Charlieholtz

10 ms·

91.
LLM Boxing – Llama 70B-chat vs. GPT3.5 blind test(llmboxing.com)
4 points by Charlieholtz 3y ago|0 comments
92.
by Charlieholtz 3y ago
Whoops! Fixed now
93.
by Charlieholtz 3y ago
This is a great question. My best answer is that there's a speed/intelligence trade-off. The smaller weights (7B) will run faster and require less memory, but you won't get the same quality responses as the 13B / 70B mod
94.
by Charlieholtz 3y ago
Thanks for pointing this out — we should've pointed out the script needs to be run on WSL. I added a note to the post to clarify (I work at Replicate). Also, you don't need a GPU to run this script! It builds and runs llama.cpp h
95.
by Charlieholtz 3y ago
There’s been a lot of talk comparing Llama 2 and GPT-3.5. So, I made a site to let the models fight it out. Each round, pick the output you like better. All open-source, and built with Elixir/Phoenix LiveView: https://github
96.
Show HN: LLM Boxing – GPT 3.5 vs. Llama 2(llmboxing.com)
1 points by Charlieholtz 3y ago|1 comments
97.
Fine-tune Llama 2 in a few lines of code(replicate.com)
2 points by Charlieholtz 3y ago|0 comments
98.
Fine tune Llama 2 (7B) with an API(replicate.com)
1 points by Charlieholtz 3y ago|0 comments
99.
Llama-2 70B parameter model(replicate.com)
2 points by Charlieholtz 3y ago|0 comments
100.
by Charlieholtz 3y ago
This is really exciting. I work at Replicate, where we've already setup a hosted version for anyone to try it: https://replicate.com/a16z-infra/llama13b-v2-chat
101.
Teach-O-Matic: Make how-to videos with Langchain and Replicate(teachomatic.net)
1 points by Charlieholtz 3y ago|0 comments
102.
by Charlieholtz 3y ago
Haha, thanks! Yes the outputs can be bizarre. But it's very impressive how fast text-to-video is evolving. Imagine text-to-video a year from now!
103.
Making movie trailers with Zeroscope, ElevenLabs, and GPT-4(charlieholtz.com)
30 points by Charlieholtz 3y ago|2 comments
104.
Show HN: Make cool QR codes with multi-controlnet(github.com)
4 points by Charlieholtz 3y ago|0 comments
105.
Teach-O-Matic: AI generated how-to videos with Replicate and LangChain(teachomatic.net)
1 points by Charlieholtz 3y ago|0 comments
106.
Show HN: Zoo. A free, open source tool for playing with AI image models(zoo.replicate.dev)
2 points by Charlieholtz 3y ago|0 comments
107.
Show HN: I made a free, open-source playground for AI image models(zoo.replicate.dev)
1 points by Charlieholtz 3y ago|0 comments
108.
Show HN: Zoo – a playground for comparing AI image models(zoo.replicate.dev)
3 points by Charlieholtz 3y ago|0 comments
109.
Show HN: Jupyter notebook for making how-to videos with Replicate and LangChain(colab.research.google.com)
2 points by Charlieholtz 3y ago|0 comments
110.
by Charlieholtz 3y ago
Should be fixed now!
111.
by Charlieholtz 3y ago
Ah shoot! Fixing now. I think the redirect went to an old version of the notebook
112.
Show HN: A Jupyter notebook for creating how-to videos with GPT4 and LangChain(teachomatic.net)
4 points by Charlieholtz 3y ago|6 comments
113.
Show HN: Teach-O-Matic – create instructional videos with GPT4 and Replicate(colab.research.google.com)
1 points by Charlieholtz 3y ago|0 comments
114.
by Charlieholtz 4y ago
woah these are great
115.
by Charlieholtz 4y ago
interesting! I got it to work but I wonder if the "bottoms" threw things off
116.
by Charlieholtz 4y ago
ahah yeah that is terrifying. thanks :)
117.
TileMaker – Create seamless tiled images with material diffusion(tilemaker.app)
38 points by Charlieholtz 4y ago|14 comments
118.
Show HN: Tic Tac Toe but with blind auctions(bidtactoe.com)
2 points by Charlieholtz 4y ago|0 comments
119.
Show HN: BidTacToe – tic tac toe with a twist(bidtactoe.com)
1 points by Charlieholtz 4y ago|0 comments
120.
When Robots Outwrite Humans, Maybe English Class Will Suck Less(elbow.substack.com)
1 points by Charlieholtz 4y ago|0 comments

More ›