3 ms·
Maybe not 10x yet, but deepcoder has done some impressive things recently. Instead of a generic LLM, they have a relatively smaller one which is coding specific
by viraptor 2y ago
Maybe not 10x yet, but deepcoder has done some impressive things recently. Instead of a generic LLM, they have a relatively smaller one which is coding specific and gpt4-class in quality. This makes it cheaper. In addition, they can do caching which ~10x reduces the cost of follow-up request. And there are still improvements around Star, which reduces the need for learning datasets (models can self-reflect and improve without additional data)
So while we're not 10x-ing everything, it's not like there's no significant improvements in many places.
- viraptor 2y agoI meant deepseek coder. Can't edit anymore.
- tmikaeld 2y agoUnfortunately the smaller model is not anywhere near GPT4 in quality and no one seems to want to host the bigger model (it was even removed from fireworks ai this week). And no one in their right mind want to send their code to deepmind chinese API hosting.
- viraptor 2y agoI'm perfectly fine sending my open source code to them. I'm also happy to send 95% of my private repos. Let's be honest, it's just boilerplate code not doing anything fancy, just routing/validating data for the remaining 5%. Nobody cares about that and it's exactly why I want AI to handle it. But I wouldn't send that remaining 5% to OpenAI either.