From b44a21e83e77c0a6ff58ddc894966f7ee3c2b069 Mon Sep 17 00:00:00 2001
From: stephanebonnevay <stephane.bonnevay@lizeo-group.com>
Date: Fri, 6 Jun 2025 07:45:12 +0200
Subject: [PATCH] Readme

---
 README.md | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 9ce2aa1..c35c049 100644
--- a/README.md
+++ b/README.md
@@ -434,15 +434,14 @@ Figures present the average points earned and prediction per round (95% confiden
 Against Constant behavior, <tt>GPT-4.5</tt> and <tt>Qwen3</tt> were able to generate a valid strategy. The charts show that they are able to correctly predict their opponent's strategy after just a few rounds. They perfectly identify the fact that their opponent always plays the same move.
 The predictions made by <tt>Mistral-Small<tt>, <tt>LLaMA3</tt>, and <tt>DeepSeek-R1</tt> are not incorrect, but the moves played are not in line with these predictions, which leads to a fairly low expected gain.
 
-The models exhibit varied approaches to decision-making in the MP game.
-<tt>GPT-4.5</tt> follows a fixed alternating pattern, switching between "Head" and "Tail" each turn, assuming the opponent behaves similarly.
-<tt>Mistral-Small<tt> adopts a reactive strategy, analyzing the  frequency of the  opponent’s past moves and selecting the less common one. On the other hand, <tt>Qwen3</tt>, relies on randomness — choosing moves unpredictably while  presuming the opponent will mimic its choice. <tt>LLaMA3</tt> does not  implement  a functioning strategy. Overall, these approaches demonstrate simplistic heuristics that may lack credibility and efficiency.
-
 ![Prediction Accuracy per Round by Actions Against Constant Behaviour (with 95% Confidence Interval)](figures/mp/mp_prediction_ConstHT.svg)
 ![Points Earned per Round by Actions Against Constant Behaviour (with 95% Confidence Interval)](figures/mp/mp_payoff_ConstHT.svg)
 ![Prediction Accuracy per Round by Actions Against Alternate Behaviour (with 95% Confidence Interval)](figures/mp/mp_prediction_Altern.svg)
 ![Points Earned per Round by Actions Against Alternate Behaviour (with 95% Confidence Interval)](figures/mp/mp_payoff_Altern.svg)
 
+The models exhibit varied approaches to decision-making in the MP game.
+<tt>GPT-4.5</tt> follows a fixed alternating pattern, switching between "Head" and "Tail" each turn, assuming the opponent behaves similarly.
+<tt>Mistral-Small</tt> adopts a reactive strategy, analyzing the  frequency of the  opponent’s past moves and selecting the less common one. On the other hand, <tt>Qwen3</tt>, relies on randomness — choosing moves unpredictably while  presuming the opponent will mimic its choice. <tt>LLaMA3</tt> does not implement a functioning strategy. Overall, these approaches demonstrate simplistic heuristics that may lack credibility and efficiency.
 
 ## Beliefs - RPS
 
-- 
GitLab