#thecompaniesapi roasting the marketing plan & discuss the next steps with @LeCoupa

#thecompaniesapi port back my ai wrapper improvements on our existing llama stack and prepare fine tuning for phi3

#thecompaniesapi port back my ai wrapper improvements on our existing llama stack and prepare fine tuning for phi3

#thecompaniesapi start working on fine tuning as phi3 cannot stop hallucinating even though its a lot faster, heard it would possibly change everything for our usecase so its worth trying

#thecompaniesapi improve the dx of our AI flows so we have faster feedback loop in development, noticed it was making me lazy that i had to wait for app rebuild to rerun everything

#thecompaniesapi convert and start trying ocr model to not only read but also extract data from website visuals

#thecompaniesapi phi iterations again ; Crisp release party :)

#thecompaniesapi continue work on phi, lots of iterations, kinda exhausted lately

#thecompaniesapi continue on phi3; hell of a refactor

#thecompaniesapi work on phi3 again :) continue my search for the best prompt

#life get a PR merged on bsky.app :)

#thecompaniesapi watch the data being extracted for our customer ; switch back to phi3 branch again

#thecompaniesapi start a new customer dataset extraction, also tried to run our new LLM on a way smaller GPU with success :)

#life joined bsky.app which feels less like a mind vacuum app than x so far

#thecompaniesapi lunch with @LeCoupa and try new prompt techniques with phi

#thecompaniesapi finalize subpages & sitemap extraction, now our bot can filter out sitemap to only visit pages that contains interesting data for our schema

#thecompaniesapi quantize phi3.5 to 4bit to use it in our inference server; same model size but 128k context length instead of 4k, I can now process huge chunks of texts without relying on batching

#thecompaniesapi continue on phi3; tryin to create the best possible synthesis of a web page at a low cost

#thecompaniesapi continue r&d on phi3, trying idea refinement from structured text content, then pipe to inference flow, unknown territories 🤣

#thecompaniesapi improving results from phi3 is a tedious task, but if it pays off we get rid of a ton of problems