Falcon-H1 LLMs: Now on Amazon Bedrock & SageMaker
The Technology Innovation Institute (TII) has launched its Falcon-H1 large language models (LLMs) on Amazon Bedrock Marketplace and Amazon SageMaker JumpStart. These six instruction-tuned models (0.5B, 1.5B, 1.5B-Deep, 3B, 7B, and 34B parameters) utilize a hybrid architecture combining attention mechanisms with State Space Models (SSMs) for efficient performance. Key features include multilingual support for 18 languages (scalable to over 100), a context length of up to 256,000, and performance exceeding other LLMs of comparable size. The models are available under the open-source Falcon LLM license, offering cost-effectiveness and energy efficiency. Both Bedrock and SageMaker JumpStart provide streamlined deployment and access, allowing developers to easily integrate Falcon-H1 into their applications. Bedrock offers a model catalog with unified APIs and customizable instance selection, while JumpStart facilitates quick deployment in a secure environment with options for customization and fine-tuning. The article details deployment steps using both platforms, including prerequisites and code examples for interaction via APIs. This collaboration between TII and AWS expands access to advanced AI models globally, enabling users to build cutting-edge generative AI applications.
Amazon's ai automation bedrock provides the infrastructure foundation that makes deploying advanced Falcon-H1 language models more accessible for enterprise applications.
While ChatGPT automation bedrock solutions have dominated the market, Falcon-H1 LLMs now offer compelling alternatives through Amazon's enterprise-grade platforms.
