This is a small, randomly initialized Granite MoE hybrid model hosted on Hugging Face for internal testing of the Optimum Intel library. It provides a minimal test case for inference optimization on Intel hardware using the transformers and Optimum stack. The model includes tokenizer configuration and chat templates for basic text generation tasks and is intended for developers integrating or benchmarking Intel-accelerated AI workflows.
In the Foundation models & chat space, Tiny Random Granitemoehybrid takes a focused approach. It focuses on testing and validating Intel-optimized inference for small Mixture-of-Experts language models. Tiny Random Granitemoehybrid is an open-source project aimed at AI developers, ML engineers. The project is open source (Open Source). The product ships for the web and API.
Behind Tiny Random Granitemoehybrid is Hugging Face, based in the United States, and the product first shipped in 2024. Among its 3 catalogued features are mixture of Experts, Model Testing, and Hugging Face Integration.
Latest indexed changes and source events
optimum-intel-internal-testing/tiny-random-granitemoehybrid verified by the PulseGate indexer
Other apps tracked under the same category.
