LocalLLaMA

3 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 1 year ago

MODERATORS

communick@poweruser.forum

So what's the story with Phind CodeLlama? (alien.top)

submitted 1 year ago by SomeOddCodeGuy@alien.top to c/localllama@poweruser.forum

4 comments fedilink hide all child comments

I keep seeing people posting about how the new Phind is the most amazing thing on the planet, and I kept thinking "We already have Phind... see, I have the gguf right here!"

I finally looked at the phind paper on their newest model, and it says that their current model is v7.

https://www.phind.com/blog/phind-model-beats-gpt4-fast

:O Huggingface only goes up to v2.

I can't tell if Phind is a proprietary model that just happened to give us an older version, if there will be newer versions coming out, or what. Does anyone happen to know?

top 4 comments

sorted by: hot top controversial new old

[–] Ilforte@alien.top 1 points 1 year ago

What are you surprised about? Yes, it's 5 versions younger, they're running a data flywheel, definitely scraping GPT-4 interactions when people use it through their website; v2 was just a shout-out to the community, to build goodwill and bring attention to their project. They've said on HN that they might release newer models – when they're no longer close to SoTA.

Thank you. Yes, it is the 7th iteration that started with our open-source models. We do plan to open source this model as well down the road, once we've released a few more generations.

It's business, not charity.

[–] m98789@alien.top 1 points 1 year ago (2 children)

V7 is propriety.

This is the current business model / strategy:

Release an open model that performs well
Let the community embrace it and get hype
Find investors and sell story of next OpenAI
Get huge valuation and raise a ton of cash
Go closed source due to “competitive reasons”
Cash out some founder equity on next round with “dumb money” investors (pyramid scheme)
Party!

[–] SomeOddCodeGuy@alien.top 1 points 1 year ago

Ok, this makes sense.

I think that something that had confused me for a long time, and maybe I was the only one, was everyone kept saying "I prefer Phind!" for coding and I legitimately thought that they meant the v2 on huggingface.

I always had better luck with CodeFuse-CodeLlama and couldn't figure out why no one ever talked about it and everyone just talked about Phind. Now I realize I was off in left field using v2 all by myself while everyone else was on a website using Phind v6 lol

I feel pretty silly now =D

[–] peterwu00@alien.top 1 points 1 year ago

I have a question regarding the meta licensing below. Does it imply that we can use the open-source Llama 2 model as a foundational model, train it with additional data, and retain the retrained model as proprietary?

"v. You will not use the Llama Materials or any output or results of the Llama Materials to improve any other large language model (excluding Llama 2 or derivative works thereof)."