You are not alone mate :p

Cycle Count: 434 Condition: Normal Maximum Capacity: 88%

Long way to go

I assume you are referring to fine tuning a model here?

You could also just continue pre-training of an existing foundation model. Would still be cheaper by not starting from zero.

The amount of accuracy while doing fine tuning or distillation is usually better than pre-training an existing model, not to mention the graph against the cost.

I mean, at one point this goat did like almost 7000 commits a year. That's around 20 a day.

- 128-bit compatibility with UUID - 1.21e+24 unique ULIDs per millisecond - Lexicographically sortable! - Canonically encoded as a 26 character string, as opposed to the 36 character UUID - Uses Crockford's base32 for better efficiency and readability (5 bits per character) - Case insensitive - No special characters (URL safe) - Monotonic sort order (correctly detects and handles the same millisecond)

ULID's claim according to their official documentation are the above of which everything is taken care of by UUIDv7 officially now. Why should I or not shift back to UUIDv7?

One big practical benefit of UUIDv7 is that it is hex, and simple substring extraction and from_hex functions, which are available pretty much everywhere, will be able to extract the timestamp portion from the high bits.

Having had to massage lots of data with base62 and other weirdnesses, this lowest common denominator approach is a really big selling point.

People often bikeshed about shaving a few bytes off the ID or something, but in most applications this is not very important and data storage formats are usually compressed so costs are about entropy not alphabet.

Some databases directly support uuid, so they can save it as the binary data itself.

Do NOT do that with MS-SQL. The uniqueidentifier type shuffles byes around and does not give you the ordering you want.

So whether you use ULID or UUIDv7, save it as binary(16) in MSSQL.

(There is a special custom UUID implementation someone made for MSSQL that will sort time-ordered on MSQL but nowhere else; IMO that is too niche...just use binary(16)...)

I'm trying to build a custom domain network on my website but I couldn't figure out how people can use my subdomain to see their content

All depends by your domain registrar, in my company we simply can add/remove every type of domain records via API keys (so with a "simple" curl to our registrar endpoint)

It's high time Python got some upgrade. Game changer for managing packages and dealing with unnecessary fallbacks. UV for Python is like pnpm for Javascript.

The github trending repositories are repeated... Multiple times over and over again. At first, I thought it was a glitch and refreshed a couple times, then realised something was seriously wrong when I saw the same pattern in other accounts & PCs. Links are below:

https://ibb.co/LrrK5t2 https://ibb.co/2PdL0jr https://ibb.co/ZhdTkQS https://ibb.co/ncxTrV3 https://ibb.co/JqcQDjw

If you don't face the same issue, please let me know if there is something wrong with my account.

i can see repeated repositories too.

I see it too

The field is advancing too fast to catch up with!