Very likely your synthetic data for LLM prompt evals is terrible because it grossly over-estimates people's grammar, ability to provide context, and general coherence. Production queries are extremely bad.

Iran? The world leader in civilian nuclear power generation?

Do we have to add Iran to the pride flag now?

I'm working on AI memory == I'm working on saving and retrieving data from a database

When Gemini code assist *gets* your library and executes multi-file changes to update your DAG from very vague instructions 😍🥰☺️ junjo

Gemini Code Assist has gotten significantly better in the past month. Codebase contextual awareness and multi-file agentic work is way ahead of Github Copilot.

people are mad at this, but don't realize how a corporation is responsible for systematically optimizing outrage event turnout and mobilization for shareholder returns. Not exactly organic grass roots.

People keep accusing my profile of being fake because it's not anon

When you ask Gemini what Reddit says is the best headset, and it says pink catgirl headset from temu

Grok is really slow I think the generators are running out of diesel.

Does Iran have a weapon of mass destruction?

predicting human irrelevancy is the latest mental health crisis

asking every spammer and cold caller to star my GitHub repo

instead of a draft for infantry, it should be a draft for killer robot and drone manufacturing school.

you can replace pretty much any vegetable with meat and it's really good.

> Open discord on mac > Installing update 1 of 1 > Disappears for literally 30 seconds > Trying to open again restarts update process > Suddenly installs more updates > Finally opens

where is your JIRA ticket

im trying real scrum this time

i need a third monitor dedicated to monitoring the situation

The problem with making a library for python developers is that none of them know how to code.