SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1809.05188 · 2018

CM3: Cooperative Multi-goal Multi-stage Multi-agent Reinforcement Learning

arXiv · PDF · Open in the Atlas

Code that ran

2 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/1809.05188.json · badge: /api/ran/1809.05188.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
011235813/cm3 canonical (classifier label)
classifier basis: census
2 ran · 3 no recorded run
FunctionStatusWhere it lives
test_particle Ran
ran on a synthesized input
011235813/cm3/alg/evaluate.py
code served (permissive licence) · get_code("67cc3c6cd167d79c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch
import torch.nn as nn

torch.manual_seed(0)

class MockAlgorithm(nn.Module):
    def __init__(self, n_agents, l_goal):
        super().__init__()
        self.n_agents = n_agents
        self.l_goal = l_goal
        
    def run_actor(self, local_others, local_v, goals, epsilon, sess):
        # Returns n_agents actions (integers)
        return np.array([0, 1], dtype=int)

class MockLandmark:
    def __init__(self, pos):
        self.state = type('obj', (object,), {'p_pos': pos})()

class MockWorld:
    def __init__(self, n_agents, l_goal):
        self.landmarks = [MockLandmark(np.random.rand(l_goal)) for _ in range(n_agents)]

class MockEnv:
    def __init__(self, n_agents, l_goal):
        self.n_agents = n_agents
        self.l_goal = l_goal
        self.world = MockWorld(n_agents, l_goal)
        self.step_count = 0
        self.max_steps = 5
        
    def reset(self):
        self.step_count = 0
        global_state = np.random.rand(10)
        local_others = np.random.rand(2, 6)
        local_v = np.random.rand(2, 4)
        done = False
        return global_state, local_others, local_v, done
    
    def step(self, actions):
        self.step_count += 1
        next_global_state = np.random.rand(10)
        next_local_others = np.random.rand(2, 6)
        next_local_v = np.random.rand(2, 4)
        reward = 1.0
        local_rewards = np.array([0.5, 0.3])
        done = self.step_count >= self.max_steps
        return next_global_state, next_local_others, next_local_v, reward, local_rewards, done
    
    def render(self):
        pass

ARGS = {
    'n_eval': 2,
    'env': MockEnv(n_agents=2, l_goal=3),
    'sess': None,
    'n_agents': 2,
    'l_goal': 3,
    'alg': MockAlgorithm(n_agents=2, l_goal=3),
    'render': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
test_particle_coma Ran
ran on a synthesized input
011235813/cm3/alg/evaluate.py
code served (permissive licence) · get_code("f5cd32a31728c1df")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch
import torch.nn as nn

torch.manual_seed(0)

class DummyEnv:
    def __init__(self, n_agents):
        self.n_agents = n_agents
        self.step_count = 0
        self.max_steps = 3
    
    def reset(self):
        self.step_count = 0
        global_state = np.array([0.1, 0.2, 0.3, 0.4], dtype=np.float32)
        local_v = [np.array([0.5, 0.6], dtype=np.float32) for _ in range(self.n_agents)]
        done = False
        return global_state, local_v, done
    
    def step(self, actions):
        self.step_count += 1
        next_global_state = np.array([0.1, 0.2, 0.3, 0.4], dtype=np.float32)
        next_local_v = [np.array([0.5, 0.6], dtype=np.float32) for _ in range(self.n_agents)]
        reward = 1.0
        local_rewards = np.array([0.5, 0.5], dtype=np.float32)
        done = self.step_count >= self.max_steps
        return next_global_state, next_local_v, reward, local_rewards, done
    
    def render(self):
        pass


class DummyAlgorithm(nn.Module):
    def __init__(self, n_agents):
        super().__init__()
        self.n_agents = n_agents
        self.linear = nn.Linear(2, 2)
    
    def run_actor(self, local_v, epsilon, sess):
        actions = np.array([0, 1], dtype=np.int64)
        return actions


class DummySession:
    pass


ARGS = {
    'n_eval': 2,
    'env': DummyEnv(n_agents=2),
    'sess': DummySession(),
    'n_agents': 2,
    'alg': DummyAlgorithm(n_agents=2),
    'render': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
convnet Not yet run
unverified (harvested, no recorded run)
011235813/cm3/alg/networks.py
code served (permissive licence) · get_code("1c28a1c3d2124c04")
How it ran

no invocation recorded.

fc2 Not yet run
unverified (harvested, no recorded run)
011235813/cm3/alg/networks.py
code served (permissive licence) · get_code("c760e2b7dc1ba768")
How it ran

no invocation recorded.

fc3 Not yet run
unverified (harvested, no recorded run)
011235813/cm3/alg/networks.py
code served (permissive licence) · get_code("622a4f24ec30dbd7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

A variety of cooperative multi-agent control problems require agents to achieve individual goals while contributing to collective success. This multi-goal multi-agent setting poses difficulties for recent algorithms, which primarily target settings with a single global reward, due to two new challenges: efficient exploration for learning both individual goal attainment and cooperation for others' success, and credit-assignment for interactions between actions and goals of different agents. To address both challenges, we restructure the problem into a novel two-stage curriculum, in which single-agent goal attainment is learned prior to learning multi-agent cooperation, and we derive a new multi-goal multi-agent policy gradient with a credit function for localized credit assignment. We use a function augmentation scheme to bridge value and policy functions across the curriculum. The complete architecture, called CM3, learns significantly faster than direct adaptations of existing algorithms on three challenging multi-goal multi-agent problems: cooperative navigation in difficult formations, negotiating multi-vehicle lane changes in the SUMO traffic simulator, and strategic cooperation in a Checkers environment.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("1809.05188")
get_code_for_paper("1809.05188")
have("1809.05188")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/1809.05188.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/1809.05188.svg)](https://syntology.ai/paper/1809.05188)

Connect an agent — have() is free.