yunusabd

Now imagine what you could do in 6 minutes!

But honestly I really like the short turnaround times. Makes it easy to experiment with different parameters and develop an intuition for what they do.

Sorry, just saw this.

I absolutely agree, but it's really stubborn with the flowery language. I tried adding things like "DO NOT USE EMPTY PHRASES LIKE 'EVER-EVOLVING TECH LANDSCAPE'!!!!!" to the prompt, but it just can't resist.

I want to give the whole system an overhaul, maybe newer models are better at this. Or maybe a second LLM pass to de-flowerize (lol) the language.

GP asked the model to _create_ a riddle, not solve a given one.

Impressive, might use this for https://hnup.date

Very nice!

> I have a mechanism to quickly delete problem submissions.

Did you build a male genitalia swastika classifier like the fish guy? (What a sentence)

https://news.ycombinator.com/item?id=44719222

One of those articles that you'd love to share with certain people but it seems awkward when they receive a message from you with the link preview saying "Face it: you're a crazy person".

I'm building an app for language learning with Youtube. I realized that yt probably has the largest collection of spoken language that ever existed, so I wanted to make it accessible, especially on mobile.

I'm focusing on Chinese (Mandarin) right now, because that's what I've been learning, and the language learning community on reddit likes it too. But other languages are also available.

Link: https://lingolingo.app

I mean yeah, but also

Doctor: do this

Patient: I tried doing this and it's not good

Doctor: actually you need a device for $5000 lol

I tried it on a M1 Pro MBP using Docker. It's quite slow (no MPS) and there are no timestamps in the resulting transcript. But the basics are there. Truncated output:

  Fetching video metadata...
  Downloading from YouTube...
  Generating transcript using medium model...

  === System Information ===
  CPU Cores: 10
  CPU Threads: 10
  Memory: 15.8GB
  PyTorch version: 2.7.1+cpu
  PyTorch CUDA available: False
  MPS available: False
  MPS built: False
  
  Falling back to CPU only
  Model stored in: /home/app/.cache/whisper
  Loading medium model into CPU...
  100%|| 1.42G/1.42G [02:05<00:00, 12.2MiB/s]
  Model loaded, transcribing...
  Model size: 1457.2MB
  Transcription completed in 468.70 seconds
  === Video Metadata ===
  Title: 厨师长教你：“酱油炒饭”的家常做法，里面满满的小技巧，包你学会炒饭的最香做法，粒粒分明！
  Channel: Chef Wang 美食作家王刚
  Upload Date: 20190918
  Duration: 5:41
  URL: https://www.youtube.com/watch?v=1Q-5eIBfBDQ
  === Transcript ===
  
  哈喽大家好我是王刚本期视频我跟大家分享...

Did you build this? I'm looking for an API that does this.

So you're saying that they should analyze both audio and video to increase the quality of the captions, if the video has hard-coded captions? I guess that's possible, just a question of effort vs. payoff.

Inaccurate auto-captions for videos with hard coded captions probably isn't a big enough pain to warrant big investments?

I'm working on an app that's based around youtube videos for language learning. I had to solve the same problem of youtube automatically changing the audio track to match the device locale.

Even thought about making a spin off app with only the no-translate feature, that simply always uses the original title and audio. I guess revanced can do this too, but maybe there's enough people who don't use revanced, or don't know about this feature. Thoughts?

Okay, but it looks like in the paper, they are actually adding the question twice in the prompt, not just instructing the model to read it twice. Or am I missing something?

That's an interesting space to explore! I'm wondering about the baseline in the benchmarks. Which prompts did you use for those? I'm asking because some of the resulting prompts seem fairly generic, and I'm wondering if you could just blanket add them to each prompt and also see an improvement. Things like "Identify the question (what are you trying to find?)".

In the same vein, wouldn't it be interesting to measure which part of the prompt most contributed to better solving the problem? Surely some parts will be just noise and can be trimmed away.

Also wondering what this does, since the model probably won't (can't?) actually read the problem multiple times:

  > Read the problem carefully (multiple times).

I was curious if the sensor would pick up other things like trees or other cyclist, but it seems like they accounted for that:

> We then log a sensor events [sic] if the majority of cells in the sensor frame agree to the same value within a threshold parameter [...]. This ensures that sensor events are only logged when large objects like cars block the sensor’s field-of-view , i.e., one or more small objects like branches or distance pedestrians in the sensor’s field-of-view will not trigger this condition. While there is no guarantee that this approach strictly identifies cars, we empirically saw during testing that passing cyclists and pedestrians rarely satisfied this condition at the typical passing distance due to the wide field-of-view of the VL53L8.

Also interesting that it's quite cheap to build:

> The whole system can cost less than $25 [...]

From the paper https://dl.acm.org/doi/10.1145/3706598.3713325

True, easy for a human, not so easy for a robot to go through those extra steps. I wonder if they made it work with the robots, because in the video they only show the robots building from the bottom up.

Cool project, but judging from the videos, it looks like some of them can't actually be built using those instructions. E.g. "A backless bench with armrest" would require some bricks to float in the air with no support while you're assembling the rest.

Demo of the pronunciation feature (currently only in the extension): https://youtu.be/d42i4httuao

I was wondering how it went for the author. Didn't see any mention of traffic in the article. Probably a lot of open roads, but at least in the populated areas traffic must have been a factor?

Your brother learned to read at the age of 4, but learning it at 7 1/2 is also fantastic!

Such a well-tuned machine, there has to be some grease somewhere.. Otoh, there's a lot more shady things going on everywhere in business.

If anything, I'm envious that _I_ don't have access to a system like that (only half joking)

Saw this on HN a while ago [1], really eye-opening: https://www.calcalistech.com/ctechnews/article/b1a1jn00hc

> The first sales come from the loyal CISOs who work with the fund.

> This "loyalty program" - which encourages deepening the relationship between the CISO and a party other than his employer - is seen by many in the industry as a red line crossed by Ra'anan and Cyberstarts.

> Cyberstarts vehemently denies [...] and claims that CISOs were never remunerated for purchasing the products of the portfolio companies.

[1] https://news.ycombinator.com/item?id=41042462

The author literally asked if the people were willing to take part in an experiment and clearly stated the goal. Doesn't sound very manipulative to me.

I'll look into it for the next iteration! I could just take the transcript that's already on the page and put it somewhere separate from the audio.

But thinking about it a little more, what would the use case for a text version actually look like? I feel like if you're already on HN, navigating somewhere else to get a TLDR would be too much friction. Or are we talking RSS/blog type delivery?

Same, it was using the high quality openai voice until my account ran out of funds.. Now it's using edge-tts which is free. So far it seems like the best option in terms of price/performance, but I'm happy to switch it up if something better comes along.

https://hnup.date/ ;)

In my experience, our current AIs work best when heavily grounded with real data. Maybe try and work with actual footage, reconstructions, or existing simulations as an input to make the videos more realistic

Hi, I wanted to try this, but account creation doesn't work for me. With GitHub it gives me "invalid auth code" and with email PW it gives me this, even though my password fulfills all these and I tried many different ones

"Password

Password Strength: Very Strong

Password must contain at least one uppercase letter, one lowercase letter, one number, and one special character"

Edit: Now I received an email that redirects to "http://localhost:3000/"

> For many reasons, that's just too big, we have folks in Europe that can't even clone the repo due to it's size.

What's up with folks in Europe that they can't clone a big repo, but others can? Also it sounds like they still won't be able to clone, until the change is implemented on the server side?

> This meant we were in many occasions just pushing the entire file again and again, which could be 10s of MBs per file in some cases, and you can imagine in a repo

The sentence seems to be cut off.

Also, the gifs are incredibly distracting while trying to read the article, and they are there even in reader mode.

This is perfect. If I was on the team, I would have this framed and hang it up in the office, to get a good chuckle every time I walk by.