fsndz

9 ms·

61.
by fsndz 2y ago
Anthropic is back and cementing its place as the creator of the best coding models—bravo! With Claude Code, the goal is clearly to take a slice of Cursor and its competitors' market share. I expected this to happen eventually. The app
62.
by fsndz 2y ago
absolutely on point!
63.
by fsndz 2y ago
DeepSeek is just the gift that keeps on giving. I now agree with people who say open source AI will win: https://open.substack.com/pub/transitions/p/deepseek-is-comi...
64.
by fsndz 2y ago
lol with that kind of reasoning, nobody should use statistics or any kind of machine learning model...
65.
by fsndz 2y ago
I am starting to believe this is a lie spread by AI companies because if AI-slop starts to be detected at scale, it kills their primary use case. True, AI detection tools are not perfect, like any classification algo, they don't have a
66.
by fsndz 2y ago
you can inject personal stuff to make it feel original, but huge chunks are still AI-generated. Just get the first 4/5 paragraphs and paste in gptzero
67.
by fsndz 2y ago
Lol, by just reading I knew it was. Then I used an AI detection tool and it says 100% sure it is AI-generated. You know how hard it is to get 100% sure it is AI-generated ?
68.
by fsndz 2y ago
the article is AI-generated isn't it ?
69.
by fsndz 2y ago
True. If the U.S. loses the AI race, it won't be because of China. https://open.substack.com/pub/transitions/p/the-ai-enemy-fro...
70.
by fsndz 2y ago
True. That's the reason I see a lot of people dropping similar frameworks like LangChain recently: https://medium.com/thoughts-on-machine-learning/drop-langcha...
71.
by fsndz 2y ago
"You Will Never Win an Argument On the Internet"=> True, but participation is important because even if the person you are arguing with will rarely change their mind in the process, you might convince some bystanders. So for ve
72.
by fsndz 2y ago
I think a one year gap in adoption of new tech is not that bad. Isn't it better to always go for the mature tech first ? The real change will come from the fact that because of AI, compute will be so cheap in the coming years: https:&
73.
by fsndz 2y ago
Frankly, it is now clear that open source AI will win at this rate. I just see innovations after innovations on a weekly basis. Exciting times ahead. https://open.substack.com/pub/transitions/p/the-ai-enemy-fr
74.
by fsndz 2y ago
If the U.S. loses the AI race, it won't be because of China.
75.
The AI Enemy from Within(transitions.substack.com)
2 points by fsndz 2y ago|1 comments
76.
by fsndz 2y ago
yeah that's mostly hype to get the next funding round
77.
by fsndz 2y ago
to get the score they had on ARC AGI, they had to fine-tune o3 on ARC AGI. That is hardly a sign of general intelligence or emergent capability. PhD novel research ? What is the novel research discovered by an LLM ever since the emergence o
78.
by fsndz 2y ago
The chief hype officer is back at it again. Altman is defending exponential progress when everything points to incremental progress with signs of diminishing returns. Altman thinks benefits will naturally trickle down when everything points
79.
by fsndz 2y ago
exactly I was surprised as well
80.
by fsndz 2y ago
I think we will have to move with pre-training and post-training efforts in parallel. What DeepSeek showed is that you first need to have a strong enough pretrained model. For that, we have to continue the acquisition of high quality, multi
81.
by fsndz 2y ago
The secret is to basically use RL to create a model that will generate synthetic data. Then you use the synthetic dataset to fine-tune a pretrained model. The secret is basically synthetic data imo: https://medium.com/though
82.
by fsndz 2y ago
I think the secret of DeepSeek is basically using RL to train a model that will generate high quality synthetic data. You then use the synthetic dataset to fine-tune a pretrained model and the result is just amazing: https://open
83.
by fsndz 2y ago
really good use case there haha
84.
by fsndz 2y ago
possible to use it with R1 for the reasoning part ?
85.
by fsndz 2y ago
This is mainly to reclaim mindshare from DeepSeek that has done incredible launches recently. R1 was particularly a strong demonstration of what cracked team of former quants can do. The demo of Operator was nice but I still feel like R1 is
86.
by fsndz 2y ago
What do you prefer ? Letting DeepSeek and China lead the AI war ? DeepSeek R1 is a big wake up call https://open.substack.com/pub/transitions/p/deepseek-is-comi...
87.
by fsndz 2y ago
true. I think simulations will help a lot in that direction. Imagine if you can do RL a bit like DeepSeek for R1 but on corporate tasks. https://open.substack.com/pub/transitions/p/deepseek-is-comi...
88.
by fsndz 2y ago
Damn, one year at google and the author was basically set for life.
89.
by fsndz 2y ago
emphasis on corporate
90.
by fsndz 2y ago
this ! and the truth is is there that much corporate domains without "clear success metrics" ?

More ›