tmikaeld

"Please note that normal charges for CDN bandwidth are billed separately"

Compared to cloudflare workers, which has free bandwidth, bunnys bandwidth is not that cheap at 0.01$/GB

So while their example suggests stream-encoding video is possible, it would probably be cost-prohibitive.

Impressive! Definitely going to try this, thanks!

It's inserted under the skin?! Seems quite risky, especially removal of it.

Fat

When factoring in motherboard, CPU etc, then yes. The max speed is only theoretical, unlike the Apple chips which actually benchmark on the speed specified.

When you factor in memory bandwidth (80GB/s for DDR4) - that's not even close to the M4 (120GB/s base model).

I work with full text search where this is common. Here is some points.

Stemming: Reducing words to their base or root form (e.g., “working,” “worked” becoming “work”).

Lemmatization: Similar to stemming, but more sophisticated, accounting for context (e.g., “better” lemmatizes to “good”).

Token normalization: Standardizing tokens, such as converting “wrk” to “work” through predefined rules (case folding, character replacement).

Fuzzy matching: Allowing approximate matches based on edit distance (e.g., “wrk” matches “work” due to minimal character difference).

Phonetic matching: Matching words that sound similar, sometimes used to match abbreviations or common misspellings.

Thesaurus-based search: Using a predefined list of synonyms or alternative spellings to expand search queries.

Most of these are open and free lists you can use, check the sources on manticore search for example.

My gripe with any of the open source phones is that id apps such as bankid don’t work without the secure token storage of Android. Is this Solvable?

You're right that reads and writes are immediate in the same client connection, this is how it works with CF KV as well - but not across the entire network.

On KV they expect up to 30 second latency before a write can be written everywhere, I expect similar here.

> ..each DO constantly streams a sequence of WAL entries to object storage - batched every 16MB or every ten seconds.

Which also means it may take 10 seconds before you can (reliably) read the write globally.

I keep failing to see how this can replace regionally placed database clusters which can serve a continent in milliseconds.

Edit: I know it uses streams, but those are only to 5 followers and CF have hundreds of datacenters. There is no physical way to guarantee reads in seconds unless all instances of the SQLite are always connected and even then, packet latency will cause issues.

I just tested it with ioredis node.js package for production use and Deno 2 could handle:

- Automatic connection pooling

- 400MB/s+ of throughput

- 20 000+ keys and larger values (10-50kb)

- 1000+ concurrent reads/writes

- 200-250MB of RAM usage max

Without breaking a sweat, the limitation was my keydb test server.

Does it affect libido?

The most famous project is Bun, the Javascript runtime

Since this is hackernews, here's a free and open source version of the same that works from the statusbar.

https://github.com/Gaulomatic/AirPodsSanity

I didn't create this, but I do detest subscription models on bugfixes..

Everything looks superb, except for one thing, the limit of 1Gbit/second per bucket is not useable in production.

This would be saturated by a single file download. Hopefully it's only a beta limitation!

This is really nice, is this converting to SVG from Raster like similar services?

This seems to be a growing issue..

It can’t, it’s the most likely answer to the question based on what it knows. Which kind of makes it funnier

And loose all of it when it fails.

Solved, see:

https://github.com/01-ai/Yi-Coder/issues/6#issuecomment-2334...

Works very well now! 65K input tokens with 8192 output tokens is no longer an issue on my 4090. (It maxes out on 22GB/VRAM)

Turns out that Ollama on windows will run multiple models in parallell consuming all available VRAM and RAM. Changing it to 1 fixed the issue, now it's working great! However, the context length for the output is very small - only 1024 tokens.

No, I’m using 9b-chat-q8_0 on a 4090

Click on "View more" in the dropdown on their page, it has many many quantized versions to choose from.

I have the same experience, hallucinates and rambles on and on about "solutions" that are not related.

Unfortunately, this has always been my experience with all open source code models that can be self-hosted.

This is why I don't use code LLM code generators for citing or outputting a solution that includes the current code, because it's inclined to remove parts that it thinks don't matter, but rather matter a lot further down the line. And if it's not caught in code reviews, that can cause severe and difficult to debug issues. I'm sure there will be an epidemic of these issues in a few years, because developers are definitely lazy enough to rely on it.

Please don't do this unless you have a backup yubikey with the same key on it..

Unfortunately the smaller model is not anywhere near GPT4 in quality and no one seems to want to host the bigger model (it was even removed from fireworks ai this week). And no one in their right mind want to send their code to deepmind chinese API hosting.

"Modern tech" often require significant corporate backing and/or significant amount of funds. I'm amazed that Haiku OS is still going considering it's surviving on donations.

Hm, I wish this could use a smartphone with lidar (Like iPhones/iPads) to scan the environment and create the 3D models. The viewer itself is very impressive and I see it will soon support VR. But I've never seen the viewer as the problem - capturing of the environment in an easy and affordable way - that's what's needed. Right now you have to resort to expensive 3D camera rigs.

You’d get that quality with a bigscreenvr headset at a lower cost even if you would have to buy a new computer.

https://youtu.be/tIIWYh601tg?si=3azdig7logfSWYKu

Check what this simmer is saying about it, it was an eyeopener for me as well