5 ms·

This looks very fake to me. I might be wrong. Yet, there is no "document" that was leaked, the original source is some blog post. If there is a document, share

by vlaaad 3y ago

This looks very fake to me. I might be wrong. Yet, there is no "document" that was leaked, the original source is some blog post. If there is a document, share the document. Shared by "anonymous individual on discord who granted permission for republication"... I don't know. If it was shared by anonymous, why ask for permission? Which discord server?

simonw 3y ago
Did you read it? I honestly don't care if it's really a leak from inside Google or not: I think the analysis stands on its own. It's a genuinely insightful summary of the last few months of activity in open source models, and makes a very compelling argument as to the strategic impact those will have on the incumbent LLM providers. I don't think it's a leak though, purely because I have trouble imagining anyone writing something this good and deciding NOT to take credit for the analysis themselves.
- ftxbro 3y ago
  > It's a genuinely insightful summary of the last few months of activity in open source models Yes this is an amazing summary! Just for its summary alone, it is probably one of the top five writings I saw on LLMs and I read every one! > because I have trouble imagining anyone writing something this good and deciding NOT to take credit for the analysis themselves. Not everyone has a substack that they spam onto hacker news every time a thought enter their head. Or imagine that INTP exist. In my opinion the best take ever on LLMs is the simulators essay https://generative.ink/posts/simulators/ https://generative.ink/posts/simulators/ and the author is so shy they went pseudonymous and put their twitter as private and don't even want their name in conferences.
  - heliophobicdude 3y ago
    I think the another great paper is RLHF article from Chip Huyen https://huyenchip.com/2023/05/02/rlhf.html https://huyenchip.com/2023/05/02/rlhf.html
    - ftxbro 3y ago
      Thanks for sharing it but I'm sorry I don't agree that it's so important. In my opinion almost every interesting thing about LLMs comes from the raw base model, which is before the RLHF is applied. For example the simulators paper was written before ChatGPT was even released, based on research with the GPT-3 base model that only had text completion and no instruction tuning or any kind of RLHF or lobotomization. In another example, in the interviews with the people who had access to the base model of GPT-4 like the red teamers and the ones at microsoft who integrated it with bing, they consistently explain that the raw base pretrained model has the most raw intelligence which is deadened as they put RLHF and guardrails onto it.
16bitvoid 3y ago
Coder Radio podcast uploaded the document to the show notes for their latest episode[1]. The first link[2] in the PDF does link to some internal Google resource that requires an @google.com email address. 1: https://jblive.wufoo.com/cabinet/af096271-d358-4a25-aedf-e56dab169109 https://jblive.wufoo.com/cabinet/af096271-d358-4a25-aedf-e56... 2: http://goto.google.com/we-have-no-moat http://goto.google.com/we-have-no-moat
jsnell 3y ago
Presumably they weren't getting permission in the sense of "this publication is authorized by the original author, or by Google" but in the sense of "thanks for leaking the document; can we publish it more widely, or will you get into trouble?"

by vlaaad 3y ago

simonw 3y ago

Did you read it? I honestly don't care if it's really a leak from inside Google or not: I think the analysis stands on its own. It's a genuinely insightful summary of the last few months of activity in open source models, and makes a very compelling argument as to the strategic impact those will have on the incumbent LLM providers. I don't think it's a leak though, purely because I have trouble imagining anyone writing something this good and deciding NOT to take credit for the analysis themselves.

ftxbro 3y ago
> It's a genuinely insightful summary of the last few months of activity in open source models Yes this is an amazing summary! Just for its summary alone, it is probably one of the top five writings I saw on LLMs and I read every one! > because I have trouble imagining anyone writing something this good and deciding NOT to take credit for the analysis themselves. Not everyone has a substack that they spam onto hacker news every time a thought enter their head. Or imagine that INTP exist. In my opinion the best take ever on LLMs is the simulators essay https://generative.ink/posts/simulators/ https://generative.ink/posts/simulators/ and the author is so shy they went pseudonymous and put their twitter as private and don't even want their name in conferences.
- heliophobicdude 3y ago
  I think the another great paper is RLHF article from Chip Huyen https://huyenchip.com/2023/05/02/rlhf.html https://huyenchip.com/2023/05/02/rlhf.html
  - ftxbro 3y ago
    Thanks for sharing it but I'm sorry I don't agree that it's so important. In my opinion almost every interesting thing about LLMs comes from the raw base model, which is before the RLHF is applied. For example the simulators paper was written before ChatGPT was even released, based on research with the GPT-3 base model that only had text completion and no instruction tuning or any kind of RLHF or lobotomization. In another example, in the interviews with the people who had access to the base model of GPT-4 like the red teamers and the ones at microsoft who integrated it with bing, they consistently explain that the raw base pretrained model has the most raw intelligence which is deadened as they put RLHF and guardrails onto it.

16bitvoid 3y ago

Coder Radio podcast uploaded the document to the show notes for their latest episode[1]. The first link[2] in the PDF does link to some internal Google resource that requires an @google.com email address. 1: https://jblive.wufoo.com/cabinet/af096271-d358-4a25-aedf-e56dab169109 https://jblive.wufoo.com/cabinet/af096271-d358-4a25-aedf-e56... 2: http://goto.google.com/we-have-no-moat http://goto.google.com/we-have-no-moat

jsnell 3y ago

Presumably they weren't getting permission in the sense of "this publication is authorized by the original author, or by Google" but in the sense of "thanks for leaking the document; can we publish it more widely, or will you get into trouble?"