HackerTrans
TopNewTrendsCommentsPastAskShowJobs

gerlv

no profile record

Submissions

ScreenAI: A Vision-Language Model for UI and Infographics Understanding

arxiv.org
1 points·by gerlv·قبل سنتين·0 comments

An Interactive Agent Foundation Model

arxiv.org
2 points·by gerlv·قبل سنتين·0 comments

Universal Android Debloater

github.com
1 points·by gerlv·قبل سنتين·0 comments

The Annotated Transformer (2022)

nlp.seas.harvard.edu
1 points·by gerlv·قبل 3 سنوات·0 comments

comments

gerlv
·قبل 4 أشهر·discuss
Building DynoWizard [1] - tool for designing single table DynamoDB tables.

I first used DynamoDB 8 years ago and have been designing single-table schemas heavily since. For me, the best way to create drafts was always pen and paper (and then excel/confluence tables), but in reality it's a process (based on The DynamoDB Book) that can be automated to an extent.

Decided to build an app while on paternity leave. You define entities and access patterns, create (or get suggested) key and GSI design, and generate code for access patterns (TypeScript and Python), infrastructure (CDK, CloudFormation, Terraform), and documentation you can share with stakeholders.

There's more I want to build beyond the MVP - things around understanding and validating designs that you can't get from a chatbot - but for now focusing on the core.

If anyone wants to try it out, sign up for the waitlist on the landing page. MVP should be ready in the next few weeks.

[1] https://www.dynowizard.com
gerlv
·السنة الماضية·discuss
1. Treat it as a stream - you can't catch all of it as there is so much info, just catch something interesting, read it, and don't stress out if you didn't catch something else. There will always be something else to read.

2. Most of the sources you've mentioned are push-based - i.e. someone else is pushing this new info onto you (newsletters, youtube, podcasts, news). This increases FOMO. Instead, try to implement a pull-based approach and only seek and read info that is relevant to what you want to learn, read. It's a lot harder than it seems, but my guess it's harder due to default.

Last year, I re-tried[1] the experiment of not using the internet for entertainment for a few months, only for work and life admin. To catch up with news, I subscribed to a paper-based weekly newspaper. If there is something important in the world, you will find out about it, someone will tell you. But this will help a ton with anxiety and mental health.

The other thing I realised - when I listen to podcast and go into info overload, I get burned out a lot quicker. Listening to podcasts while working is the worst. I removed all podcast subscriptions and only started adding those that I want to listen + limit when I listen to these episodes.

[1] https://oleggera.com/blog/life-with-no-internet/
gerlv
·قبل سنتين·discuss
This looks cool and I would definitely use it as at the moment manually type "Act as ..., this is the context, ... etc" to improve the responses.

Side note - this is a single page with a few paragraphs, why is it 1.4MB in size (300kb gzipped)? It's just insane size for the amount of functionality it provides.
gerlv
·قبل سنتين·discuss
For API responses that require valid JSON - you can make requests in JSON mode - https://platform.openai.com/docs/api-reference/chat/create#c...

Edit: url to API docs
gerlv
·قبل 3 سنوات·discuss
Google Wave, there were some features that were amazing like splitting messages in their own threads to branch off some ideas and keep conversations on track.