obvithrowaway34434

joined 1 year ago
sorted by: new top controversial old
Why is no one releasing 70b models? in c/localllama@poweruser.forum
[–] obvithrowaway34434@alien.top 1 points 11 months ago

Mistral has already shown that it's mostly about the data rather than the model. So why waste loads of money and time on training something that no average consumer can run locally?

Please enlighten me, why are people building LLM Twitter bots? in c/localllama@poweruser.forum
[–] obvithrowaway34434@alien.top 1 points 11 months ago (1 children)

That's an order of magnitude improvement from the average Twitter post in terms of grammar, composition, ability to hold a coherent thought for a few seconds etc. and most importantly does not make your blood boil with rage. Why are you complaining?

Many people here are probably getting started on machine learning, here's a blog post from Greg Brockman who recently left OpenAI with Sam Altman. I found this quite inspiring. in c/localllama@poweruser.forum
[–] obvithrowaway34434@alien.top 1 points 11 months ago

Look him up (he has a Wikipedia page) or ask ChatGPT. It will be much more easier than explaining in Reddit.

1
Many people here are probably getting started on machine learning, here's a blog post from Greg Brockman who recently left OpenAI with Sam Altman. I found this quite inspiring. (blog.gregbrockman.com)
1
Has anyone tried the OpenGPTs by Langchain? Is there any curated list of good opengpts? (alien.top)
 

For those who're not aware Langchain released OpenGPTs in response to GPTs by OpenAI. I'm curious about how this works with different OSS models. So far with GPT 3.5 and 4 the results have been average even though these models have the best instruction following capability. Is anyone tracking different opengpts being made?

Llama 3 will be released in the first quarter of 2024? in c/localllama@poweruser.forum
[–] obvithrowaway34434@alien.top 1 points 1 year ago (1 children)

I'm more interested in the next Mistral release, none of that corporate "safety" bs. It would also be good to have a truly open-source model (that releases both weight and the training data).

1
New Microsoft codediffusion paper suggests GPT-3.5 Turbo is only 20B, good news for open source models? (alien.top)
 

Wondering what everyone thinks in case this is true. It seems they're already beating all open source models including Llama-2 70B. Is this all due to data quality? Will Mistral be able to beat it next year?

Edit: Link to the paper -> https://arxiv.org/abs/2310.17680

https://preview.redd.it/kdk6fwr7vbxb1.png?width=605&format=png&auto=webp&s=21ac9936581d1376815d53e07e5b0adb739c3b06