Fat Fritz 2

[gaard]

New Fat Fritz information here today:

https://en.chessbase.com/post/interview ... -fritz-2-0

Lol that's even bigger scam than before.
The headline starts that way:

2/12/2021 – Albert Silver is the head coach of Fat Fritz, and its successor. His path to inventing the 2.0 engine is worth talking about...

It's impossible to take CB serious anymore, they have lost their mind - really someone should sue them.

His path to inventing the 2.0 engine hahaha

Now surely the unwashed masses think he really has invented something, a new engine. My god it cannot get worse anymore.

All while infringing on someone else's copyright. Shameless.

[Ozymandias]

New Fat Fritz information here today:

https://en.chessbase.com/post/interview ... -fritz-2-0

His path to inventing the 2.0 engine is worth talking about, because once he was almost ready to publishing it he came across a new neural network technology from Japan – "NNUE". Silver reconsidered his close-to-finished project, and dedicated his mind and soul to the efficiently updatable neural network.

Translation:

"We were pestering him to give us something we could market as a new Fritz, and when he was about to hand out FF1.2, which was surely going to be less than a 100 Elo points worse, than the best 20x256 Lc0 nets out there, he was saved by the bell."

[connor_mcmonigle]

It would appear ChessBase came to their senses and are now providing the entirety of the source (including the network) here https://github.com/DanielUranga/Stockfish/tree/ff2. The FatFritz2 network can be downloaded here: https://github.com/DanielUranga/Stockfi ... tz2_v1.bin. Following this change, there is now no GPL license violation in my understanding. ChessBase's marketing is still appalling, but such is life.

[connor_mcmonigle]

You said correct as in it wasn’t a violation then you are saying it was a violation. I’m confused. If the violation is not selling it then what is the violation?

Selling is perfectly fine and allowed by the GPL. But they are not providing the complete source, and that's a requirement - regardless if you sell it or not.

They can fix the violation by publishing the full source - what Dan was proposing - but they can't change what has already happened, i.e. that they're violating it now.

Slowing down won't save you from a speeding ticket if you already got caught by a Gatso.

GPLv3 grants you 30 days following notification of a violation to cure the violation:

8. Termination.

You may not propagate or modify a covered work except as expressly provided under this License. Any attempt otherwise to propagate or modify it is void, and will automatically terminate your rights under this License (including any patent licenses granted under the third paragraph of section 11).

However, if you cease all violation of this License, then your license from a particular copyright holder is reinstated (a) provisionally, unless and until the copyright holder explicitly and finally terminates your license, and (b) permanently, if the copyright holder fails to notify you of the violation by some reasonable means prior to 60 days after the cessation.

Moreover, your license from a particular copyright holder is reinstated permanently if the copyright holder notifies you of the violation by some reasonable means, this is the first time you have received notice of violation of this License (for any work) from that copyright holder, and you cure the violation prior to 30 days after your receipt of the notice.

Termination of your rights under this section does not terminate the licenses of parties who have received copies or rights from you under this License. If your rights have been terminated and not permanently reinstated, you do not qualify to receive new licenses for the same material under section 10.

[glennsamuel32]

It would appear ChessBase came to their senses and are now providing the entirety of the source (including the network) here https://github.com/DanielUranga/Stockfish/tree/ff2. The FatFritz2 network can be downloaded here: https://github.com/DanielUranga/Stockfi ... tz2_v1.bin. Following this change, there is now no GPL license violation in my understanding. ChessBase's marketing is still appalling, but such is life.

Committed 22 hrs ago and also changed the layers from 32 to 16 !!!
Probably a net created off-hand by you-know-who

It's only because people stood up that these changes are happening.
Just another reason Stockfish needs to fight this.
It's disgusting to see the work of the community being manipulated and marketed for financial gains.

To say considerable efforts went into FF2 is laughable...
FF2 will fall by the wayside just like FF1.

I'm very glad some have already offered their services to fight this.

[abgursu]

It would appear ChessBase came to their senses and are now providing the entirety of the source (including the network) here https://github.com/DanielUranga/Stockfish/tree/ff2. The FatFritz2 network can be downloaded here: https://github.com/DanielUranga/Stockfi ... tz2_v1.bin. Following this change, there is now no GPL license violation in my understanding. ChessBase's marketing is still appalling, but such is life.

Hi,
Did anybody test this net to see if it's the same with FF2?

[connor_mcmonigle]

It would appear ChessBase came to their senses and are now providing the entirety of the source (including the network) here https://github.com/DanielUranga/Stockfish/tree/ff2. The FatFritz2 network can be downloaded here: https://github.com/DanielUranga/Stockfi ... tz2_v1.bin. Following this change, there is now no GPL license violation in my understanding. ChessBase's marketing is still appalling, but such is life.

Hi,
Did anybody test this net to see if it's the same with FF2?

I'm unsure. Someone who has actually purchased it from ChessBase would have to determine whether the benches match. I'll likely do some more testing, but initial results look to be in the same ball park as other third party testing groups' results.

The change of the intermediate layer sizes from 32 to 16 is interesting. This results in the FF2 network (512x2-16-16-1) actually requiring slightly fewer ops than the 256x2-32-32-1 networks currently used by SF dev to evaluate. However, the incremental updates will be more costly. An interesting tradeoff.

[connor_mcmonigle]

It would appear ChessBase came to their senses and are now providing the entirety of the source (including the network) here https://github.com/DanielUranga/Stockfish/tree/ff2. The FatFritz2 network can be downloaded here: https://github.com/DanielUranga/Stockfi ... tz2_v1.bin. Following this change, there is now no GPL license violation in my understanding. ChessBase's marketing is still appalling, but such is life.

Hi,
Did anybody test this net to see if it's the same with FF2?

I'm unsure. Someone who has actually purchased it from ChessBase would have to determine whether the benches match. I'll likely do some more testing, but initial results look to be in the same ball park as other third party testing groups' results.

The change of the intermediate layer sizes from 32 to 16 is interesting. This results in the FF2 network (512x2-16-16-1) actually requiring slightly fewer ops than the 256x2-32-32-1 networks currently used by SF dev to evaluate. However, the incremental updates will be more costly. An interesting tradeoff.

Hmm. It has come to my attention that the benches might actually not match suggesting some foul play (a "fake"/unofficial, perhaps weakened, network).

This seems like an incredibly odd decision on ChessBase's part if true. If someone having purchased/otherwise obtained, the official ChessBase FF2 binary could determine whether the benches match the published source, that would be appreciated.

[kinderchocolate]

Sorry. I may have missed if the YouTube video has already been posted (too many pages).

Chessbase is trying to paint Mr Silver as a computer chess expert who "developed" the new NNUE with his "hundreds and hundreds" of hard hours.

[crem]

Sorry. I may have missed if the YouTube video has already been posted (too many pages).

Chessbase is trying to paint Mr Silver as a computer chess expert who "developed" the new NNUE with his "hundreds and hundreds" of hard hours.

Last time it was "months, many months":