Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

68 Pages«<7891011>»
scottbreak's Meta Decks for Vision AI
Falconer210
#172 Posted : Sunday, April 23, 2017 9:18:15 AM(UTC)
Rank: Member

Groups: Registered
Posts: 29

Thanks: 2 times
Was thanked: 6 time(s) in 5 post(s)
Hey Scottbreak,

Thanks for the excelent work on the development.

Here's one of my runs with VisionAI on your settings en one with the new silverfish which is included with hearthranger 9.1.0.0 onwards.
Both runs are made on casual with high MMR so these runs will likely be indicative of runs made at ranked, (roughly) rank 5 or so.

VisionAI:


Updated siverfish:


As you can see the silverfish run is quite a bit more successful than the VisionAI one.

Cheers,

Falconer
JoyAdmin
#171 Posted : Sunday, April 23, 2017 10:23:44 AM(UTC)
Rank: Administration

Groups: Administrators
Posts: 4,871

Thanks: 807 times
Was thanked: 4545 time(s) in 1642 post(s)
moon4717;48465 wrote:

Why did you use equality??????


Bot guesses Mage has secret of Counterspell, so it plays a spell to trigger it.

This issue will be fixed in next update.
tomatus.bot
#173 Posted : Sunday, April 23, 2017 11:42:16 AM(UTC)
Rank: Newbie

Groups: Registered
Posts: 7

Thanks: 1 times
Falconer210;48467 wrote:
Hey Scottbreak,

Thanks for the excelent work on the development.

Here's one of my runs with VisionAI on your settings en one with the new silverfish which is included with hearthranger 9.1.0.0 onwards.
Both runs are made on casual with high MMR so these runs will likely be indicative of runs made at ranked, (roughly) rank 5 or so.

VisionAI:


Updated siverfish:


As you can see the silverfish run is quite a bit more successful than the VisionAI one.

Cheers,

Falconer



I've seen silverfish also behave better. But it's also doing super stupid plays. For example it doesn't notice that Southsea Deckhand has charge with a weapon equipped. It seems that the state of the bot right now it not good enough to play ladder.
Also the silverfish dll included (version 129 I think) doesn't work with the silver.exe that everyone was using. And it's not reading the decks customizations from the Data directory.
NightlyKnight
#174 Posted : Sunday, April 23, 2017 11:59:12 AM(UTC)
NightlyKnight


Rank: Advanced Member

Groups: Registered
Posts: 44
Man
Location: Sofia, Bulgaria

Thanks: 19 times
Was thanked: 53 time(s) in 13 post(s)
tomatus.bot;48470 wrote:

Also the silverfish dll included (version 129 I think) doesn't work with the silver.exe that everyone was using. And it's not reading the decks customizations from the Data directory.


They are two different branches of silverfish.
gloryholegal
#175 Posted : Sunday, April 23, 2017 2:52:12 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 38

Thanks: 8 times
Was thanked: 10 time(s) in 9 post(s)
I am all for upgrades, and i understand that there are a lot of issues to workout with vision ai, but after loading up these files for pirate warriors, the number of misplays is just silly. Like doesn't attack with deck hand when it has charge, or always plays bloodsail raider before a weapon... on the same turn... I have a very very low mmr in casual, where i used to have a 90%+ win rate (not counting concede after wins). Now I get about a 35% winrate, anyone who isn't playing a basic deck has a chance to win with the number of misplays the bot does.

I knew it is in sort of a 'beta' testing phase and problems are to be expected, but at the moment it is nigh unusable. So far the only thing i like is that it seems a lot quicker. But speed doesn't mean much if it plays so badly.
fullhalfcake
#176 Posted : Sunday, April 23, 2017 6:27:54 PM(UTC)
Rank: Member

Groups: Registered
Posts: 13

Thanks: 6 times
Was thanked: 1 time(s) in 1 post(s)
Hey Scott, thanks for doing all the hard work.

My question is...
For Taunt Warrior mulligan, why does it discard "Fire Plumb's Heart" for Rogue, Hunter and Warlock?
The mulligan file is set to accept all opponents except those three.
hugo1970
#157 Posted : Sunday, April 23, 2017 6:49:53 PM(UTC)
Rank: Member

Groups: Registered
Posts: 10

Thanks: 1 times
Was thanked: 3 time(s) in 2 post(s)
u4ea;48461 wrote:
How is it possible? My winrate like in above post in casual 33%. I don't understand what's wrong. Hunters 0 wins.
Yeah, the only reason the winrate is that high, is because of weak opponents. Just make sure to concede after every win and you will get there. My winrate around rank 15 is like 40% max.
scottbreak
#177 Posted : Sunday, April 23, 2017 7:21:22 PM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
fullhalfcake;48477 wrote:
Hey Scott, thanks for doing all the hard work.

My question is...
For Taunt Warrior mulligan, why does it discard "Fire Plumb's Heart" for Rogue, Hunter and Warlock?
The mulligan file is set to accept all opponents except those three.

Because they are fast decks, and it is more important to get fast cards like Firey Win Axe to level out early.
1 user thanked scottbreak for this useful post.
fullhalfcake on 4/24/2017(UTC)
hugo1970
#178 Posted : Monday, April 24, 2017 2:51:49 AM(UTC)
Rank: Member

Groups: Registered
Posts: 10

Thanks: 1 times
Was thanked: 3 time(s) in 2 post(s)
decks

Standard, EU, unranked

Are those small sample sizes even useful?
kratosdev
#184 Posted : Monday, April 24, 2017 4:39:57 AM(UTC)
Rank: Newbie

Groups: Registered
Posts: 3

For token shaman seems that the bot does't play bloodlust when it is lethal
scottbreak
#179 Posted : Monday, April 24, 2017 5:17:45 AM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
hugo1970;48483 wrote:
decks

Standard, EU, unranked

Are those small sample sizes even useful?

Anything helps; are you using "concede if won last game"?
scottbreak
#185 Posted : Monday, April 24, 2017 5:19:01 AM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
kratosdev;48486 wrote:
For token shaman seems that the bot does't play bloodlust when it is lethal

Can you please report in the Debug Tool?
Cloud
#182 Posted : Monday, April 24, 2017 5:30:56 AM(UTC)
Cloud


Rank: Advanced Member

Groups: Registered
Posts: 332
Man
Location: Germany

Thanks: 29 times
Was thanked: 60 time(s) in 45 post(s)
scottbreak;48487 wrote:
Anything helps; are you using "concede if won last game"?


no need in unranked mode.
DarkFennec
#187 Posted : Monday, April 24, 2017 5:55:35 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 90

Thanks: 15 times
Was thanked: 18 time(s) in 17 post(s)
Scott, i've found an error (or a missplay in the coding) in the mulligan for the quest warrior:

scottbreak;48122 wrote:
scottbreak's Meta Decks for Vision AI

Taunt Warrior
Win rate: being tested
Download


This is the line in the mulligan, seems not working during the mulligan..

Quote:
{
"comment": "hold Dirty Rat (CFM_790) when opponent is mage or rogue or shaman or warrior",
"mulligan":"CFM_790",
"condition":"opp_hero(count(),mage=true)>0 OR opp_hero(count(),rogue=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}


this is the screenshot with HsReplay. The bot discarded the dirty rat even if the code is clear..








and if the mulligan you wrote is good, it should discard the quest. ( in my Replay's when the my enemy was a rogue, it keeped the quest)
maybe Joy should check if it is wrong..

Quote:
{
"comment": "hold Fier Plume's Heart (UNG_934) when opponent is druid or mage or paladin or priest or shaman or warrior",
"mulligan":"UNG_934",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),mage=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),priest=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}


another missplay i've found (this did not happened everytime) is with Stonehill Defender:

Quote:
{
"comment": "hold Stonehill Defender (UNG_072) when opponent is druid or hunter or mage or paladin or priest or shaman or warrior",
"mulligan":"UNG_072",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),hunter=true)>0 OR opp_hero(count(),mage=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),priest=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}




missplay with the ghoul:

Quote:
{
"comment": "hold Ravaging Ghoul (OG_149) when opponent is druid or hunter or paladin or shaman or warlock",
"mulligan":"OG_149",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),hunter=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warlock=true)>0",
"value":"hold",
}





Please tell me if i need to report this on the bot or it is ok in this way. hope it'll help you in the better way :)

1 user thanked DarkFennec for this useful post.
NightlyKnight on 4/24/2017(UTC)
Thagann
#191 Posted : Monday, April 24, 2017 6:16:31 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 80

Thanks: 6 times
Was thanked: 16 time(s) in 12 post(s)
Quote:
Scott, i've found an error (or a missplay in the coding) in the mulligan for the quest warrior:

scottbreak;48122 wrote:
scottbreak's Meta Decks for Vision AI

Taunt Warrior
Win rate: being tested
Download


This is the line in the mulligan, seems not working during the mulligan..

Quote:
{
"comment": "hold Dirty Rat (CFM_790) when opponent is mage or rogue or shaman or warrior",
"mulligan":"CFM_790",
"condition":"opp_hero(count(),mage=true)>0 OR opp_hero(count(),rogue=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}


this is the screenshot with HsReplay. The bot discarded the dirty rat even if the code is clear..








and if the mulligan you wrote is good, it should discard the quest. ( in my Replay's when the my enemy was a rogue, it keeped the quest)
maybe Joy should check if it is wrong..

Quote:
{
"comment": "hold Fier Plume's Heart (UNG_934) when opponent is druid or mage or paladin or priest or shaman or warrior",
"mulligan":"UNG_934",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),mage=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),priest=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}


another missplay i've found (this did not happened everytime) is with Stonehill Defender:

Quote:
{
"comment": "hold Stonehill Defender (UNG_072) when opponent is druid or hunter or mage or paladin or priest or shaman or warrior",
"mulligan":"UNG_072",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),hunter=true)>0 OR opp_hero(count(),mage=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),priest=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warrior=true)>0",
"value":"hold",
}




missplay with the ghoul:

Quote:
{
"comment": "hold Ravaging Ghoul (OG_149) when opponent is druid or hunter or paladin or shaman or warlock",
"mulligan":"OG_149",
"condition":"opp_hero(count(),druid=true)>0 OR opp_hero(count(),hunter=true)>0 OR opp_hero(count(),paladin=true)>0 OR opp_hero(count(),shaman=true)>0 OR opp_hero(count(),warlock=true)>0",
"value":"hold",
}





Please tell me if i need to report this on the bot or it is ok in this way. hope it'll help you in the better way :)



that's clearly another bug on the mulligan json files , we've already found some already , @JoyAdmin take a look at this
u4ea
#186 Posted : Monday, April 24, 2017 6:22:08 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 70
Man

Thanks: 44 times
Was thanked: 17 time(s) in 13 post(s)
scottbreak;48488 wrote:
Can you please report in the Debug Tool?

BL issue reported via forum and debug tool a week ago by me and one more person, so the work should be done or in progress …
hugo1970
#180 Posted : Monday, April 24, 2017 6:33:37 AM(UTC)
Rank: Member

Groups: Registered
Posts: 10

Thanks: 1 times
Was thanked: 3 time(s) in 2 post(s)
scottbreak;48487 wrote:
Anything helps; are you using "concede if won last game"?
Yes

Cloud;48489 wrote:
no need in unranked mode.
Yes it is needed. You will get stronger opponents if you keep winning.
Cloud
#181 Posted : Monday, April 24, 2017 7:52:19 AM(UTC)
Cloud


Rank: Advanced Member

Groups: Registered
Posts: 332
Man
Location: Germany

Thanks: 29 times
Was thanked: 60 time(s) in 45 post(s)
hugo1970;48494 wrote:
Yes

Yes it is needed. You will get stronger opponents if you keep winning.


Oh, good to know :)
scottbreak
#183 Posted : Monday, April 24, 2017 8:04:28 AM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
Cloud;48489 wrote:
no need in unranked mode.

There is a MMR in unranked as well as ranked.
scottbreak
#188 Posted : Monday, April 24, 2017 8:06:40 AM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
DarkFennec;48490 wrote:
Scott, i've found an error (or a missplay in the coding) in the mulligan for the quest warrior:



This is the line in the mulligan, seems not working during the mulligan..



this is the screenshot with HsReplay. The bot discarded the dirty rat even if the code is clear..








and if the mulligan you wrote is good, it should discard the quest. ( in my Replay's when the my enemy was a rogue, it keeped the quest)
maybe Joy should check if it is wrong..



another missplay i've found (this did not happened everytime) is with Stonehill Defender:





missplay with the ghoul:






Please tell me if i need to report this on the bot or it is ok in this way. hope it'll help you in the better way :)


Are you using HR ver 9.2.0.0?
68 Pages«<7891011>»
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.

Powered by YAF | YAF © 2003-2011, Yet Another Forum.NET
This page was generated in 0.484 seconds.