Hacker News Reader

ash-ishhon Live video feed for every multimodal model not just Gemini

For those interested, you can try VideoDB's real-time streaming notebooks here: https://github.com/video-db/videodb-cookbook/tree/main/real_...

ash-ishhon Man-Computer Symbiosis (1960)

https://karpathy.ai/blog/licklider1960.html

ash-ishhon 'Entropy bagels' and other complex structures emerge from simple rules

Reminded me of this recent observation of fractals in nn training https://sohl-dickstein.github.io/2024/02/12/fractal.html

ash-ishhon Show HN: Free image generation using Cloudflare Workers AI

Awesome!

It would be interesting to see how Cloudflare is going to charge with their Regular / Fast Twitch Neurons in future.

ash-ishhon Many RAGs for conversations, docs and PDFs but what about videos and audios?

Demo: https://publish.spext.co/video/Demo_video-mp4_a57b2d78

Sample Highlight Extraction Workflow On Audio: https://publish.spext.co/chat/New-York-Mets-vs-San-Francisco...

ash-ishhon Ask HN: Extracting Knowledge Graphs from LLMs

Checkout InstaGraph by Yohei: https://github.com/yoheinakajima/instagraph

Sample: https://twitter.com/yoheinakajima/status/1706848028014068118

ash-ishhon Ask HN: GPT-4 has 1.7T parameters. What's a parameter?

Checkout this tweet https://twitter.com/ylecun/status/1706545305762582580 by Yan LeCun.

"Parameters are coefficients inside the model that are adjusted by the training procedure. The dataset is what you train the model on. Language models are trained with tokens that are subword units (e.g. prefix, root, suffix)."

His comment on GPT-4 parameters count

"Also: a model with more parameters is not necessarily better. It's generally more expensive to run and requires more RAM than a single GPU card can have. GPT-4 is rumored to be a "mixture of experts", i.e. a neural net consisting of multiple specialized modules, only one of which is run on any particular prompt. So the effective number of parameters used at any one time is smaller than the total number."

ash-ishhon Show HN: ChatGPT vs. Jurassic vs. Bard vs. Claude

Interesting!

Here are some features that I will find useful.

1 - Option to customize the model parameters.

2 - Selective model comparison, once you extend the supported models it can be inefficient to run all the models all the time, selective execution can give more control to users to execute only required comparison models.

3 - Share option.

ash-ishhon Show HN: I build a dedicated news site for AI developers (think HN for AI/ML)

Nice! I see there is an option of communities to organize the posts, should there be a way for user to create or submit the name of community that is not present?

Also, I have a crawler which fetches blogs from OpenAI and DeepMind if there is a community for blogs I can plugin the code to post those articles there as well.