AI model library

data-mixture-doremi-1b is a 1B-parameter open-weight text language model from Sea AI Lab, trained using the DoReMi data mixture approach and released under the MIT license; it is primarily of interest for data mixture optimization research.

An open-weight 1B-scale language model released in July 2024 by Singapore's Sea AI Lab under the MIT license. Trained on a data mixture of The Pile and Common Crawl, it is intended for text generation research.

Sea AI Lab's 1B parameter language model with chat capabilities, released under MIT license.

Sea AI Lab's 1B parameter text generation model with chat capability, released under MIT license.

Sea AI Lab's 3.1B parameter language model supporting 32k context length, specialized in long-text conversations.

Sea AI Lab's language model with 43k vocabulary supporting chat capability.

Sea AI Lab's 14.2B parameter open-source LLM with chat capabilities.

Sea AI Lab's 0.6B parameter language model with chat capabilities, released under Apache 2.0 license.

Sea AI Lab's 4B parameter open-source LLM with chat capability, released under Apache 2.0 license.

Sea AI Lab's 1.8B parameter open-source LLM with chat capabilities.

Sea AI Lab's 7B parameter open-source LLM with chat capabilities, released under Apache 2.0 license.