KV-Rescue: Recovering Reasoning Language Model KV Eviction Loss via Stepwise Interleaving

KV-cache eviction caps the memory cost of long reasoning traces but is inherently lossy because the model decodes from a partial view of its history. Under aggressive budgets, this not only lowers accuracy but can also cause runaway degeneration, where the model produces incoherent or repetitive tokens until reaching the length limit. We characterize much of this loss as an information gapf caused by missing context, rather than a capability gap caused by limited model capacity. An evicted 7B model and a full-context 1.5B model make complementary errors, and an oracle choice between their answers recovers 79% of the accuracy gap to the full-KV 7B model. Based on this observation, we propose KV-Rescue, a training-free inference framework that bridges the information gap introduced by KV eviction using a lightweight full-context helper. KV-Rescue interleaves reasoning steps from the two models into a shared trajectory. An online detector uses entropy and compressibility to terminate the generation of incoherent or repetitive base-model candidates early. Across five math benchmarks with Qwen2.5-Math 7B and 72B, KV-Rescue recovers an average of 87% of the accuracy lost to eviction at eviction budget B=64. A decode-cost analysis further shows that preventing runaway degeneration cuts base-model token generation by 43% on average.

Output-Aware Rotation for INT2 KV-Cache Quantization

The key-value (KV) cache has become a major memory and bandwidth bottleneck in long-context large language model inference, making ultra-low-bit quantization increasingly important. However, existing rotation-based INT2 methods optimize cache statistics or proxy errors before the complete attention readout, even though the model is ultimately affected by the error propagated through attention and the output projection $W_O$. To address this mismatch, we propose \textit{OptR}, an output-aware rotation method that minimizes post-$W_O$ attention-output error. OptR decomposes the post-$W_O$ attention-output error into key- and value-induced terms and learns per-head orthogonal corrections through the full INT2 quantization and attention path. OptR further applies an attention-equivalent key reparameterization to reduce large channel-wise offsets without changing the softmax distribution. Across three models and five reasoning and coding benchmarks, OptR consistently improves both QuaRot and OSCAR and strengthens long-context retrieval, while preserving the paged KV-cache format with negligible inference overhead.

EntropyCache: Decoded Token Entropy Guided KV Caching for Diffusion Language Models

Diffusion-based large language models (dLLMs) rely on bidirectional attention, which prevents lossless KV caching and requires a full forward pass at every denoising step. Existing approximate KV caching methods reduce this cost by selectively updating cached states, but their decision overhead scales with context length or model depth. We propose EntropyCache, a training-free KV caching method that uses the maximum entropy of newly decoded token distributions as a constant-cost signal for deciding when to recompute. Our design is grounded in two empirical observations: (1) decoded token entropy correlates with KV cache drift, providing a cheap proxy for cache staleness, and (2) feature volatility of decoded tokens persists for multiple steps after unmasking, motivating recomputation of the $k$ most recently decoded tokens. The skip-or-recompute decision requires only $O(V)$ computation per step, independent of context length and model scale. Experiments on LLaDA-8B-Instruct and Dream-7B-Instruct show that EntropyCache achieves $15.2\times$-$26.4\times$ speedup on standard benchmarks and $22.4\times$-$24.1\times$ on chain-of-thought benchmarks, with competitive accuracy and decision overhead accounting for only $0.5\%$ of inference time. Code is available at https://github.com/mscheong01/EntropyCache.

InFerActive: Towards Scalable Human Evaluation of Large Language Models through Interactive Inference

Human evaluation remains the gold standard for evaluating outputs of Large Language Models (LLMs). The current evaluation paradigm reviews numerous individual responses, leading to significant scalability challenges. LLM outputs can be more efficiently represented as a tree structure, reflecting their autoregressive generation process and stochastic token selection. However, conventional tree visualization cannot scale to the exponentially large trees generated by modern sampling methods of LLMs. To address this problem, we present InFerActive, an interactive inference system for scalable human evaluation. InFerActive enables on-demand exploration through probability-based filtering and evaluation features, while bridging the semantic gap between computational tokens and human-readable text through adaptive visualization techniques. Through a technical evaluation and user study (N=12), we demonstrate that InFerActive significantly improves evaluation efficiency and enables more comprehensive assessment of model behavior. We further conduct expert case studies that demonstrate InFerActive's practical applicability and potential for transforming LLM evaluation workflows.