Roman
Veteran
Hey. Thanks for the report! Can you check if it is fixed now?
ZBLL trainer
- Thread starter Roman
- Start date
Hey. Thanks for the report! Can you check if it is fixed now?
PapaSmurf
Veteran
It is! Thanks! Now back to the L grind.
Roman
Veteran
Issue with short / 2gen / stupid scrambles is now fixed.
2-gen scrambles listed here were removed from scrambles list. They were replaced by a bunch of alternative algs found on birdflu website. Short (12- moves) scrambles were also replaced.
related git commit: click.
The scramble generator tends to give you less B moves now.
related git commit: click.
@Pyjam @Aerma @Neel Gore
2-gen scrambles listed here were removed from scrambles list. They were replaced by a bunch of alternative algs found on birdflu website. Short (12- moves) scrambles were also replaced.
related git commit: click.
The scramble generator tends to give you less B moves now.
related git commit: click.
@Pyjam @Aerma @Neel Gore
Pyjam
Premium Member
Very well. Thanks a lot for this improvement.
Awesome, thanks for doing this 
I'm still going to continue working on my own trainer, partially because there's a bunch of other features I plan on implementing, and partially because it's a bit too late now to turn back xD
Neither of us are earning any money from these so I wouldn't call it a "competitor" though haha.
I'm still going to continue working on my own trainer, partially because there's a bunch of other features I plan on implementing, and partially because it's a bit too late now to turn back xD
Neither of us are earning any money from these so I wouldn't call it a "competitor" though haha.
D
Deleted member 51471
Guest
Have you tried jperm.net
PapaSmurf
Veteran
This or train yu are the best alg trainers. This specifically for ZBLL, OLL or PLL, otherwise train yu.
Duncan Bannon
Veteran
I have not, thanks for the resource! I don't really need it anymore, but thank you!
Nilsibert
Veteran
I'm a bit confused. In the bottom right window, the case and the scramble shown are always different from the main scramble above. Why is that?
Pyjam
Premium Member
It's the previous case.
Nilsibert
Veteran
Oh, silly me. That is actually quite useful now that I think about it.
Roman
Veteran
You can now switch between top view and 3D view during practising by clicking at the picture.
Switching between flat and 3D view in the selection layout is yet to come.
Switching between flat and 3D view in the selection layout is yet to come.
GenTheSnail
Veteran
This has been my main gripe with your trainer so far; the other one you fixed in the earlier post.You can now switch between top view and 3D view during practising by clicking at the picture.
Switching between flat and 3D view in the selection layout is yet to come.
Thank you so much for making a great trainer and now for making it even better!
Roman
Veteran
Flat view by default.
You can still switch between flat and 3D view in the timer by clicking on the picture.
related git commit: 42b7d49
PS If there are any misoriented pictures or if short/2gen scrambles still appear, please let me know.
If you don't tell me how to make ZBLL trainer better then how else would I know
You can still switch between flat and 3D view in the timer by clicking on the picture.
related git commit: 42b7d49
PS If there are any misoriented pictures or if short/2gen scrambles still appear, please let me know.
If you don't tell me how to make ZBLL trainer better then how else would I know
For the 2GLL pi case with H perm edges, all of the algs are the same 11-move optimal alg.
The short RUL and RUF algs on Birdflu for this case are either Niklas combos or Sune combos, so please don't include those either.
Roman
Veteran
Fair.
Now I have manually replaced this and a lot of scrambles sets so it should be a lot better on the scrambling side.
A few more:
"B' U2 F U' B U F' U B' U B",
"F U F' U B' U F U' B U2 F'",
F U' F2 U2 D' F U F' U D F2 U F'
(Probably the mirror/inverse cases are also affected but I haven't looked yet.)
Nilsibert
Veteran
I like the idea of the random mode, but it seems like when using it, some cases tend to show up significally less often than others. That is of course possible with true randomness. Is there a way to stay in recap mode, with only the order changing? Or better yet, I'd like it to have a sort of "semi random" mode, where cases can appear multiple times, but it would still guarantee that the number of times all the cases that come up is somewhat consistant, if you know what I mean.
Roman
Veteran
Yep, I will do this.
In the beginning each case will have equal probability, and when the case is shown, its probability is decreased 2 or 3 times.
Roman
Veteran
ZBLL trainer now efficiently gives you scrambles to practise all cases evenly during the session.
This feature is rather subtle, but those who practise with ZBLL a lot will feel the imporovement right off.
Related git commit: click
This feature is rather subtle, but those who practise with ZBLL a lot will feel the imporovement right off.
The downside of the recap mode is that after you've done with some cases, you don't expect them to appear anymore, which defeats the purpose. And with "true random" mode, some cases may not appear in the session at all, while others may appear 3,4, even 5 times!
With new probability feature introduced, both problems are mitigated.
When you select N cases to practise, a probability of (1/N) is assigned to each of them. When a certain scramble is given to you, its probability is decreased in half, making this case twice less likely to appear next time. Note that "twice less likely" does not mean "impossible". Right after that, all the probabilities are normalized so that their sum is equal to 1 (or 100%).
Example: you have selected 4 cases. Initial probabilities:
[25%, 25%, 25%, 25%]
Let's say the first one had been given to you, and it's probability is now 12.5%. After normalization, the probs become:
[14.28%, 28.57%, 28.57%, 28.57%].
The probability of first scramble appear again is only 14% — twice less than all others, but they all sum up to 100%.
As you practise, more and more cases get their probabilities cut in half, so at some point the first case will have about the same prob as all others and will likely to appear again. Nontheless, you can't predict when it appears exactly.
This can be interpreted as something between "recap mode" and "just random" mode.
With new probability feature introduced, both problems are mitigated.
When you select N cases to practise, a probability of (1/N) is assigned to each of them. When a certain scramble is given to you, its probability is decreased in half, making this case twice less likely to appear next time. Note that "twice less likely" does not mean "impossible". Right after that, all the probabilities are normalized so that their sum is equal to 1 (or 100%).
Example: you have selected 4 cases. Initial probabilities:
[25%, 25%, 25%, 25%]
Let's say the first one had been given to you, and it's probability is now 12.5%. After normalization, the probs become:
[14.28%, 28.57%, 28.57%, 28.57%].
The probability of first scramble appear again is only 14% — twice less than all others, but they all sum up to 100%.
As you practise, more and more cases get their probabilities cut in half, so at some point the first case will have about the same prob as all others and will likely to appear again. Nontheless, you can't predict when it appears exactly.
This can be interpreted as something between "recap mode" and "just random" mode.
This is solved now.
Related git commit: click
Similar threads
