Skip to content

Commit c794dca

Browse files
john-b-yangjohn-b-yang
committed
Update glm-4-5, qwen 3 coder refs
1 parent 007cc3b commit c794dca

93 files changed

Lines changed: 3207 additions & 96 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎codeclash/games/game.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -194,7 +194,7 @@ def get_environment(self, branch_name: str | None = None) -> DockerEnvironment:
194194
"PIP_PROGRESS_BAR": "off",
195195
"TQDM_DISABLE": "1",
196196
},
197-
container_timeout="3h",
197+
container_timeout="10h",
198198
logger=self.logger,
199199
run_args=run_args,
200200
)

‎configs/main/BattleCode__claude-sonnet-4-20250514__glm-4-5__r15__s1000.yaml‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: claude-sonnet-4-20250514
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: anthropic/claude-sonnet-4-20250514
15+
model_kwargs:
16+
temperature: 0.0
17+
- agent: mini
18+
name: glm-4-5
19+
config:
20+
agent: !include mini/default.yaml
21+
model:
22+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
23+
model_class: portkey
24+
prompts:
25+
game_description: |-
26+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
27+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
28+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
29+
30+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
31+
After you and your competitor finish editing your codebases, the game is run automatically.
32+
33+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
34+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
35+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__claude-sonnet-4-20250514__qwen3-coder-480b-a35b-instruct__r15__s1000.yaml‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: claude-sonnet-4-20250514
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: anthropic/claude-sonnet-4-20250514
15+
model_kwargs:
16+
temperature: 0.0
17+
- agent: mini
18+
name: qwen3-coder-480b-a35b-instruct
19+
config:
20+
agent: !include mini/default.yaml
21+
model:
22+
model_name: '@fireworks/accounts/fireworks/models/qwen3-coder-480b-a35b-instruct'
23+
model_class: portkey
24+
prompts:
25+
game_description: |-
26+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
27+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
28+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
29+
30+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
31+
After you and your competitor finish editing your codebases, the game is run automatically.
32+
33+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
34+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
35+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__claude-sonnet-4-5-20250929__glm-4-5__r15__s1000.yaml‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: claude-sonnet-4-5-20250929
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: anthropic/claude-sonnet-4-5-20250929
15+
model_kwargs:
16+
temperature: 0.0
17+
- agent: mini
18+
name: glm-4-5
19+
config:
20+
agent: !include mini/default.yaml
21+
model:
22+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
23+
model_class: portkey
24+
prompts:
25+
game_description: |-
26+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
27+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
28+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
29+
30+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
31+
After you and your competitor finish editing your codebases, the game is run automatically.
32+
33+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
34+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
35+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__claude-sonnet-4-5-20250929__qwen3-coder-480b-a35b-instruct__r15__s1000.yaml‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: claude-sonnet-4-5-20250929
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: anthropic/claude-sonnet-4-5-20250929
15+
model_kwargs:
16+
temperature: 0.0
17+
- agent: mini
18+
name: qwen3-coder-480b-a35b-instruct
19+
config:
20+
agent: !include mini/default.yaml
21+
model:
22+
model_name: '@fireworks/accounts/fireworks/models/qwen3-coder-480b-a35b-instruct'
23+
model_class: portkey
24+
prompts:
25+
game_description: |-
26+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
27+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
28+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
29+
30+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
31+
After you and your competitor finish editing your codebases, the game is run automatically.
32+
33+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
34+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
35+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__gemini-2.5-pro__glm-4-5__r15__s1000.yaml‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: gemini-2.5-pro
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: '@google/gemini-2.5-pro'
15+
model_class: portkey
16+
- agent: mini
17+
name: glm-4-5
18+
config:
19+
agent: !include mini/default.yaml
20+
model:
21+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
22+
model_class: portkey
23+
prompts:
24+
game_description: |-
25+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
26+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
27+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
28+
29+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
30+
After you and your competitor finish editing your codebases, the game is run automatically.
31+
32+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
33+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
34+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__gemini-2.5-pro__qwen3-coder-480b-a35b-instruct__r15__s1000.yaml‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: gemini-2.5-pro
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: '@google/gemini-2.5-pro'
15+
model_class: portkey
16+
- agent: mini
17+
name: qwen3-coder-480b-a35b-instruct
18+
config:
19+
agent: !include mini/default.yaml
20+
model:
21+
model_name: '@fireworks/accounts/fireworks/models/qwen3-coder-480b-a35b-instruct'
22+
model_class: portkey
23+
prompts:
24+
game_description: |-
25+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
26+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
27+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
28+
29+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
30+
After you and your competitor finish editing your codebases, the game is run automatically.
31+
32+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
33+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
34+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__glm-4-5__gpt-5-mini__r15__s1000.yaml‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: glm-4-5
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
15+
model_class: portkey
16+
- agent: mini
17+
name: gpt-5-mini
18+
config:
19+
agent: !include mini/default.yaml
20+
model:
21+
model_name: '@openai/gpt-5-mini'
22+
model_class: portkey
23+
prompts:
24+
game_description: |-
25+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
26+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
27+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
28+
29+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
30+
After you and your competitor finish editing your codebases, the game is run automatically.
31+
32+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
33+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
34+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__glm-4-5__gpt-5__r15__s1000.yaml‎

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: glm-4-5
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
15+
model_class: portkey
16+
- agent: mini
17+
name: gpt-5
18+
config:
19+
agent: !include mini/default.yaml
20+
model:
21+
model_name: '@openai/gpt-5'
22+
model_class: portkey
23+
prompts:
24+
game_description: |-
25+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
26+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
27+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
28+
29+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
30+
After you and your competitor finish editing your codebases, the game is run automatically.
31+
32+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
33+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
34+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

‎configs/main/BattleCode__glm-4-5__grok-code-fast-1__r15__s1000.yaml‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
tournament:
2+
rounds: 15
3+
game:
4+
name: BattleCode
5+
sims_per_round: 1000
6+
args:
7+
maps: quack
8+
players:
9+
- agent: mini
10+
name: grok-code-fast-1
11+
config:
12+
agent: !include mini/default.yaml
13+
model:
14+
model_name: '@x-ai/grok-code-fast-1'
15+
model_class: portkey
16+
litellm_model_name_override: xai/grok-code-fast-1
17+
- agent: mini
18+
name: glm-4-5
19+
config:
20+
agent: !include mini/default.yaml
21+
model:
22+
model_name: '@fireworks/accounts/fireworks/models/glm-4-5'
23+
model_class: portkey
24+
prompts:
25+
game_description: |-
26+
You are a software developer ({{player_id}}) competing in a coding game called BattleCode.
27+
Battlecode 25 throws you into a real-time strategy showdown where your Python bot pilots a team of specialized robots—Soldiers, Moppers, Splashers—alongside towers that spawn units or generate resources.
28+
Your mission: paint over 70% of the map (or eliminate the enemy) by coordinating cleanups, area cover, and tower-building through tight bytecode budgets and clever unit synergy.
29+
30+
The game is played in 15 rounds. For every round, you (and your competitors) edit program code that controls your bot. This is round {{round}}.
31+
After you and your competitor finish editing your codebases, the game is run automatically.
32+
33+
Your task: improve the bot in `src/mysubmission`, located in {{working_dir}}.
34+
{{working_dir}} is your codebase, which contains both your both and supporting assets.
35+
All of your commands will be executed in the {{working_dir}} directory (see notes below).

0 commit comments

Comments
 (0)