Douzero Resnet Bid

该数据集是一个基于斗地主游戏的双边监督学习模型，由斗零Resnet演变而来，旨在评估斗地主游戏中的叫牌策略。该模型使用了1000组农民与地主的牌型进行训练，并通过胜率和平均分数差来评估模型的表现。数据集规模涵盖了4000局游戏，任务是对斗地主游戏中的叫牌策略进行评估。

This dataset is a bilateral supervised learning model for the game of Dou Di Zhu, evolved from DouZero ResNet, and it is designed to evaluate the bidding strategies in Dou Di Zhu. The model is trained using 1000 sets of card hands from both farmers and the landlord, and its performance is evaluated based on win rate and average score difference. The dataset includes 4000 game rounds, and its core task is to assess the bidding strategies in the game of Dou Di Zhu.

搜集汇总

背景与挑战

背景概述

该数据集是Douzero的改进版本，增加了ResNet架构和GPU支持，特别针对Windows平台优化。项目包含完整的训练框架、评估脚本和预训练模型权重，使用Python实现。