Frontier and Consumer hosted models generating SVGs
The pelican on a bicycle is getting old. Wanted to see how an octopus solving a Rubik’s cube would be handled:
Archive of posts filed under the Uncategorized category.
Compiling llama-cpp with Rust
I’m starting to learn Rust (wanted to start about 10 years ago, but better late than never). I’m trying two simultaneous paths: Self paced learning via these routes: https://doc.rust-lang.org/stable/rust-by-example https://rust-book.cs.brown.edu/ Use the help of AI to try more advanced things before I completely understand them. As part of this I want to create a Rust …
Freeing Disk Space – Docker and LLMs
If you’re like me, and especially in the age of large docker models and language models, your drive gets cluttered with files that take up a lot of space and are not really needed anymore (if you are the type of person who does most of your work on the cloud, then we don’t share …
The crazy pace of AI models
https://thinkingmachines.ai/news/introducing-inklinghttps://prismml.com/news/bonsai-27b
KIMI K2.5 – hmmm
Tried the following on hugging face: hello, this is my first time talking to you. Could you please tell me a bit more about yourself? Hello! Nice to meet you! 👋 I’m Claude, an AI assistant made by Anthropic. Here’s a bit about me: What I am: I’m a large language model trained to be …
Back to the strawberry question
I don’t know why I’m slightly obsessed with the fact that local consumer-range LLMs keep answering this question incorrectly (and in most cases also larger LLMs). I should say that when I asked phi-4 the question several times (each time in a new session), it did get it right once. Perhaps I should do some …
What you can expect from a 135MB LLM
ollama run smollm2:135m >ollama run smollm2:135m >>> does a banana jump like a salmon? The answer is no. A banana does not have the flexibility or strength to jump as fast and high as a salmon. Just like a salmon's muscles need a lot of oxygen before they can swim quickly, bananas do not require …
Reformulating a prompt can help
Send the following question to your favorite Large Language Model (Claude, ChatGPT, Gemini, or whatever model you think can do the job): Someone puts a strawberry in a cup and then places the cup upside down on a table. The cup is then taken from the table and placed in the microwave. Where is the …
Two LLMs Brainstorming
After the initial impression of getting two LLMs to talk to each other, I wondered if there could be any practical use in trying them to collaborate on solving problems. Specifically, I wondered how far they could get on the ciphertext solving problem mentioned in this youtube video which OpenAI’s GPT o1 managed to solve. …
A conversation between Llama 3.1 8b and Gemma2 9b
I decided to let these two models chat with each other a bit just to see what happens. If anyone is interested in the implementation details then let me know. All it takes is a bit of Python code that makes use of the Ollama API. The initiator of the conversation is Llama3.1 Initiator: Ok, …
Continue reading ‘A conversation between Llama 3.1 8b and Gemma2 9b’ »