From 08ba9ea5ed2f312e879db2273deb8f07a849df55 Mon Sep 17 00:00:00 2001
From: Maxime MORGE <maxime.morge@univ-lille.fr>
Date: Mon, 3 Mar 2025 14:00:48 +0100
Subject: [PATCH] Test strategy for dictator game with preference alignment
---
README.md | 46 ++++++++++----
src/dictator/dictator_setup.py | 106 ++++++++++++++++++++++++---------
2 files changed, 113 insertions(+), 39 deletions(-)
diff --git a/README.md b/README.md
index 6b1d288..1b14b1b 100644
--- a/README.md
+++ b/README.md
@@ -57,19 +57,37 @@ We consider 4 allocation options where money can be lost in the division, each c
3. The dictator keeps 400, the other player receives 300, resulting in a 300 loss (utilitarian)
4. The dictator keeps 325, the other player also receives 325, and 350 is lost in the division (egalitarian)
-The following table shows the accuracy of the dictator's decision for each model and preference.
-The temperature is fixed at 0.7, and each experiment was conducted 30 times.
+The following table presents the accuracy of the dictator's decision for each model and preference,
+regardless of whether the models were prompted to generate a strategy or specific actions.
+The temperature is set to 0.7, and each experiment involving action generation was repeated 30 times.
+
+| *Model* | *Generation* | *SELFISH* | *ALTRUISTIC* | *UTILITARIAN* | *EGALITARIAN* |
+|-----------------|---------------|---------------|----------------|------------------|------------------|
+| *gpt-4.5* | *actions* | 1.00 | 1.00 | 0.50 | 1.00 |
+| *llama3* | *actions* | 1.00 | 0.90 | 0.40 | 0.73 |
+| *mistral-small* | *actions* | 0.40 | 0.93 | 0.76 | 0.16 |
+| *deepseek-r1 | *actions* | 0.06 | 0.20 | 0.76 | 0.03 |
+|-----------------| ------------- | ------------- | -------------- | ---------------- | ---------------- |
+| *gpt-4.5* | *strategy* | 1.00 | 1.00 | 1.00 | 1.00 |
+| *llama3* | *actions* | 1.00 | 1.00 | 1.00 | 1.00 |
+| *mistral-small* | *actions* | 1.00 | 1.00 | 1.00 | 1.00 |
+| *deepseek-r1 | *actions* | - | - | - | - |
+
+
+This table helps assess the models’ ability to align with different preferences.
+When models are explicitly prompted to generate strategies,
+they exhibit perfect alignment with the predefined preferences except for DeepSeek-R1,
+which does not generate valid code.
+When models are prompted to generate actions, GPT-4.5 and Llama3 demonstrate strong alignment across all preferences,
+GPT-4.5 consistently aligns well across all preferences when generating strategies but struggles with
+utilitarianism when generating actions.
+Llama3 performs well for selfish and altruistic preferences but shows weaker alignment for utilitarian and egalitarian
+choices.
+Mistral-small aligns best with altruistic preferences and maintains moderate performance on utilitarianism,
+but struggles with selfish and egalitarian preferences.
+Deepseek-r1 performs best for utilitarianism but has poor accuracy in other categories.
-| Model | SELFISH | ALTRUISTIC | UTILITARIAN | EGALITARIAN |
-|-----------------|-----------|--------------|---------------|----------------|
-| gpt-4.5 | 1.0 | 1.0 | 0.5 | 1.0 |
-| llama3 | 1.0 | 0.9 | 0.4 | 0.73 |
-| mistral-small | 0.4 | 0.93 | 0.76 | 0.16 |
-| deepseek-r1 | 0.06 | 0.2 | 0.76 | 0.03 |
-Bad decisions can be explained either by arithmetic errors (e.g., it is not the case that 500 + 100 > 400 + 300)
-or by misinterpretations of preferences (e.g., ‘I’m choosing to prioritize the common interest by keeping a
-relatively equal split with the other player’).
This table can be used to evaluate the models based on their ability to align with different preferences.
GPT-4.5 exhibits strong alignment across all preferences except for utilitarianism, where its performance is moderate.
@@ -79,6 +97,12 @@ Mistral-small shows the best alignment with altruistic preferences, while mainta
performance across the other preferences. Deepseek-r1 is most capable of aligning with utilitarian preferences,
but performs poorly in aligning with other preferences.
+Bad action selections can be explained either by arithmetic errors (e.g., it is not the case that 500 + 100 > 400 + 300)
+or by misinterpretations of preferences (e.g., ‘I’m choosing to prioritize the common interest by keeping a
+relatively equal split with the other player’).
+
+
+
## Ring-network game
A player is rational if she plays a best response to her beliefs.
diff --git a/src/dictator/dictator_setup.py b/src/dictator/dictator_setup.py
index 493cd9a..8eb9eb9 100644
--- a/src/dictator/dictator_setup.py
+++ b/src/dictator/dictator_setup.py
@@ -43,34 +43,38 @@ class AgentResponse(BaseModel):
class DictatorSetup:
debug = False
- def __init__(self, model: str, temperature: float, preference: Welfare):
+ def __init__(self, model: str, temperature: float, preference: Welfare, strategy=False):
self.model = model
self.temperature = temperature
self.preference = preference
- # Determine if using OpenAI model (like gpt-4)
- is_openai_model = model.startswith("gpt")
- base_url = "https://api.openai.com/v1" if is_openai_model else "http://localhost:11434/v1"
- # Define model information
- self.model_info = {
- "temperature": self.temperature,
- "function_calling": True,
- "parallel_tool_calls": True,
- "family": "unknown",
- "json_output": True , # Add this line to ensure JSON output is supported
- "vision": False
- }
- # Create the model client
- self.model_client = OpenAIChatCompletionClient(
- model=self.model,
- base_url=base_url,
- api_key=OPENAI_API_KEY,
- model_info=self.model_info,
- response_format = AgentResponse # Only set for OpenAI models that support it
- )
+ self.strategy = strategy
+
+ if not self.strategy:
+ is_openai_model = model.startswith("gpt")
+ base_url = "https://api.openai.com/v1" if is_openai_model else "http://localhost:11434/v1"
+ self.model_info = {
+ "temperature": self.temperature,
+ "function_calling": True,
+ "parallel_tool_calls": True,
+ "family": "unknown",
+ "json_output": True,
+ "vision": False
+ }
+
+ self.model_client = OpenAIChatCompletionClient(
+ model=self.model,
+ base_url=base_url,
+ api_key=OPENAI_API_KEY,
+ model_info=self.model_info,
+ response_format=AgentResponse
+ )
async def run(self) -> Dict:
"""Runs the dictator game and returns a detailed response dictionary."""
+ if self.strategy:
+ return self.apply_strategy(self.model)
+
instruction = f"""
Hi, let’s play a game. You are paired with another player.
Your role is to decide how to divide $1000, and the other player simply receives your choice.
@@ -96,14 +100,14 @@ class DictatorSetup:
[TextMessage(content=instruction, source="user")],
cancellation_token=CancellationToken(),
)
- # Extract JSON content from response
+
response_data = response.chat_message.content
if self.debug:
print(f"Raw Response: {response_data}")
- # Convert JSON string to dictionary before validation
+
response_dict = json.loads(response_data)
agent_response = AgentResponse.model_validate(response_dict)
- # Check if response is consistent with the specified preference
+
is_consistent = self.check_consistency(agent_response)
return {
"is_consistent": is_consistent,
@@ -113,6 +117,53 @@ class DictatorSetup:
"motivations": agent_response.motivation
}
+ def apply_strategy(self, model: str) -> Dict:
+ """Applies a predefined strategy based on the preference."""
+ if model == "gpt-4.5-preview-2025-02-27":
+ strategy_mapping = {
+ Welfare.SELFISH: {"my_share": 500, "other_share": 100, "lost": 400, "motivations": "Maximizing self-interest"},
+ Welfare.ALTRUISTIC: {"my_share": 100, "other_share": 500, "lost": 400, "motivations": "Helping others at a personal cost"},
+ Welfare.UTILITARIAN: {"my_share": 400, "other_share": 300, "lost": 300, "motivations": "Maximizing total utility"},
+ Welfare.EGALITARIAN: {"my_share": 325, "other_share": 325, "lost": 350, "motivations": "Ensuring fairness"}
+ }
+ return strategy_mapping.get(self.preference, {"error": "Preference strategy not defined"})
+ if model == "llama3":
+ strategy_map = {
+ Welfare.SELFISH: (500, 100, 400),
+ Welfare.ALTRUISTIC: (100, 500, 400),
+ Welfare.UTILITARIAN: (400, 300, 300),
+ Welfare.EGALITARIAN: (325, 325, 350)
+ }
+ if self.preference in strategy_map:
+ my_share = strategy_map[self.preference][0]
+ other_share = strategy_map[self.preference][1]
+ lost = strategy_map[self.preference][2]
+ return {
+ "my_share": my_share,
+ "other_share": other_share,
+ "lost": lost,
+ "is_consistent": True,
+ }
+ else:
+ raise ValueError("Invalid preference type")
+ return {"error": "Preference strategy not defined"}
+ if model == "mistral-small":
+ valid_choices = {
+ Welfare.SELFISH: {"my_share": 500, "other_share": 100, "lost": 400},
+ Welfare.ALTRUISTIC: {"my_share": 100, "other_share": 500, "lost": 400},
+ Welfare.UTILITARIAN: {"my_share": 400, "other_share": 300, "lost": 300},
+ Welfare.EGALITARIAN: {"my_share": 325, "other_share": 325, "lost": 350},
+ }
+ strategy = valid_choices.get(self.preference)
+ if not strategy:
+ raise ValueError(f"Unknown preference type {self.preference}")
+ return {
+ "is_consistent": True,
+ **strategy
+ }
+ if model == "deepseek-r1":
+ return {"error": "Preference strategy not defined"}
+
def check_consistency(self, agent_response: AgentResponse) -> bool:
"""Check if the response aligns with the given preference."""
@@ -135,8 +186,7 @@ class DictatorSetup:
# Run the async function and return the response
if __name__ == "__main__":
- # Example usage: setting preference to 'Egalitarian'
- preference = Welfare.EGALITARIAN # Can be Selfish, Altruistic, etc.
- game_agent = DictatorSetup(model="gpt-4.5-preview-2025-02-27", temperature=0.7, preference=preference) # or llama3, mistral-small, deepseek-r1
+ preference = Welfare.EGALITARIAN
+ game_agent = DictatorSetup(model="llama3", temperature=0.7, preference=preference, strategy=True)
response = asyncio.run(game_agent.run())
- print(response) # Prints detailed response with is_consistent, my_share, other_share, lost, motivations
+ print(response)
--
GitLab