ibm-granite/granite-3.3-2b-instruct
Granite-3.3-2B-Instruct is a 2-billion parameter language model developed by IBM, fine-tuned for enhanced reasoning and instruction-following across a 128K context length. It improves upon previous Granite versions in general performance, mathematics, and coding benchmarks. This model supports structured reasoning using and tags, making it suitable for general instruction-following tasks and integration into AI assistants.
Loading preview...
What is Granite-3.3-2B-Instruct?
Granite-3.3-2B-Instruct is a 2-billion parameter instruction-tuned language model developed by the Granite Team at IBM. It features a substantial 128K context length and is built upon the Granite-3.3-2B-Base model. This iteration focuses on significantly improving reasoning and instruction-following capabilities, with notable gains observed in benchmarks for generic performance, mathematics, and coding.
Key Differentiators & Capabilities
- Enhanced Reasoning: The model supports structured reasoning through
<think>and<response>tags, allowing for clear separation between internal thought processes and final outputs. - Improved Benchmarks: It demonstrates significant gains on benchmarks like AlpacaEval-2.0 and Arena-Hard, and shows strong improvements in mathematics (e.g., AIME24 score of 3.28, MATH-500 score of 58.09) and coding (HumanEval score of 80.51).
- Multilingual Support: While primarily English, it supports 12 languages including German, Spanish, French, Japanese, and Chinese, with potential for fine-tuning in other languages.
- Broad Task Handling: Capabilities include thinking, summarization, text classification, extraction, question-answering, RAG, code-related tasks, function-calling, and long-context tasks.
Ideal Use Cases
- General Instruction Following: Designed for a wide range of instruction-following tasks.
- AI Assistants: Suitable for integration into AI assistants across various domains, including business applications.
- Structured Reasoning: Benefits applications requiring explicit internal thought processes before generating a response.
- Long Document Processing: Excels in tasks involving long documents, such as summarization and question-answering over extended contexts.