeek-V4-Pro
Input (Cache Miss)
Previous: $0.435
New - Off-Peak: $0.66 (approx. 1.51x)
New - Peak: $1.32 (approx. 3.03x)
Input (Cache Hit)
Previous: $0.003625
New - Off-Peak: $0.022 (approx. 6.07x)
New - Peak: $0.044 (approx. 12.14x)
Output
Previous: $0.87
New - Off-Peak: $1.98 (approx. 2.27x)
New - Peak: $3.96 (approx. 4.55x)
Re:Run Kimi K3 using 29 GB of RAM at 0.50 tok/s
on Saturday August 01, 2026 @03:13AM (#49131912)
Yes. They are.
Re:ChatGPT claims rogue AI attacked more companies
on Wednesday July 29, 2026 @07:15AM (#49095941)
They somehow forgot to mention that Hugging Face tried to use frontier models to analyze the attack but all models rejected. They had to use GLM 5.2 deployed locally.
Re:Using Go for Mobile Apps
on Thursday July 16, 2026 @07:32PM (#48941656)
Also for WASM? It's just not really there compared to Rust's wasm bindgen.
Re:Using Go for Mobile Apps
on Wednesday July 15, 2026 @02:24PM (#48925035)
I can highly recommend Rust for this. The same Rust engine powering web app (WASM + React), iOS (SwiftUI), Android (Kotlin Compose) and desktop (Tauri).
Re:Benchmarking Coding Agents on Databricks' Multi-Million Line Codebase
on Thursday July 09, 2026 @01:33AM (#48841379)
Could it be that users of Pi are more senior and know better how to prompt and that's why the pass rate is higher?
Re:Benchmarking Coding Agents on Databricks' Multi-Million Line Codebase
on Thursday July 09, 2026 @01:29AM (#48841354)
Yes, this is a known issue. A significant amount of Edit tool calls fails in Pi witg newer models.
https://lucumr.pocoo.org/2026/7/4/better-models-worse-tools/
Re:GLM 5.2 and the coming AI margin collapse
on Tuesday July 07, 2026 @08:51AM (#48817021)
Well, Microsoft just started offering Kimi K2.7 through Copilot hosted on Azure.
https://github.blog/changelog/2026-07-01-kimi-k2-7-is-now-av...
Cursor Composer 2 and 2.5 are also fine tunes of Kimi K2.5
It looks like politics don't matter when it comes to economics.
Re:Netflix Simplified Batch Compute with Kueue
on Tuesday June 30, 2026 @01:04AM (#48728663)
It's refreshing to see a tech article that isn't about AI. It feels like 5 years ago.
Re:NUMA: Cores, memory, and the distance between them
on Monday June 29, 2026 @06:01AM (#48717145)
NUMA can cause really crappy performance. We deployed a Go based LLM gateway in Kubernetes deployed on a server with hundreds of CPU cores. We didn't explicitly set GOMAXPROCS so Go runtime scheduled goroutines over different CPUs and it constantly used 200% CPU and GC was causing latency spikes. Then we set GOMAXPROCS 8 and all performance issues went away. Until recently Kubernetes didn't work well with NUMA.
Re:Nearly half of LG smart TV apps contain residential proxy SDKs
on Monday June 22, 2026 @06:08PM (#48636940)
Well, that's how data for training LLMs is scraped.
Re:Claude Fable 5
on Wednesday June 10, 2026 @02:21PM (#48480489)
Zero data retention policies.
Re:Theseus: Translating Win32 to WASM
on Wednesday May 27, 2026 @03:12PM (#48299023)
There is also Retrotick.
It simulates x86 (win32 and win16) and implements Windows APIs in javascript and renders window frames with DOM and contents with canvas (e.g. GDI translates to browser canvas operations). A lot of programs run already but a lot of APIs are not yet implemented.
I successfullt spent a few days extending it to run a Click & Create based game from my childhood.
Re:Xiaomi MiMo-v2.5 Series API Permanent Price Reduction Up to 99%
on Tuesday May 26, 2026 @03:02PM (#48284320)
Huawei Ascend AI Accellerators. DeepSeek V4 model architecture was optimized for Chinese hardware.
Re:What Color is Your Function? (2015)
on Tuesday May 26, 2026 @02:34PM (#48283888)
That's just not true. Let's say you have a form validation library with a public api that supports custom validators Validate(name string, value string) bool. Then you decide that your validator now needs to make an HTTP request. This request needs context so that tracing is propagated and needs to return (bool, error) so that error is propagated up instead of silently ignoring it or logging it and returning false. This is coloring. You can use context.Background the same way you can use blocking in other languages. It just doesn't feel right and it breaks things.
Re:Antigravity 2.0 Tops the OpenSCAD Architectural 3D LLM Benchmark
on Saturday May 23, 2026 @01:49AM (#48245084)
I wonder what would happen if they used Kimi 2.5 directly instead of Cursor Composer 2.5. Composer is a fine tune of Kimi. Probably they didn't want to test "Chinese" models.
Re:GitHub is having issues now
on Monday April 27, 2026 @02:47PM (#47925601)
It looks like migration to Azure is not going very well
Re:GitHub is having issues now
on Monday April 27, 2026 @02:46PM (#47925571)
They are migrating from their own datacenters to Azure
Re:It's OK to compare floating-points for equality
on Saturday April 18, 2026 @07:31AM (#47815042)
See the implementation of Python's math.isclose
https://github.com/python/cpython/blob/d61fcf834d197f0113a6a...
Re:It's OK to compare floating-points for equality
on Saturday April 18, 2026 @07:24AM (#47815009)
You generally want both relative and absolute tolerances. Relative handles scale, absolute handles values near zero (raw EPSILON isn’t a universal threshold per IEEE 754).
The usual pattern is abs(a - b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol) to avoid both large-value and near-zero pitfalls.
Re:Someone Bought 30 WordPress Plugins and Planted a Backdoor in All o...
on Monday April 13, 2026 @03:43PM (#47756926)
Do you really need to roll your own NIO HTTP server? You could just use Jetty with virtual threads (still uses NIO under the hood though) and enjoy the synchronous code style (same as Go)
Re:Someone Bought 30 WordPress Plugins and Planted a Backdoor in All o...
on Monday April 13, 2026 @03:36PM (#47756860)
Rust wasm ecosystem also needs a lot of crates to do anything useful, a lot of them unmaintained.