Skip to content

RLM spawning #751

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
eligotts wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

RLM spawning #751

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3ca157e
implemented protocol wrapper on top of environment. protocol owns dat…
eligotts Jan 21, 2026
48f6698
call run_rollout directly rather than run_group
eligotts Jan 21, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions verifiers/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,9 @@
from .envs.environment import Environment # noqa # isort: skip
from .envs.multiturn_env import MultiTurnEnv # noqa # isort: skip
from .envs.tool_env import ToolEnv # noqa # isort: skip
from .envs.actor import Actor # noqa # isort: skip
from .envs.protocol import Protocol # noqa # isort: skip
from .envs.multiagent_env import MultiAgentEnv, SingleTurnMAEnv # noqa # isort: skip

# main imports
from .envs.env_group import EnvGroup
Expand Down Expand Up @@ -62,12 +65,16 @@
"MCPEnv",
"Environment",
"MultiTurnEnv",
"MultiAgentEnv",
"SingleTurnMAEnv",
"SingleTurnEnv",
"PythonEnv",
"SandboxEnv",
"StatefulToolEnv",
"ToolEnv",
"EnvGroup",
"Actor",
"Protocol",
"extract_boxed_answer",
"extract_hash_answer",
"load_example_dataset",
Expand All @@ -80,6 +87,7 @@
"get_model_and_tokenizer",
"RLTrainer",
"RLConfig",
"MultiAgentOrchestrator",
"GRPOTrainer",
"GRPOConfig",
"grpo_defaults",
Expand All @@ -94,6 +102,7 @@
"get_model_and_tokenizer": "verifiers.rl.trainer.utils:get_model_and_tokenizer",
"RLConfig": "verifiers.rl.trainer:RLConfig",
"RLTrainer": "verifiers.rl.trainer:RLTrainer",
"MultiAgentOrchestrator": "verifiers.rl.trainer:MultiAgentOrchestrator",
"GRPOTrainer": "verifiers.rl.trainer:GRPOTrainer",
"GRPOConfig": "verifiers.rl.trainer:GRPOConfig",
"grpo_defaults": "verifiers.rl.trainer:grpo_defaults",
Expand Down Expand Up @@ -135,6 +144,7 @@ def __getattr__(name: str):
from .rl.trainer import ( # noqa: F401
GRPOConfig,
GRPOTrainer,
MultiAgentOrchestrator,
RLConfig,
RLTrainer,
grpo_defaults,
Expand Down
19 changes: 19 additions & 0 deletions verifiers/envs/actor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,19 @@
"""
Actor: A trainable entity in multi-agent environments.

Actors are registered to a Protocol and define the system prompt
used when making model calls.
"""
from dataclasses import dataclass


@dataclass
class Actor:
"""
A trainable actor. Registered to Protocol.

The system_prompt is used when this actor makes model calls.
"""

id: str
system_prompt: str = ""
Loading