gtwomd

If you wanna try them all from A to Z: https://beza1e1.tuxen.de/one_letter_proglangs.html

I wonder what DAN[0] has to say about Ron Desantis?

[0] https://gist.github.com/coolaj86/6f4f7b30129b0251f61fa7baaa8...

Has anyone tried quantizing some metric of brain activity (EEG etc.) into a set of tokens, and then training a Transformer model on it?

I wonder if hidden in that is some good prior for human-cognition-related activities, i.e. extend the token space to add human language tokens, train on that and see if it trains significantly faster than a randomly initialized model.

I guess no one else is allowed to have indirect lighting anymore (/s)

As someone not well versed in how startups work, why do they need to acquire the entire company to have the team work for them, as opposed to negotiating with the employees individually? Is it essentially a lump sum payment to stop what they were currently working on.

Tangentially related:

I hear a lot about people talking about politics becoming ingrained in certain scientific fields, but it's hard for me to imagine this happening in certain fields. It is amusing to try to think how one might even begin to "ideologically subvert" a field like Condensed Matter Physics or Knot Theory.

Can you find anyone who would be upset about the Quantum Hall Effect or the Călugăreanu Theorem.

Sounds like he forgot to "space" the "X".

In any case I imagine that trying to create a commerce "everything" app like WeChat in the US will cause trouble with Apple and Google, even Epic Games w/ all their Fortnite users couldn't dodge the 30% fee.

Late to the party, here is some code to solve the path between each word, just requires numpy and a txt file of 4 letter words "words.txt" separated by lines

    import numpy as np
    with open("words.txt") as f:
        words = f.read().splitlines()
        f.seek(0)
        chars = np.array(list(ord(c) for c in f.read() if c != '\n')).reshape((-1, 4))

    word_to_index = dict(zip(words, range(len(words))))

    char_flip = np.sum(chars[:, None, :] != chars[None, :, :], axis=2) == 1
    chars_sorted = np.sort(chars, axis=1)
    char_shuffle = np.all(chars_sorted[:, None, :] == chars_sorted[None, :, :], axis=2)

    adj = char_flip | char_shuffle
    np.fill_diagonal(adj, False)
    # import itertools

    def path_recursive(current, end, adj, path, depth):
        if depth <= 0:
            return []
        if current == end:
            return [[*path, end]]
        
        # all_paths = []
        for neighbor in adj[current]:
            if neighbor in path:
                continue
            _path = path_recursive(neighbor, end, adj, [*path, current], depth - 1)
            # all_paths.append(_path)
            if _path:
                return _path

        
        # return list(itertools.chain(*all_paths))
        return []

    def binary_shortest_dist(start, end, adj):
        depth = 1
        adj_n = np.eye(adj.shape[0])
        while depth < 20:
            depth += 1
            adj_n = adj_n @ adj
            if adj_n[start, end]:
                return depth


    def path(start, end, adj):
        start = word_to_index[start.upper()]
        end = word_to_index[end.upper()]
        depth = binary_shortest_dist(start, end, adj)

        print(depth)

        adj = [[i for i in np.nonzero(_adj)[0]] for _adj in adj]

        paths = path_recursive(start, end, adj, [], depth)
        return [[words[i] for i in path] for path in paths]

    path("disc", "zero", adj)

Only a matter of time until Ryan Gosling shows up in someone's X Ray[0]

[0] https://petapixel.com/2020/08/17/gigapixel-ai-accidentally-a...

I was looking for "NeRF" in here but reading the paper it seems like they don't use neural networks to represent anything.

They seem to just optimize the positions and shapes of gaussian primitives as well as the reflectance properties.

Certainly a lot more "explainable" than a NeRF.

+1