Hacker News: cpburns2009

New comment by cpburns2009 in "Ask HN: How much time do you spend cooking every day?"

cpburns2009 — Sat, 06 Jun 2026 14:32:58 +0000

Parmesan chicken, gulasch and spatzle, vegetable soup, summer tortellini skillet, lasagna (~3 hr), baked mac and cheese, baked spaghetti. I usually cook some frozen vegetables if the dish doesn't have any.

New comment by cpburns2009 in "Ask HN: What was your "oh shit" moment with GenAI?"

cpburns2009 — Sat, 06 Jun 2026 00:19:31 +0000

My "oh shit" is the enshitification, people blindly accepting the output without thought or review. LLMs are a remarkable technology. But despite the capability, they're vastly oversold.

New comment by cpburns2009 in "Ask HN: How much time do you spend cooking every day?"

cpburns2009 — Fri, 29 May 2026 12:41:47 +0000

I usually spend 2 hours cooking on Mondays to make dinner for the rest of the week. Breakfast and lunch are quick meals requiring 0-10 min prep day of.

New comment by cpburns2009 in "Is my blue your blue? (2024)"

cpburns2009 — Tue, 28 Apr 2026 12:49:16 +0000

Yeah that scene doesn't make any sense. A dark teal that could be confused between blue or green would look nothing like navy blue.

New comment by cpburns2009 in "Is my blue your blue? (2024)"

cpburns2009 — Tue, 28 Apr 2026 12:43:28 +0000

I think of purple as encompassing both indigo (blueish purple) and violet (pinkish purple).

New comment by cpburns2009 in "Mistral built a $14B AI empire by not being American"

cpburns2009 — Mon, 27 Apr 2026 17:44:57 +0000

There's a method to the madness:

- Mistral-Nemo: the actual model developed by Mistral and Nvidia.

- 2407: likely the release date of the base model, July of 2024.

- 12B: the model has 12 billion parameters.

- Thinking: the model operates in thinking mode (generates output plan and injests it before producing actual output).

- Claude-Gemini-GPT5.2: I think this means the model was finetuned with session data from Claude, Gemini, and GTP5.2 to replicate their behavior.

- Uncensored-HERITIC: the model was uncensored using the automated Heretic method.

- Q4_k_m: the model is quantized (lossy compression) to ~5 bpw from orignal 16 bpw.

New comment by cpburns2009 in "Refuse to let your doctor record you"

cpburns2009 — Sat, 25 Apr 2026 00:44:34 +0000

What competition? When enrollment comes around or you switch jobs you get maybe two insurance providers to choose from.

New comment by cpburns2009 in "Ubuntu 26.04"

cpburns2009 — Fri, 24 Apr 2026 18:25:32 +0000

It used to be true. I've never had problems using the proprietary Nvidia drivers on Ubuntu. You used to have to jump through a bunch of hoops to get them installed on Debian. Now Ubuntu lags behind on kernel versions leaving new hardware less usable.

New comment by cpburns2009 in "Refuse to let your doctor record you"

cpburns2009 — Fri, 24 Apr 2026 17:41:59 +0000

Profit margins are capped by percentage. That creates the perverse intensive for insurance companies to pursue ever increasing costs in order to increase profits.

New comment by cpburns2009 in "If America's so rich, how'd it get so sad?"

cpburns2009 — Fri, 24 Apr 2026 13:25:00 +0000

My pink crabapple is developing beautiful buds this year. Last year there was a windstorm that blew off all the flowers right before it reached full bloom.

New comment by cpburns2009 in "Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model"

cpburns2009 — Wed, 22 Apr 2026 18:28:36 +0000

Do you get early access so you can prep the quants for release?

New comment by cpburns2009 in "Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model"

cpburns2009 — Wed, 22 Apr 2026 18:27:06 +0000

For this model results are identical. In my experience it can go either way by up to 10%.

New comment by cpburns2009 in "Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model"

cpburns2009 — Wed, 22 Apr 2026 18:26:30 +0000

Results are nearly identical running on a Strix Halo using Vulkan, llama.cpp b8884:

    $ llama-batched-bench -dev Vulkan2 -hf unsloth/Qwen3.6-27B-GGUF:IQ4_XS -npp 1000,2000,4000,8000,16000,32000 -ntg 128 -npl 1 -c 34000
    |    PP |     TG |    B |   N_KV |   T_PP s | S_PP t/s |   T_TG s | S_TG t/s |      T s |    S t/s |
    |-------|--------|------|--------|----------|----------|----------|----------|----------|----------|
    |  1000 |    128 |    1 |   1128 |    3.288 |   304.15 |    9.873 |    12.96 |   13.161 |    85.71 |
    |  2000 |    128 |    1 |   2128 |    6.415 |   311.79 |    9.883 |    12.95 |   16.297 |   130.57 |
    |  4000 |    128 |    1 |   4128 |   13.113 |   305.04 |    9.979 |    12.83 |   23.092 |   178.76 |
    |  8000 |    128 |    1 |   8128 |   27.491 |   291.01 |   10.155 |    12.61 |   37.645 |   215.91 |
    | 16000 |    128 |    1 |  16128 |   59.079 |   270.83 |   10.476 |    12.22 |   69.555 |   231.87 |
    | 32000 |    128 |    1 |  32128 |  148.625 |   215.31 |   11.084 |    11.55 |  159.709 |   201.17 |

New comment by cpburns2009 in "Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model"

cpburns2009 — Wed, 22 Apr 2026 18:09:41 +0000

~25-26 tok/s with ROCm using the same card, llama.cpp b8884:

    $ llama-batched-bench -dev ROCm1 -hf unsloth/Qwen3.6-27B-GGUF:IQ4_XS -npp 1000,2000,4000,8000,16000,32000 -ntg 128 -npl 1 -c 34000
    |    PP |     TG |    B |   N_KV |   T_PP s | S_PP t/s |   T_TG s | S_TG t/s |      T s |    S t/s |
    |-------|--------|------|--------|----------|----------|----------|----------|----------|----------|
    |  1000 |    128 |    1 |   1128 |    1.034 |   966.90 |    4.851 |    26.39 |    5.885 |   191.67 |
    |  2000 |    128 |    1 |   2128 |    2.104 |   950.38 |    4.853 |    26.38 |    6.957 |   305.86 |
    |  4000 |    128 |    1 |   4128 |    4.269 |   937.00 |    4.876 |    26.25 |    9.145 |   451.40 |
    |  8000 |    128 |    1 |   8128 |    8.962 |   892.69 |    4.912 |    26.06 |   13.873 |   585.88 |
    | 16000 |    128 |    1 |  16128 |   19.673 |   813.31 |    4.996 |    25.62 |   24.669 |   653.78 |
    | 32000 |    128 |    1 |  32128 |   46.304 |   691.09 |    5.122 |    24.99 |   51.426 |   624.75 |

New comment by cpburns2009 in "M 7.4 earthquake – 100 km ENE of Miyako, Japan"

cpburns2009 — Mon, 20 Apr 2026 19:26:10 +0000

It's an unpopular opinion for sure. Evangelion had great potential, a mysterious world, but it never reached it. I kept waiting for it to get good and then it just abruptly ended. I couldn't stand Shinji either. His situational paralysis was so frustrating. If he'd been a coward it would at least be understandable. But no, he sits there frozen half the show.

New comment by cpburns2009 in "IntelliJ IDEA and IntelliJ Platform and become open source"

cpburns2009 — Fri, 17 Apr 2026 14:00:30 +0000

The community edition has been open source since 2009. This isn't new.

New comment by cpburns2009 in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

cpburns2009 — Thu, 16 Apr 2026 21:42:14 +0000

Apparently it's a known issue with CUDA 13.2 [1].

[1] https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/discussi...

New comment by cpburns2009 in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

cpburns2009 — Thu, 16 Apr 2026 19:18:26 +0000

Personally, I wouldn't trust any foreign or domestic LLM providers to not train on your data. I also wouldn't trust them to not have a data breach eventually which is worse. If you're really worried about your data, run it locally. The Chinese models (Qwen, GLM, etc.) are really competitive to my understanding.

New comment by cpburns2009 in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

cpburns2009 — Thu, 16 Apr 2026 19:09:23 +0000

Sir, this is 2026. You're not getting any 128GB of RAM for under $1k.

New comment by cpburns2009 in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

cpburns2009 — Thu, 16 Apr 2026 19:04:35 +0000

Anyone else getting gibberish when running unsloth/Qwen3.6-35B-A3B-GGUF:UD-IQ4_XS on CUDA (llama.cpp b8815)? UD-Q4_K_XL is fine, as is Vulkan in general.