tkgally

10 ms·

121.
by tkgally 1y ago
> I’ve been using them forever. Many other HN contributors have, too. Here’s the pre-ChatGPT em dash leaderboard: https://www.gally.net/miscellaneous/hn-em-dash-user-leaderbo...
122.
by tkgally 1y ago
> LLMs don't learn from a project. How long do you think that will remain true? I've bootstrapped some workflows with Claude Code where it writes a markdown file at the end of each session for its own reference in later session
123.
by tkgally 1y ago
Thanks for the explanation. Very interesting. I notice that that particular tokenization deviates from the morphemic divisions in several cases, including ‘dec-entral-ization’, ‘食べ-させ-られた-くな-かった’, and ‘面白-くな-さ-そうだ.’ ‘dec’ and ‘entral’ are n
124.
by tkgally 1y ago
One reason for testing this is that it might indicate how accurately models can explain natural language grammar, especially for agglutinative and fusional languages, which form words by stringing morphemes together. When I tested ChatGPT a
125.
by tkgally 1y ago
With a $20/user/month tier, I thought this was in competition with similarly priced services from OpenAI, Anthropic, and Google. But then I saw this: “Accounts are charged a flat $250 per account per month infrastructure fee for u
126.
by tkgally 1y ago
Around the time GPT-4 was released in early 2023, a similar issue arose with another profession: translation. It was at that point that machine translation between languages like English and Japanese (the language pair I have worked with) s
127.
by tkgally 1y ago
I was thinking about consumer-facing AI products, where md files controlled by the user presumably wouldn’t fly. I find it annoying that, when prompting ChatGPT, Claude, Gemini, etc. on personal tasks through their chat interfaces, I have t
128.
by tkgally 1y ago
I started to suspect a few paragraphs in that this post was written with a lot of AI assistance, but I continued to read to the end because the content was interesting to me. Here's one point that resonated in particular: "There’s
129.
by tkgally 1y ago
That’s the problem, I think: Using AI will make some people stupider overall, it will make other people smarter overall, and it will make many people stupider in some ways and smarter in other ways. It would have been nice if the author had
130.
by tkgally 1y ago
The main reason I prefer email over messaging is that it allows me to correspond with the widest range of people. I work in academia and publishing and have various personal interests, and I want to be able to interact with people in many c
131.
by tkgally 1y ago
A couple of those articles, in case anyone is interested: “The Internet? Bah! Hype alert: Why cyberspace isn't, and will never be, nirvana” by Clifford Stoll (1995) Excerpt: “How about electronic publishing? Try reading a book on disc.
132.
by tkgally 1y ago
That’s how I use it. I’m not a developer, and using Claude Code with Git turned out to be more complicated than I wanted. Now I just give it access to a folder on my Mac, put my prompt and any associated files in that folder, and have it wo
133.
by tkgally 1y ago
I agree that traditional dictionaries are still very useful. I wouldn’t trust today’s LLMs, for example, for information about etymologies, pronunciations, alternate spellings, conjugations and declensions, etc., unless they had access to h
134.
by tkgally 1y ago
Like the author, I used to be an avid user of dictionaries. In fact, my interest in them led to me freelance part-time as a lexicographer for a number of years. And, like the author, I find online dictionaries and apps to be a mixed bag. Bu
135.
Replit Agent 3(replit.com)
1 points by tkgally 1y ago|0 comments
136.
Terence Tao: Research Powers Progress [video](youtube.com)
3 points by tkgally 1y ago|0 comments
137.
by tkgally 1y ago
I have had the same problem with artifacts, and I had similar problems several months ago with Claude Desktop. I stopped using those features mostly and use Claude Code instead. I don't like CC's terminal interface, but it has bee
138.
What It Costs to Be a Sorority Girl(theatlantic.com)
3 points by tkgally 1y ago|0 comments
139.
by tkgally 1y ago
That matches my experience, too. I passed JLPT N1—then called 1-kyū—back in 1985 (!). I did spend a lot of time memorizing vocabulary with flashcards, but I spent even more time on extensive reading—novels, newspapers, magazines, anything I
140.
by tkgally 1y ago
Anthropic just posted an interview with Boris Cherny, the creator of Claude Code. He also offers some ideas on how to use it. “The future of agentic coding with Claude Code” https://youtu.be/iF9iV4xponk
141.
by tkgally 1y ago
Due to the interest in this project, I created a second, more comprehensive version of the leaderboard: https://www.gally.net/miscellaneous/hn-em-dash-user-leaderbo... This second version was vibe-coded with Codex CLI.
142.
by tkgally 1y ago
As mentioned in the thread that included dang’s suggestion [1], examples of one’s use of em dashes timestamped before ChatGPT could be used as a defense if one is accused, on the basis of em dashes, of having written with AI. Whether this i
143.
by tkgally 1y ago
Done: https://news.ycombinator.com/item?id=45071722
144.
Show HN: Hacker News em dash user leaderboard pre-ChatGPT(gally.net)
377 points by tkgally 1y ago|266 comments
145.
by tkgally 1y ago
> LLMs ... lack the ability to hold a series of precepts "in mind" and stick to those precepts. That is perhaps the biggest weakness I've noticed lately, too. When I let Claude Code carry out long, complex tasks in YOLO mo
146.
by tkgally 1y ago
Ah, I am in good company!
147.
by tkgally 1y ago
A couple of months ago, riding the subway through Ginza Station for the first time in a while, I noticed that the door-closing melody was from the 1949 song Ginza Kankan Musume [1, 2]. I’m normally not very familiar with Japanese pop musi
148.
by tkgally 1y ago
Fortunately, em-dash users who have been posting to HN long enough can point to evidence of our pre-ChatGPT use: https://news.ycombinator.com/threads?id=tkgally&next=3380763...
149.
by tkgally 1y ago
That definitely deserves proverb status! Around the same time I was collecting those ghost proverbs, I spent a pleasant afternoon in Shinjuku, Tokyo, taking pictures of T-shirts: https://www.gally.net/tshirts/index.html
150.
by tkgally 1y ago
More than twenty years ago, I had fun tracing a similar phenomenon: English “proverbs” that appeared in English dictionaries and textbooks published in Japan but that did not seem to have any actual currency in English. It became clear that

More ›