dynatrace-oss/llama-embed-mamba2-7b
Sentence Similarity • 7B • Updated • 45 • 3
Embed Mamba2
Text embedding models based on Mamba2 with linear-time and constant-memory inference through vertical chunking.
-
-
dynatrace-oss/llama-embed-mamba2-1.3b
Sentence Similarity • 1B • Updated • 115 • 5 -
Linear-Time and Constant-Memory Text Embeddings Based on Recurrent Language Models
Paper • 2604.18199 • Published
dynatrace-oss/chunkable-mamba2
UpdatedNote Adjusted Mamba2 model code and kernels to support chunked inference with recurrent state passing.
dynatrace-oss/chunkable-sentence-transformer
UpdatedNote Chunkable transformer and last index pooling modules for sentence transformers enabling constant-memory encoding.