Metricon

If you're working with images that need to be lossless or are simple in color diversity, WEBP is probably still a better choice (though not by a large margin). The biggest issue you might run into with AVIF is legacy support, particularly amongst Apple devices who were slower to implement support for it.

Otherwise the massive size reduction for storage and transfer speeds makes it incredibly attractive for modern web applications.

Ah, I actually wasn't aware that "avifenc" was a separate executable. Thanks for the information. I will definitely do some comparison testing.

These settings have worked very well for me so far, but if you are aware of even better settings, I'd certain like to give it a try.

AVIF is fantastic format and I use it in my production applications. Spent quite a bit of time looking to the best settings (at least for up to 2k images) and for anyone interested in converting images to AVIF, here's what I've found to work the best for creating the smallest size (10x+ smaller) while still retaining good quality and even supports PNG transparency:

  ffmpeg -i {input_file} -filter_complex "[0:v]format=pix_fmts=yuva444p,split[color][alpha];[alpha]alphaextract[alpha_chan]" -map "[color]" -map "[alpha_chan]" -c:v libaom-av1 -crf 25 -cpu-used 3 -still-picture 1 -aom-params "sharpness=2:enable-chroma-deltaq=1:arnr-strength=0" -color_range pc -colorspace bt709 -color_primaries bt709 -color_trc bt709 {output_file}

Now launched => Verse Draft: https://versedraft.com - An all-in-one writing studio where fiction writers can keep all the details for their universes in one place while crafting stories, novels, movie scripts, TV series, or stage plays.

Also created a fun simple "card game" activity for writers looking for story ideas or stuck with writers block: https://talemancy.com/

(This was somewhat inspired by memories of playing Ultima IV as a teenager)

Currently working on something a little more visually focused in creative design.

I use the "Comments Owl for Hacker News" Firefox extension - https://soitis.dev/comments-owl-for-hacker-news

It has a "Dark Mode" under options, plus it allows you to revisit a thread later and know which comments are new.

What you want to look for is called "Commercial Signage Display". These are basically large "TV" monitors that are sold to businesses like sports bars.

On Amazon, you can find a few here: https://www.amazon.com/gp/bestsellers/industrial/18746933011... (which are mostly Samsung).

Still in beta with Verse Draft: https://versedraft.com - An all-in-one writing studio where fiction writers can keep all the details for their universes in one place while crafting stories, novels, movie scripts, TV series, or stage plays.

One surprising finding is how few people are interested in real-time voice interaction (transcribing, dictating, conversing with assistant), so that's been removed for now.

It's got what plants crave. It's got electrolytes!

Keep in mind that unlike purchasing a stock where the most amount of money you can lose is the amount of money you spend buying the stock (assuming you didn't buy it on margin), if you directly short a stock, there's technically no limit to the amount of money you could lose. If a stock goes up 1000% after you short it, then you could lose far more money than you put into it.

This is why I mentioned the plateauing. If for some reason that just doesn't happen and things continue at a relatively linear or even exponential rate, then all bets are off.

I'd like to imagine that it's ultimately going to work out for humanity to be in a better condition overall if that happens.

However, if there reaches a level of near or full autonomy in all aspects of knowledge work, then there is a strong possibility that the field with the highest growth potential is going to be in "Private Security"; as those who have access to the most resources seek to defend their own positions in a societal return towards aristocracy and serfdom.

Let's hope not though.

I completely agree with you, and I certainly meant no offense to anyone who's a practicing nurse.

I spent some time in the military, and my expression of medics and nurses are mostly derived from that experience, where I'm referring to a nurse as just any warm body who is able to provide aid.

For professional nurses who might work in hospitals, I'm sure that many of them have significant knowledge and experience to be very effective in providing medical assistance.

I have been writing software for over 40 years and have had a long time interest in and some work in AI over that time. I wouldn’t say this gives me any more prognosticating power about how all of this is ultimately going to go, but I believe we're soon nearing an area of plateauing; whether that's because the science itself is plateauing or the intervention of governments is going to force plateau it.

So if things continue as they are today, I think in the near future, being a software developer is going to be more analogous to the medical field, where in the medical field you have different levels of professional expertise.

Some will be like nurses, and some will be closer to a medic and a smaller set will be like doctors. Each with increasingly required knowledge and experience to fulfill a needed role.

Those who used to be actual software developers are going to be (or have to become) more in the doctor role with years of internship and practical experience to be the architects guiding the overall AI implementation of software development in organizations.

The medics are going to be people who are semi-technical, where they have some technical understanding but they don't dedicate themselves to it, like say product managers, where they jump in to help development along, but don't need to have many years of experience or very deep technical knowledge.

At the nurse level, it's probably going to be similar to what people would do in the past with no code tools, where somebody in marketing who knows very little to nothing about coding at all is just going to directly converse with AI systems, but they'll never be likely to get anything more advanced than the tools they could think up for themselves.

Of course, it's so hard to tell what the next big discovery or changes to the nature of world society might push things in one direction or another.

I've just released in beta, a web-based application called Verse Draft: https://versedraft.com

I wanted to create an all-in-one writing studio where fiction writers can keep all the details for an entire universe in one place while crafting stories, novels, movie scripts, TV series, or stage plays.

I also wanted the ability to allow for the limited use of AI in a way that only functions as a sounding board and does not write for the user; Where fiction writers could have access to tools such as a virtual assistant that they can converse with about their stories and world-building, but without it writing anything for them.

There is also an option to use the application without any AI tools at all.

Wouldn't a subjectively boring picture present an even greater challenge to the creativity of the writer involved in the competition? It could have been a picture of a Barton Newman painting; that would be a significant challenge to structure a 1000 word essay on.

Another to add to this list would be Isaac Asimov. Most well known as a Science Fiction writer, he actual wrote over 500 books (mostly non-fiction) on a wide variety of subjects.

Many word processors (Microsoft Word, LibreOffice Writer, etc.) and some online editors will auto convert double hyphens "--" as they are typed into an em dash.

This might cover what you are looking for: https://www.khanacademy.org/

It seems like some things always remain the same: https://www.youtube.com/watch?v=G0ZZJXw4MTA

You may want to have a look at Mistral OCR: https://mistral.ai/news/mistral-ocr

If you run the `gguf_orpheus.py` file in that repository, it will capture the audio tokens and convert them to a .wav file. With a little more work, you can feed the streaming audio directly using `sounddevice` and `OutputStream`

On a Nvidia 4090, it's producing:

  prompt eval time =      17.93 ms /    24 tokens (    0.75 ms per token,  1338.39 tokens per second)

         eval time =    2382.95 ms /   421 tokens (    5.66 ms per token,   176.67 tokens per second)

        total time =    2400.89 ms /   445 tokens

*A Correction to the llama.cpp server command above, there are 29 layers so it should read "-ngl 29" to load all the layers to the GPU.

GGUF version created by "isaiahbjork" which is compatible with LM Studio and llama.cpp server at: https://github.com/isaiahbjork/orpheus-tts-local/

To run llama.cpp server: llama-server -m C:\orpheus-3b-0.1-ft-q4_k_m.gguf -c 8192 -ngl 28 --host 0.0.0.0 --port 1234 --cache-type-k q8_0 --cache-type-v q8_0 -fa --mlock

To be clear, my amusement isn't that I find this technique to not be useful for the purpose it was created, but that 40 years later, we find ourselves in pursuit for the advancement of AI to be somewhat back where we already were; albeit, in a more semi-automated fashion as someone still has to create the underlying rule-set.

I do feel that the introduction of generative neural network models in both natural language and multi-media creation has been a tremendous boon for the advancement of AI, it just amuses me to see that which was old is new again.

This amuses me tremendously. I began programming in the early 1980s and quickly developed an interest in Artificial Intelligence. At the time there was a great interest in the advancement of AI by the introduction of "Expert Systems" (which would later play a part in the ‘Second AI Winter’).

What Amazon appears to have done here is use a transformers based neural network (aka LLM) to translate natural language into symbolic logic rules which are collectively used together in what could be identified as an Expert System.

Full Circle. Hilarious.

For reference to those on the younger side: The Computer Chronicles (1984) https://www.youtube.com/watch?v=_S3m0V_ZF_Q

There are a number of ways this might get solved, but I would speculate that it will generally be solved by adding image metadata that is signed by a certificate authority similar to the way SSL certificates are assigned to domains.

I think eventually all digital cameras and image scanners will securely hash and sign images just as forensic cameras do to certify that an image was "captured" instead of generated.

Of course this leaves a grey area for image editing applications such as Photoshop, so there may also need to be some other level of certificate base signing introduced there as well.

For those who might not be aware of this, there is also an open source project on GitHub called "Twinny" which is an offline Visual Studio Code plugin equivalent to Copilot: https://github.com/rjmacarthy/twinny

It can be used with a number of local model services. Currently for my setup on a NVIDIA 4090, I'm running both the base and instruct model for deepseek-coder 6.7b using 5_K_M Quantization GGUF files (for performance) through llama.cpp "server" where the base model is for completions and the instruct model for chat interactions.

llama.cpp: https://github.com/ggerganov/llama.cpp/

deepseek-coder 6.7b base GGUF files: https://huggingface.co/TheBloke/deepseek-coder-6.7B-base-GGU...

deepseek-coder 6.7b instruct GGUF files: https://huggingface.co/TheBloke/deepseek-coder-6.7B-instruct...

Started 1982 on Tandy Color Computer. Still at it.

#1 Advice - Focus on getting things done (as many will not) and Lego build interconnected/isolated simplicity as much as possible.

Currently, if you disable chat history, you'll see this message:

Chat History is off for this browser. When history is turned off, new chats on this browser won't appear in your history on any of your devices, be used to train our models, or stored for longer than 30 days. This setting does not sync across browsers or devices.

I haven't personally tried that one, but on the HuggingFace LLM Leaderboard:

Open-Orca/Mistral-7B-SlimOrca - AVG: 60.37, ARC: 62.54, HellaSwag: 83.86, MMLU: 62.77, TruthfulQA: 54.23, Winogrande: 77.43, GSM8k: 21.38

Intel/neural-chat-7b-v3-2 - AVG: 68.29, ARC: 67.49, HellaSwag: 83.92, MMLU: 63.55, TruthfulQA: 59.68, Winogrande: 79.95, GSM8k: 55.12

BTW, for anyone who might not be aware of it, this model trained by Intel based on the Mistral architecture is probably the single best general 7B model available currently:

https://huggingface.co/Intel/neural-chat-7b-v3-2 (also see https://huggingface.co/Intel/neural-chat-7b-v3-1 from the previous version for more details)

It's licensed Apache 2.0 and unaligned (uncensored).