NousResearch/atropos
1
0
Fork 0
mirror of https://github.com/NousResearch/atropos.git synced 2026-04-28 17:29:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

merged latest

Browse source
This commit is contained in:
Shannon Sands 2025-05-15 11:54:33 -07:00
parent ba6ba173c1
commit d768ad68aa
9 changed files with 28 additions and 3232 deletions
Download patch file Download diff file Expand all files Collapse all files

8
environments/game_environments/gymnasium/blackjack/blackjack_env_no_thinking.py
View file

@ -5,7 +5,7 @@ from typing import Dict, List, Optional, Tuple
import gymnasium as gym
from atroposlib.envs.base import BaseEnv, BaseEnvConfig, OpenaiConfig, ScoredDataItem
from atroposlib.envs.base import APIServerConfig, BaseEnv, BaseEnvConfig, ScoredDataItem
from atroposlib.type_definitions import Item, Message
from atroposlib.utils.tokenize_for_trainer import tokenize_for_trainer
from atroposlib.utils.tool_call_parser import parse_tool_call
@ -35,7 +35,7 @@ class BlackjackEnvNoThinking(BaseEnv):
def __init__(
self,
config: BlackjackEnvNoThinkingConfig,
server_configs: List[OpenaiConfig],
server_configs: List[APIServerConfig],
slurm: bool = True,
testing: bool = False,
):
@ -74,7 +74,7 @@ class BlackjackEnvNoThinking(BaseEnv):
)
@classmethod
def config_init(cls) -> Tuple[BlackjackEnvNoThinkingConfig, List[OpenaiConfig]]:
def config_init(cls) -> Tuple[BlackjackEnvNoThinkingConfig, List[APIServerConfig]]:
env_config = BlackjackEnvNoThinkingConfig(
tokenizer_name="NousResearch/DeepHermes-3-Llama-3-8B-Preview",
group_size=16,
@ -87,7 +87,7 @@ class BlackjackEnvNoThinking(BaseEnv):
eval_episodes=100,
)
server_configs = [
OpenaiConfig(
APIServerConfig(
model_name="NousResearch/DeepHermes-3-Llama-3-8B-Preview",
base_url="http://localhost:9001/v1",
api_key="x",

8
environments/game_environments/gymnasium/blackjack/blackjack_env_thinking.py
View file

@ -21,10 +21,10 @@ import gymnasium
from tqdm.asyncio import tqdm_asyncio
from atroposlib.envs.base import (
APIServerConfig,
BaseEnv,
BaseEnvConfig,
EvalHandlingEnum,
OpenaiConfig,
ScoredDataGroup,
)
from atroposlib.utils.best_of_n_selection import select_best_index
@ -79,7 +79,7 @@ class BlackjackEnv(BaseEnv):
def __init__(
self,
config: BlackjackEnvConfig,
server_configs: List[OpenaiConfig],
server_configs: List[APIServerConfig],
slurm: bool = True,
testing: bool = False,
):
@ -936,7 +936,7 @@ class BlackjackEnv(BaseEnv):
await super().wandb_log(wandb_metrics)
@classmethod
def config_init(cls) -> Tuple[BlackjackEnvConfig, List[OpenaiConfig]]:
def config_init(cls) -> Tuple[BlackjackEnvConfig, List[APIServerConfig]]:
env_config = BlackjackEnvConfig(
tokenizer_name="NousResearch/DeepHermes-3-Llama-3-8B-Preview",
group_size=16,
@ -964,7 +964,7 @@ class BlackjackEnv(BaseEnv):
tiebreak_token_factor=0.01,
)
server_configs = [
OpenaiConfig(
APIServerConfig(
model_name="NousResearch/DeepHermes-3-Llama-3-8B-Preview",
base_url="http://localhost:9004/v1",
api_key="x",

4
environments/game_environments/gymnasium/blackjack/blackjack_local_server.py
View file

@ -5,7 +5,7 @@ import random
from dotenv import load_dotenv
from atroposlib.envs.base import EvalHandlingEnum, OpenaiConfig
from atroposlib.envs.base import APIServerConfig, EvalHandlingEnum
from environments.game_environments.gymnasium.blackjack.blackjack_env_thinking import (
BlackjackEnv,
BlackjackEnvConfig,
@ -47,7 +47,7 @@ async def main():
mc_samples=1,
)
server_configs = [
OpenaiConfig(
APIServerConfig(
model_name="gpt-4.1-nano",
base_url="https://api.openai.com/v1",
api_key=os.getenv("OPENAI_API_KEY"),

4
environments/game_environments/gymnasium/blackjack/blackjack_local_server_no_thinking.py
View file

@ -6,7 +6,7 @@ from typing import Optional
from dotenv import load_dotenv
from atroposlib.envs.base import EvalHandlingEnum, OpenaiConfig, ScoredDataItem
from atroposlib.envs.base import APIServerConfig, EvalHandlingEnum, ScoredDataItem
from environments.game_environments.gymnasium.blackjack.blackjack_env_no_thinking import (
BlackjackEnvNoThinking,
BlackjackEnvNoThinkingConfig,
@ -41,7 +41,7 @@ async def main():
eval_episodes=0,
)
server_configs = [
OpenaiConfig(
APIServerConfig(
model_name="gpt-4.1-nano",
base_url="https://api.openai.com/v1",
api_key=os.getenv("OPENAI_API_KEY"),