Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

On Mon, Jul 24, 2023, 3:25 PM wrote: > i seen this, was going to tell Matt on here, thought nah for some reason, > but seems to not be a big deal I guess IDK? The Gzip thing above is not > actual AI right? > Gzip compresses text to 3 bpc and humans to about 1 bpc, so no. The prediction model for

Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

On Mon, Jul 24, 2023, 1:47 PM James Bowery wrote: > > > On Mon, Jul 24, 2023 at 11:46 AM Matt Mahoney > wrote: > >> >> Well I did warn you that AGI would be expensive >> > > Not to harp on the IS vs OUGHT distinction but if one factors the decision > half to individuals, we're left with appr

Re: [agi] Re: The extinction tournament

About what test you would propose for AGI or ASI. On Sun, Jul 23, 2023, 11:25 PM wrote: > HmmLet me think. > I know of 4 ways that come to mind how to test for AGI: > > Prediction Score, such as Perplexity or Lossless Compression. > If it sounds and draws and sees etc like us. > How much mon

Re: [agi] Re: Saurabh Kumar's fast-cmix wins €5187 Hutter Prize Award!

NoOn Mon, Jul 24, 2023, 8:11 AM stefan.reich.maker.of.eye via AGI < agi@agi.topicbox.com> wrote: > If compression is superior to language models, when are we getting a > compression based chat bot? :) > Compression is a way to test language models. A model predicts the next token. A compressor pr

[agi] Re: I got GPT-4 vision

Balanced Mode no work it searches only usually Creative mdoe turned out to be mostly best. -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T39dac2ecbb39d535-M06761da0e434b7bb93eb53ec Delivery options: https://agi.

[agi] Re: I got GPT-4 vision

Someone I know's test is just wow wtf: https://ibb.co/J5Z6DN8 also one from me: https://ibb.co/BC37KT2 -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T39dac2ecbb39d535-M4f4a16c16e6c21b1381bc1d4 Delivery options:

[agi] Re: I got GPT-4 vision

I tried the robot test in my kitchen to see if it can make me a coffee or hotdog or go to another room say, it thinks the exit of the kitchen is there maybe, sometimes it didn't mention it. It is clearly there. It was still incredible though, it sees it went way to close to the 3 blinds, see

[agi] I got GPT-4 vision

https://www.reddit.com/r/GPT4/comments/156n5xv/funny_and_deep_vision_and_chat_with_gpt4_bing_ai/ -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T39dac2ecbb39d535-Ma8291b0db050de2708ad8b9d Delivery options: https:/

[agi] Re: Saurabh Kumar's fast-cmix wins €5187 Hutter Prize Award!

It's always exciting to see the score drop. I like seeing this happen. -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T919d1153741947b5-M779d8090e944b48c1b043193 Delivery options: https://agi.topicbox.com/groups/a

[agi] Re: Saurabh Kumar's fast-cmix wins €5187 Hutter Prize Award!

On Monday, July 24, 2023, at 8:09 AM, stefan.reich.maker.of.eye wrote: > If compression is superior to language models, when are we getting a > compression based chat bot? :) It's only a way to evaluate AI's prediction score. It doesn't change how you, or what you, or if you, or why you, or what

[agi] Re: Saurabh Kumar's fast-cmix wins €5187 Hutter Prize Award!

Congrats. -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T919d1153741947b5-Mc808b440166265c5b947762c Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

And doesn't scale right? -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T4dbad1e5c8d7f685-Med08561ebe997a38a1179207 Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

i seen this, was going to tell Matt on here, thought nah for some reason, but seems to not be a big deal I guess IDK? The Gzip thing above is not actual AI right? -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T

Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

On Mon, Jul 24, 2023 at 11:46 AM Matt Mahoney wrote: > On Sun, Jul 23, 2023, 8:16 PM James Bowery wrote: > >> https://aclanthology.org/2023.findings-acl.426.pdf >> > > Yes, and I think the only reason gzip didn't outperform the other text > classifiers on the largest data sets is that it only fi

Re: [agi] “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

On Sun, Jul 23, 2023, 8:16 PM James Bowery wrote: > https://aclanthology.org/2023.findings-acl.426.pdf > Yes, and I think the only reason gzip didn't outperform the other text classifiers on the largest data sets is that it only finds matching strings over a 32 KB window. Of course, text classi

[agi] Dilated Attention: Too good to be true?

LONGNET: Scaling Transformers to 1,000,000,000 Token In this work, we successfully scale the sequence length to 1 billion > tokens. Our solution is LONGNET, which replaces the attention of vanilla > Transformers with a novel component named dilated attention.

[agi] Re: Saurabh Kumar's fast-cmix wins €5187 Hutter Prize Award!

If compression is superior to language models, when are we getting a compression based chat bot? :) -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T919d1153741947b5-M150f6f8c8ff4769359d3772c Delivery options: htt

Re: [agi] Visualizing Quantum Circuit Probability: Estimating Quantum State Complexity for Quantum Program Synthesis

On Sunday, July 23, 2023, at 12:33 PM, Matt Mahoney wrote: > Right now we have we have working 10 qubit quantum computers that can factor > 48 bit numbers using a different algorithm called QAOA, a huge leap over > Shor's algorithm. They estimate RSA-2048 can be broken with 372 qubits and a > fe