Hacker News: benxh

New comment by benxh in "Qwen 3.8 27B"

benxh — Fri, 14 Aug 2026 15:15:47 +0000

Depends on what software/hardware you'll run it. GGUFs from Unsloth can run on pretty much every single potato; full weights need beefy gpus

New comment by benxh in "GLM 5.2 Performance Benchmarks"

benxh — Wed, 17 Jun 2026 13:40:24 +0000

benchmark where gemini flash is better than fable btw.

New comment by benxh in "US to suspend immigrant visa processing for 75 nations, State Department says"

benxh — Wed, 14 Jan 2026 22:36:50 +0000

Crazy calling sovereign states "US Puppets".

New comment by benxh in "Chinese AI models have lagged the US frontier by 7 months on average since 2023"

benxh — Thu, 08 Jan 2026 23:36:43 +0000

Minimax has been great for super high speed web/js/ts related work. It compares in my experience to Claude Sonnet, and at times gets stuff similar to Opus. Design wise it produces some of the most beautiful AI generated page I've seen.

GLM-4.7 like a mix of Sonnet 4.5 and GPT-5 (the first version not the later ones). It has deep deep knowledge, but it's often just not as good in execution.

They're very cheap to try out, so you should see how your mileage varies.

Ofcourse for the hardest possible tasks that GPT 5.2 only approaches, they're not up to scratch. And for the hard-ish tasks in C++ for example that Opus 4.5 tackles Minimax feels closer, but just doesn't "grok" the problem space good enough.

New comment by benxh in "Chinese AI models have lagged the US frontier by 7 months on average since 2023"

benxh — Thu, 08 Jan 2026 20:20:21 +0000

It is arguable that the new Minimax M2.1 and GLM4.7 are drastically above Sonnet 3.7 in capabilities.

New comment by benxh in "First Self-Propagating Worm Using Invisible Code Hits OpenVSX and VS Code"

benxh — Mon, 20 Oct 2025 22:54:50 +0000

cline is used by a lot of devs

New comment by benxh in "OpenAI o3-pro"

benxh — Tue, 10 Jun 2025 21:11:57 +0000

The longer "it" reasons, the more attention sinks are used to come to a "better" final output.

New comment by benxh in "DoppelBot: Replace Your CEO with an LLM"

benxh — Tue, 04 Feb 2025 16:54:28 +0000

To prove you right, you can read up on the incredible giga-brained countrywide experiments by Kardelj in Socialist Yugoslavia [0]. The result being a country where no-one wanted to work, and everyone had a great standard of living (while the IMF didn't call in its loans). And then the entire country collapsed all at once under the accumulated mismanagement.

[0] - https://en.wikipedia.org/wiki/Workers%27_self-management#Yug...

New comment by benxh in "Llama.cpp supports Vulkan. why doesn't Ollama?"

benxh — Fri, 31 Jan 2025 12:35:32 +0000

My biggest gripe with Ollama is the badly named models, e.g. under deepseek-r1, it defaults to the distill models.

New comment by benxh in "In the land of LLMs, can we do better mock data generation?"

benxh — Wed, 02 Oct 2024 12:49:50 +0000

I'm pretty sure that Neosync[0] does this to a pretty good degree, it is open source and YC funded too.

[0] https://www.neosync.dev/

New comment by benxh in "Anna’s Archive approaching 1 petabyte"

benxh — Sat, 20 Jul 2024 10:25:46 +0000

I am assuming this will be solved this year.

New comment by benxh in "Phind-70B: Closing the code quality gap with GPT-4 Turbo while running 4x faster"

benxh — Fri, 23 Feb 2024 18:05:03 +0000

If GPT4 is 220B/8 experts, that would be in-line with 3.5 Turbo being a 20B model, and GPT4 being a 55B activation out of a total 220B parameters.

It is ultimately all speculation, until Deepseek releases their own 145B MoE model, and then we can compare the activations/results

New comment by benxh in "A million ways to die on the web"

benxh — Thu, 18 Jan 2024 15:03:40 +0000

I personally was affected by this fire, although I've always kept 3 month backups of production data, encrypted, on-site, just in case of emergencies like this. Haven't touched their services for anything production related ever since

New comment by benxh in "AlphaCode2 powered by Gemini performs better than 85% of programmers"

benxh — Wed, 06 Dec 2023 18:27:52 +0000

It's buried deep in the Gemini report, but goddamn are these incredible stats.

New comment by benxh in "OpenAI's board has fired Sam Altman"

benxh — Fri, 17 Nov 2023 20:57:58 +0000

The Albanian takeover of AI continues. It's incredibly exciting!

New comment by benxh in "Phind Model beats GPT-4 at coding, with GPT-3.5 speed and 16k context"

benxh — Tue, 31 Oct 2023 18:30:22 +0000

I can't wait to see this open sourced, there's a lot of sampling strategies that help coding.

And I also can't wait to see how much Phind will improve further if the Glaive dataset is added onto it.

Edit: Contrastive search, dynamic temperatures.

New comment by benxh in "China is becoming a data black hole, says short seller Aandahl"

benxh — Mon, 23 Oct 2023 10:25:04 +0000

It was known by Polynesians for at least 1000 years before Columbus. See sweet potatoes.

New comment by benxh in "Mistral 7B"

benxh — Wed, 11 Oct 2023 13:14:35 +0000

It's missing a lot of crucial details. Nothing on the dataset used, nothing on the data mix, nothing on their data cleaning procedures, nothing on the tokens trained.

New comment by benxh in "Tiny Language Models Come of Age"

benxh — Sun, 08 Oct 2023 19:58:00 +0000

Not all of them per se, take a look at something like Mistral. It's a 7B model displaying incredible performance. IMO, we still haven't even scratched the surface of what is possible with small LLMs. Especially not with pre-filtered/classified pre-training data. (Interesting LLMs based on their data approach and relatively small size: Qwen, InternLM, Mistral, Phi)

New comment by benxh in "OpenAI's justification for why training data is fair use, not infringement [pdf]"

benxh — Sat, 07 Oct 2023 17:23:41 +0000

Added, and reached out on Twitter.