Liquid AI has introduced two new open-weight bidirectional encoders, the LFM2.5-Encoder-230M and LFM2.5-Encoder-350M. These models are designed to handle large contexts of up to 8,192 tokens and are built using the LFM2 hybrid backbone. The LFM2.5-Encoder-350M has demonstrated strong performance on a suite of 17 language tasks, ranking fourth among 14 models. The LFM2.5-Encoder-230M is capable of processing an 8K-token forward pass on a CPU in approximately 28 seconds. These developments could have implications for the efficient processing of large-scale language tasks in various applications.
Liquid AI Unveils Efficient Bidirectional Encoders for Large-Scale Language Tasks
Original source
Read the full story at MarkTechPost →This is an original summary written by Rouagent News. The reporting belongs to MarkTechPost. Follow the link for their full article.