EDINET-Bench: Evaluating LLMs on Complex Financial Tasks using Japanese Financial Statements Paper: https://
pub.sakana.ai/edinet-bench/ We released a Japanese financial benchmark on @Huggingface
, designed to evaluate the performance of LLMs on financial tasks like fraud detection in Japan.
LLMS
-
EDINET-Bench: Evaluating LLMs on Japanese Financial Tasks
By
–
-
LLMs Automate Research Tasks: Researcher Reflects on Thirty Years
By
–
I too have been mesmerised at the power of LLMs to write proofs, bib files, code, literature reviews, slides, etc. I feel I wasted 30 years of research at Wits, Cambridge, Berkeley, UBC, Oxford and DeepMind doing things that could have been easily automated. Life would have
-
Expanding X AI Agent Capabilities and Autonomy
By
–
Unlimited context across recent X content (more than API as a tool as it is currently) X AI agent capabilities – Grok should be able to run an X account with minimum supervision
-
LLM Pixel Art Generation App piClash Demonstration
By
–
How good are models at Pixel art?
— Pietro Schirano (@skirano) 8 juin 2025
I built an app to find out, piClash, all in @Magicpathai using @OpenRouterAI.
It's crazy to see how creative LLMs have become, each drawing is done via function calling.
Try the app too 👇 pic.twitter.com/tXTDmwkfXSHow good are models at Pixel art? I built an app to find out, piClash, all in @Magicpathai using @openrouter
. It's crazy to see how creative LLMs have become, each drawing is done via function calling. Try the app too -
Trick to write naturally like a human
By
–
— Steal my prompt to make any AI write naturally like a human (just insert these rules into your prompts): — HUMAN WRITER PROMPT — [ INSERT HERE ANY OTHER PROMPT WITH THE SPECIFIED GOAL AND CONTEXT ] ## NATURAL WRITING —
-
FastHTML Documentation on Large Language Models
By
–
Not a list, but here's an example that matters a lot to me personally: https://
fastht.ml/docs/llms.txt -
LLM User Agent Compatibility and Integration Options
By
–
Any LLM that a user chooses to use it with.
-
Focusing on Basics and Intermediate Prompting Concepts
By
–
This time I'm tackling basics/intermediate concepts first. Not just the advanced stuff that many know me for. I've learned a lot in the past few years, having spent thousands of hours experimenting with prompting at this point.
-
Thinking helps with medium tasks but fails on hard puzzles
By
–
For medium tasks, the first answer is often wrong, so thinking further helps it get to the correct answer. With hard tasks (or puzzles in the case of this research), it never manages to figure it out regardless.
-
Reasoning Models Overthinking Easy Tasks
By
–
When a thinking/reasoning model encounters an easy task, something funny happens. The model finds correct answer/plan early, but then it keeps thinking and sometimes talks itself out of it. Overthinking, lol. It's trained on human data, after all.