Genetic AI Project

[The Great Apple]

Ta Moham. I think I might know what is wrong. Attached is a file which I hope will fix the problem.

Unzip in Mods/GeneticAI/Assets/Python/CUSTOM

Turns out instead of fixing it I actually broke it some more. Updated the attachment on the post.

 

[The Great Apple]

New version:

UN now works.
Done some tweaks to logic of the completion popup (it used to popup too early).
Hopefully fixed a crash on some machines.

 

[Craterus22]

New version:

UN now works.
Done some tweaks to logic of the completion popup (it used to popup too early).
Hopefully fixed a crash on some machines.

just installed .28 with aidata3 - got as far as 3800bc before seeming to lockup - will give it another 5 minutes and then send you a dump

ps - was i supposed to add the file above again for the fix or was it included in this version?

 

[The Great Apple]

My fault, as usual. I got my .dll files messed up and it looks like I overwrote it with a test dll. Give me a few mins to recompile and I'll post up a working version.

The fix is included.

 

[Arlborn]

just installed .28 with aidata3 - got as far as 3800bc before seeming to lockup - will give it another 5 minutes and then send you a dump

ps - was i supposed to add the file above again for the fix or was it included in this version?

Hum, Im already testing AI3 for quite a while..try other? xD

 

[The Great Apple]

Fixed and uploaded. v0.285

 

[The Great Apple]

Ok. Turns out not to be fixed at all. I'm not quite sure what's going on - it works fine in testing (got to 1000BC in 3 games, was crashing at about 3600BC), and now it crashes again.

Sorry about this folks, I'll try and fix it. The thing I find most confusing is that I haven't actually changed much, and certainly not anything that would cause a crash at this time in the game.

I'm going to slow down a bit, do some pretty big debugging. Be back in an hour...

 

[Craterus22]

Ok. Turns out not to be fixed at all. I'm not quite sure what's going on - it works fine in testing (got to 1000BC in 3 games, was crashing at about 3600BC), and now it crashes again.

Sorry about this folks, I'll try and fix it. The thing I find most confusing is that I haven't actually changed much, and certainly not anything that would cause a crash at this time in the game.

I will send you a dump (285 with aidata13) - mine crashed at 3720bc...

Don't get frustrated - you are doing a great job...

I will gladly be donating my computer time to this project once I get a version that works for me, maybe keep the last known good version up for download for people just finding this thread and include your updated versions for people that are having trouble?

ps - just ran aidata13 again and it crashed when buddi religion was founded

 

[Gunner]

Here is the results of my first set. Is the OldAIData file supposed to be empty, because mine is. Since I thought that might be wrong I've also included the original AI file I downloaded.

http://forums.civfanatics.com/uploads/50429/Gunner_s_AI_data.zip

 

[The Great Apple]

You know that hour? It flew by.

The problem is the datasets, and not the mod. Can you try a different dataset?

Download is back up.

Thanks Gunner!

 

[Craterus22]

You know that hour? It flew by.

The problem is the datasets, and not the mod. Can you try a different dataset?

Download is back up.

Thanks Gunner!

aidata13 and aidata18 crashed for me in the 3000's BC

 

[The Great Apple]

I really do rush too much.

I found out the problem. The script I was using to generate the new AIs was basing them on an incorrect default data set. The unit AI dataset wasn't long enough so the program crashed when trying to access data which wasn't loaded in in the first place. For some reason my logging system didn't pick this up - it's picked up similar errors before.

I've replaced the datasets in the first few posts with working ones. Everything should be hunky dorey now.

Have fun!

 

[Craterus22]

I really do rush too much.

I found out the problem. The script I was using to generate the new AIs was basing them on an incorrect default data set. The unit AI dataset wasn't long enough so the program crashed when trying to access data which wasn't loaded in in the first place. For some reason my logging system didn't pick this up - it's picked up similar errors before.

I've replaced the datasets in the first few posts with working ones. Everything should be hunky dorey now.

Have fun!

I am currently using aidata8 on a small pangea with 10 ai... into the 600's ad and still going strong (thought the turns are starting to slowdown - I wonder if these things would run faster with pitboss?).

Testing summary: sweet!

 

[Surtur]

Here are my results with an older version of AIData2. I used version 0.62. I had to play tiny maps at the end because most AIs already had 8 games but 4 had only 5 games.

Team Firaxis won btw.

-edit-
added DeadAIData.txt

 

[The Great Apple]

Thanks Surtur! A few more sets and I'll have enough breed the next generation.

I wonder if these things would run faster with pitboss?).

Now there's an idea. I'll have a poke around.

 

[shiremct]

Data sets for my run. All AI's had 7-8 matches except one with 5 and one with 6. Two AI's beat Firaxis AI with the top fitness level being around 2700ish.

 

[The Great Apple]

Thanks shiremct - I'll add them to my collection.

 

[suspendinlight]

I ran the mod all day while I was at work and I think it probably ran 2 or 3 games. Here are the screens I came home to for your debugging pleasure (I think this is .27).

I will have a data set to you in a couple hours.

 

[The Great Apple]

I ran the mod all day while I was at work and I think it probably ran 2 or 3 games. Here are the screens I came home to for your debugging pleasure (I think this is .27).

I will have a data set to you in a couple hours.

Thanks.

It looks like the data didn't load correctly for some reason. These errors are telling me that for some reason there are no AIs loaded.

I can think of a number of reasons:
1) Data file has corrupted somehow. I don't know how.
2) My trim script has gone mad and killed all your AIs, leaving AIData.txt empty, and DeadAIData.txt nice and full.

I don't think it's either. The most damage my trim script can do is killing all but one AI... but that would require one AI to have a fitness of about 200000 and the rest about 2, as well as some extreme bad luck.

If you got it running again, then I suspect it's just a weird runtime error. I've taken a look through a few other bits that I suspect and I can't see anything wrong with them.

 

[suspendinlight]

It did kill 3 AIs since I last checked and I am playing with 18 civs could that be the problem?