The Avalanche framework evaluates machine learning models' resilience and security against adversarial assaults. Its major goal is to assess the resilience of these models by simulating various attack scenarios and determining their efficacy in compromising the model's functionality. Avalanche offers a complete platform for researchers and practitioners to compare alternative defense methods and assess their effectiveness in minimizing hostile threats. It provides a uniform technique for carrying out experiments and comparing findings across models and datasets, hence boosting advances in adversarial machine learning research. Finally, the framework intends to encourage the creation of more resilient and secure machine learning models capable of overcoming real$-$world adversarial situations.