Apr 28 – May 16, 2025
First-generation languages (1GL) are the rawest form of programming: binary instructions directly executed by a processor. These are sequences of bits that tell the machine exactly what to do, down to the level of moving data between memory and registers, performing arithmetic, or jumping to…
no project is complete until there are widgets
I do not have good automation of "updating the NGINX config". I do this maybe once every 3 months, so it is fine.
greenland, a post-mortem, part 3
JSON output is almost a necessity for an LLM to be usable today. All of the major LLM platforms have it in some form. But, if you are using a model from 2023, it might not support it, or it might not work very well.
greenland, a post-mortem, part 2
That link is https://spaceship.computer/greenland/ .
greenland, a post-mortem, part 1
Other than the << fit and finish , the Greenland project is done.
Trying the << qwen3-1.7b-q6_k.gguf model QWEN3, 1.7B parameters, quantized to 6-bit, there is a clear problem of runaway-thought (or at least looping responses).
more "fake thought" at the margins
QWEN3 is out. I'm starting tests with the 4B model.
The garden is mostly planted at this point. well, actually, the tomatoes and peppers haven't been transplanted yet. but they are planted. just inside.