The paper introduces a novel patching strategy for byte‑level language models, called novelty patching, which starts a new patch when upcoming bytes are unlikely to repeat recent context, inspired by LZ77 compression. Experiments on a 1.9 GB math corpus show that novelty patching improves final‑answer accuracy from 47 % to 72 % at 12 M parameters and raises computed‑result accuracy from 21 % to 57 % at 50 M parameters, while having limited impact on prose and source code. The study also analyzes patch boundary distributions and reports several failed approaches, such as typed uncertainty and context‑aware novelty prediction.
whyItMatters":"The findings demonstrate that a simple, data‑driven patching rule can significantly enhance the performance of byte‑level models on tasks involving computed results, offering a practical improvement for large language models that process raw text without explicit labels."
By Nicol\'as Vera Z\'u\~niga
Byte-level language models such as the Byte Latent Transformer (BLT) run their large global model once per patch and start a patch where a small model's next-byte entropy is high. This rule rarely sta...
The study re‑examines a reported advantage of a routed ternary (1.58‑bit) language model over a full‑precision transformer at 60K parameters. By running controlled experiments with multiple seeds and a fixed training recipe, the authors find that the apparent benefit largely stems from the choice of baseline model shape rather than the ternary architecture itself. While the routed model does outperform other shapes at a larger 130M‑byte budget, its advantage diminishes when a plain gated diagonal‑SSM block is used, and the ternary penalty varies with architecture and quantization details.
By Gautam Veldanda
RaReCache is a framework that allows a large target language model to decode accurately from a KV cache prefilled by a much smaller source model by selectively recomputing only critical token positions. It identifies these positions using a rank disagreement metric that scores tokens based on the energy of their mapped KV in output directions weakly supported by calibration data. Across two model families and five benchmarks, RaReCache achieves up to a 3.04× prefill speedup and significantly reduces prefill latency while maintaining 95–99% of target accuracy with a 30–40% recompute budget.
By Sreetama Sarkar, Saptarshi Mitra, Sitao Huang, Souvik Kundu, Peter A. Beerel
Delta N-gram introduces a forward‑only write path into a language model’s native hashed N‑gram table, preserving the pretrained table while adding evidence through a sparse patch Δ. The method uses a 4.1 M‑parameter writer that emits gated targets per event and applies a block‑averaged delta rule to keep each row a convex combination of its old value and the new targets. Experiments on a 0.6 B prototype and Qwen3.8‑Flash‑Next show that the delta writer maintains more document‑specific memory and reduces drift compared to online gradient writes, achieving stable performance over thousands of writes.
whyItMatters":"The approach allows continual updates to a language model’s memory without altering its pretrained parameters, enabling efficient, stable, and document‑specific knowledge retention during inference."
By Hao Wu
The study measured the impact of a single training example on a GPT‑2 model by running 24 counterfactual experiments. 32 models were trained from scratch on OpenWebText, and at a specific training step a single batch row was replaced with a 194‑token passage under three conditions (fluent prose, fabricated subject, random characters) or left unchanged. Results showed that the passage was learned from one exposure and decayed, with measurable differences in cross‑entropy up to 50 steps after injection but no lasting effect at the final step.
By Zachary Speck, Asa Shepard
arXiv:2607. 14431v1 Announce Type: cross Abstract: We report a way to make a frozen small language model both more capable and dramatically cheaper at once, without changing any weights.
By Sietse Schelpe
Nexus introduces a depth‑adaptive KV‑cache splicing and retrieval‑decoupled tool routing mechanism for agentic large language models that reduces the time‑to‑first‑token (TTFT) by decoupling tool routing from the expensive schema re‑encoding step. It uses an INT8 semantic lookaside buffer to select tools via retrieval and generates arguments from a compressed textual signature, maintaining about 89% routing accuracy even as the tool registry scales to 250 tools. Additionally, Nexus can splice compiled schema KV blocks into the live context, repairing the seam with a depth‑adaptive suffix redecode when rotary position embedding drift exceeds a threshold, ensuring output fidelity while achieving up to 1.7× TTFT speedup at moderate depth.
By Mustafa Arslan
EntropyCache is a training‑free KV caching technique for diffusion‑based large language models that uses the maximum entropy of newly decoded token distributions to decide when to recompute cached states. By leveraging two empirical observations—entropy correlating with cache drift and feature volatility persisting across multiple steps—EntropyCache makes a skip‑or‑recompute decision with only O(V) computation per step, independent of context length or model size. Experiments on LLaDA‑8B‑Instruct and Dream‑7B‑Instruct demonstrate speedups of 15.2×–26.4× on standard benchmarks and 22.4×–24.1× on chain‑of‑thought benchmarks while maintaining competitive accuracy and keeping decision overhead below 0.5% of inference time.
By Minsoo Cheong, Donghyun Son, Woosang Lim, Sungjoo Yoo
The paper investigates whether training Mixture-of-Experts (MoE) routers can improve memory‑bandwidth locality on consumer GPUs. Using a new zero‑surgery telemetry tool, the authors measure that a large Qwen3‑235B model is bottlenecked by disk‑based expert access, and that an LRU cache can serve a majority of requests. They pre‑register experiments training 137 M‑parameter MoE models with locality‑aware losses, finding that while cache misses can drop up to 60 % (99 % static‑pin hit rate), every configuration fails to meet a strict 1 % perplexity threshold, indicating a tight coupling between cache efficiency and model quality.
By Shriniwas Ramesh Suram
BellowsKV introduces a training‑free approach that maps the hidden states of multiple adjacent layers in a large language model to a single shared latent representation per token, allowing each layer to retrieve its own keys and values via a layer‑specific projection. The method retains the latent at full dimension and offers two compression strategies: BellowsKV‑Quant, which stores the latent at low precision, and BellowsKV‑Evict, which makes a shared eviction decision per block. Experiments on Llama‑3.1‑8B, Mistral‑7B, and Qwen2.5‑7B show that BellowsKV‑Quant preserves about 99% of full‑cache accuracy at 3.6× compression and 97‑98% at 13‑14×, outperforming existing cross‑layer methods that drop to at most 84% accuracy at 16× compression.
By Jiamu Zhang, Hanjie Chen
The paper investigates whether small models distilled from larger ones behave similarly when using byte versus token tokenization. It introduces two methods—Marginalize‑It (approximate) and End‑Of‑Token (exact)—to convert token logits to byte logits, and conducts a large‑scale study on decoder‑only dense transformers ranging from 1 billion to 1 trillion bytes of data. Results show that while token‑based models excel early, byte‑based models eventually surpass them with more compute, achieving higher performance ceilings, greater data efficiency, and lower logit storage costs.
By Kalyani Marathe, Artidoro Pagnoni, Tomasz Limisiewicz, Margaret Li, Mike Lewis, Luke Zettlemoyer, Srinivasan Iyer