





🌟 Special thanks to our amazing supporters:
✨ $10 Tier: [Geeks Love Detail]
🌈 $5 Tier: [Benedikt][David Martínez Martí]
It's been a while since I posted an update here, as I've been busy launching an API to the game called Derk's Gym. You can find more information about it here: https://gym.derkgame.com Meanwhile, I have had a bit of time to add some features and polish to the game as well. Most recently, I added an ELO based ranking system, which means you now have a chance to meet anyone in the game on the battlefield! Though the likelihood to battle someone is less for teams that are much stronger or weaker than your team. Here's the full changelog for the past couple of weeks: $$ 0.26.0 2020-10-07 ELO based ranking & matchmaking - Each player now has an associated ELO rating, which is updated when you battle someone. - There are five fixed divisions; Diamond, Ruby, Gold, Silver and Copper. Your rating determines your division. - Matchmaking is based on ELO ratings. The selection is weighted to favour teams you're likely to have an even matchup with (not too hard, not too easy), but you always have a chance to fight any active team. - You need to play a minimum of five matches to show up in the rankings. $$ 0.25.24 2020-09-15 Python Reinforcement Learning Environment I'm happy to announce that there is now an OpenAI reinforcement learning environment based on the game available called [Derk's Gym website](https://gym.derkgame.com)! $$ 0.25.23 2020-08-16 Load Derkling brain from another Derkling $$ 0.25.22 2020-08-14 Limit number of runs when using scheduling $$ 0.25.21 2020-08-10 Add item stats to item tooltips Such as damage, cast time, cooldown etc. $$ 0.25.20 2020-08-10 Kill, victory and loss bounties added When a unit is "pushed" (e.g. Magnum), their fall damage is attributed to the one that pushed them. $$ 0.25.19 2020-08-07 Move bounties to Derklings The bounties are now associated with the Derklings themselves, rather than with the training setup. There is also a "presets" feature where you can save and load bounty configurations. This also made it possible to do "dual" training again (train both sides at the same time), which has been re-enabled.
[ 6081 ]
[ 1502 ]
[ 2068 ]