bigEnotation

Try doing any kind of fantasy sports analysis and you’ll see the fallacy of the current state of LLMs.

Pinning y’all’s livelihood to this is wild.

Anyone have old Google Trends Newsletter Emails they could forward me?

I'm trying to build a model that embeds the content from the Google Trends Newsletter I've only recently signed up and I havn't been able to find any records from past emails, so was wondering if anyone would be willing to forward me copies from:

* between May 2022 and March 2023 * before March 2021

Example of what it’s being used for: https://github.com/edelauna/stock-predictor/pull/128

For a personal project I ended up using Github Actions to perform compute functions based on a cron schedule, and then commit the changes back to the repository: https://github.com/edelauna/stock-predictor/blob/prod/.github/workflows/cron-history.yml

As a solo contributor it seems to work ok, I was curious about potential drawbacks, and considerations I may want to consider if I wanted to start adding contributors, or merging dependabot PRs while a cron action was running heh.

I've created a Github action to provide code reviews as annotations to PRs [1], but it's providing some pretty odd feedback [2]. I'm not sure if it's my use of openai's functions[3], or my system prompt [4], or if gpt just isn't that good at isolated code reviews.

[1] https://github.com/marketplace/actions/chat-gpt-code-peer-review

[2] https://github.com/edelauna/discord-bot-ai/pull/74/files

[3] https://github.com/edelauna/gpt-review/blob/main/src/openai/utils/make-review.ts#L11

[4] https://github.com/edelauna/gpt-review/blob/main/src/openai/utils/message-manager.ts#L13