Pith. sign in

REVIEW 15 cited by

LLMZip: Lossless Text Compression using Large Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2306.04050 v2 pith:JOPDBC2U submitted 2023-06-06 cs.IT cs.CLcs.LGmath.IT

classification cs.ITcs.CLcs.LGmath.IT
keywords compressionlanguagelargelosslesstextciteenglishestimates
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

We provide new estimates of an asymptotic upper bound on the entropy of English using the large language model LLaMA-7B as a predictor for the next token given a window of past tokens. This estimate is significantly smaller than currently available estimates in \cite{cover1978convergent}, \cite{lutati2023focus}. A natural byproduct is an algorithm for lossless compression of English text which combines the prediction from the large language model with a lossless compression scheme. Preliminary results from limited experiments suggest that our scheme outperforms state-of-the-art text compression schemes such as BSC, ZPAQ, and paq8h.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 15 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Information Abundance Paradox: Long-Context Training Undermines Parametric Knowledge

    cs.CL 2026-08 conditional novelty 7.0 of 10

    Long-context training shifts language models from parametric knowledge to context reliance, producing an inverted-U in pretraining performance and context addiction in supervised fine-tuning.

  2. LLM-based Source Code Compression via Thresholded Symbol Ranking

    cs.IT 2026-07 conditional novelty 6.0 of 10

    Bounding LLM next-token ranks to T=63 (or T=1) with escaped exceptions beats unbounded LLM ranking by up to 37% ratio and 40% speed on source code, and beats general-purpose compressors by up to 82% ratio.

  3. StateSMix: Online Lossless Compression via Mamba State Space Models and Sparse N-gram Context Mixing

    cs.LG 2026-04 conditional novelty 6.0 of 10

    An online-trained Mamba SSM mixed with sparse n-gram logit bias beats xz on enwik8 up to 10MB without pre-trained weights or a GPU.

  4. LCIRC: A Recurrent Compression Approach for Efficient Long-form Context and Query Dependent Modeling in LLMs

    cs.CL 2025-02 conditional novelty 6.0 of 10

    A frozen LLM can process very long contexts by recurrently compressing them with a Perceiver and injecting the compressed memory through gated cross-attention, with query-dependent compression boosting QA performance.

  5. Separate Source Channel Coding Is Still What You Need: An LLM-based Rethinking

    cs.IT 2025-01 conditional novelty 6.0 of 10

    LLM-based arithmetic coding plus ECCT-enhanced LDPC decoding makes separate source and channel coding competitive with, and in these tests superior to, joint source-channel coding for text under a total-energy comparison.

  6. LMM-driven Semantic Image-Text Coding for Ultra Low-bitrate Learned Image Compression

    eess.IV 2024-11 conditional novelty 6.0 of 10

    A single LMM generates and compresses image captions for ultra-low-bitrate learned image compression, improving LPIPS BD-rate by 41.58% over MISC.

  7. PMKLC: Parallel Multi-Knowledge Learning-based Lossless Compression for Large-Scale Genomics Database

    cs.LG 2025-07 conditional novelty 5.0 of 10

    PMKLC is a parallel multi-knowledge learning-based genomic compressor that achieves the best average compression ratio and throughput among tested baselines, with gains over the strongest baseline under 1%.

  8. DualComp: End-to-End Learning of a Unified Dual-Modality Lossless Compressor

    cs.CV 2025-05 conditional novelty 5.0 of 10

    A unified RWKV-based model compresses both images and text losslessly with a shared vocabulary, modality-specific contextual layers, and mixture-of-experts routing, matching or improving on much larger SOTA compressor...

  9. ToDMA: Large Model-Driven Massive Token Communications for Semantic Multiple Access

    cs.IT 2025-05 conditional novelty 5.0 of 10

    ToDMA lets many devices share one wireless channel by transmitting token indices from a common codebook, recovering collisions with compressed sensing and masked-token prediction from pretrained models.

  10. GPT as a Monte Carlo Language Tree: A Probabilistic Perspective

    cs.CL 2025-01 conditional novelty 5.0 of 10

    GPT models trained on a corpus are shown to approximate a tree of empirical next-token probabilities derived from the same corpus, with alignment increasing with model size.

  11. L3TC: Leveraging RWKV for Learned Lossless Low-Complexity Text Compression

    cs.CL 2024-12 conditional novelty 5.0 of 10

    L3TC, a 3.2M-parameter RWKV-based compressor with an outlier-bypass tokenizer, saves roughly 50% of gzip's bitrate on enwik9 and decodes at megabyte-per-second rates.

  12. Channel-Agnostic Semantic Compression for Bandwidth-Limited Visual Communication

    cs.IT 2026-08 conditional novelty 4.0 of 10

    RQ-NAC, a residual-quantized image codec with n-gram arithmetic coding, reports 671 times compression over uncompressed dashcam frames at modest reconstruction quality (SSIM 0.74, PSNR 23.6 dB).

  13. Joint Lossless Compression and Steganography for Medical Images via Large Language Models

    eess.IV 2025-08 reject novelty 4.0 of 10

    A joint lossless compression and steganography framework for medical images that splits bit planes into a VAE-compressed global part and an LLM-compressed local part, embedding secret messages in the local part.

  14. EDPC: Accelerating Lossless Compression via Lightweight Probability Models and Decoupled Parallel Dataflow

    cs.IT 2025-07 conditional novelty 4.0 of 10

    EDPC reports state-of-the-art lossless compression ratios across text, image, float, and benchmark datasets while running about 2.7x faster than the strongest prior learned compressor, PAC.

  15. An Enhanced Text Compression Approach Using Transformer-based Language Models

    cs.CL 2024-12 reject novelty 3.0 of 10

    Removing vowels before LZW compression yields high compression ratios, but the resulting text cannot be restored without a large transformer, making the claimed state-of-the-art comparison unfair.

Pith tools