Search papers, labs, and topics across Lattice.
StateSMix, a novel lossless compressor, combines an online-trained Mamba-style State Space Model (SSM) with sparse n-gram context mixing and arithmetic coding, learning token-by-token without pre-trained weights or GPUs. By integrating a continuously-updated SSM for probability estimation with sparse n-gram hash tables for exact local and long-range pattern memorization, StateSMix achieves state-of-the-art compression ratios on the enwik8 benchmark, surpassing xz -9e by up to 8.7% on 1MB files. Ablation studies reveal the SSM as the primary driver of compression, with n-grams providing a complementary boost.
Mamba state space models can beat standard compression algorithms like LZMA2, even when trained online from scratch on a single file.
We present StateSMix, a fully self-contained lossless compressor that couples an online-trained Mamba-style State Space Model (SSM) with sparse n-gram context mixing and arithmetic coding. The model is initialised from scratch and trained token-by-token on the file being compressed, requiring no pre-trained weights, no GPU, and no external dependencies. The SSM (DM=32, NL=2, approximately 120K active parameters per file) provides a continuously-updated probability estimate over BPE tokens, while nine sparse n-gram hash tables (bigram through 32-gram, 16M slots each) add exact local and long-range pattern memorisation via a softmax-invariant logit-bias mechanism that updates only non-zero-count tokens. An entropy-adaptive scaling mechanism modulates the n-gram contribution based on the SSM's predictive confidence, preventing over-correction when the neural model is already well-calibrated. On the standard enwik8 benchmark, StateSMix achieves 2.123 bpb on 1 MB, 2.149 bpb on 3 MB, and 2.162 bpb on 10 MB, beating xz -9e (LZMA2) by 8.7%, 5.4%, and 0.7% respectively. Ablation experiments establish the SSM as the dominant compression engine: it alone accounts for a 46.6% size reduction over a frequency-count baseline and beats xz without any n-gram component, while n-gram tables provide a complementary 4.1% gain through exact context memorisation. OpenMP parallelisation of the training loop yields 1.9x speedup on 4 cores. The system is implemented in pure C with AVX2 SIMD and processes approximately 2,000 tokens per second on commodity x86-64 hardware.
