This preprint proposes a unified post-training quantization framework for large language models. It derives a Hessian-guided approximation of the output change caused by joint weight and activation quantization, using a tractable Fisher approximation. Quantization ranges, linear transformations and weight rounding are optimized together under one output-sensitivity objective. The authors report gains over state-of-the-art transformation and rounding methods in channel-wise and block-wise settings.
This paper measures the energy cost of speculative decoding in SGLang with CUDA graphs on H100 and H200 GPUs, at batch sizes up to 64 across 5 corpora. Savings depend on drafted-token acceptance, which is lower on conversation than code and lower on Romansh than English. Savings shrink as batch size grows, and block drafters can push verification into the compute-bound regime, where speculation raises energy per token.
SGLangCUDALLM+4
NVIDIAarXivRomansh
inference, researchΒ· Added 31 Oct 2026, published 29 Oct 2026
Kolesnikov and Novoselov give block basis reduction (BKZ) algorithms for linear codes that need polynomial or quasi-polynomial numbers of calls to a minimum weight codeword oracle at fixed block size. The earlier Ghentiyala and Stephens-Davidowitz bounds are exponential. The paper adds a truncated BKZ and a DeepBKZ variant, with an implementation for block sizes 2 to 24 and approximation-factor estimates on random codes.
DICE is a blockchain-powered federated learning (BCFL) algorithm for customizing large pretrained models among untrusted Web3 participants. It combines dynamic progressive learning and model slice verification, so hardware-limited participants can train part of the model, with a tree-structured committee consensus protocol and an efficiency-driven incentive scheme to scale to more data holders. Experiments cover vision and language models.
Beyond the Event Horizon of Keccak extends a 2011 observation that a path in the Keccak sponge from a zero-capacity state back to one yields infinite second-preimage families. Three theorems cover loops from any capacity value x, k-block loops and cycles, and Joux-style multicollisions with search cost independent of k. Verified examples run on Keccak-f[200] and Keccak-f[1600] with c=32; it also proposes bridge-based protocols.
FlatCake is a KV cache storage architecture for LLM serving that treats host DRAM and NVMe SSD as peer media in one 64-bit address space, instead of a DRAM-over-SSD tier. Blocks are placed once, in session-contiguous, layer-major order, and streamed via GPUDirect Storage. On a Claude Code trace with DeepSeek-V4-Flash it holds a 94% hit rate (Mooncake: 64%) and cuts peak-load mean TTFT 1.6x.
GPUDirect StorageNVMe SSDLLM+3
FlatCakeMooncakeMooncake-GDS+3
inference, researchΒ· Added 30 Oct 2026, published 28 Oct 2026
LINE-Break is a security analysis of Letter Sealing v2, the custom end-to-end encryption protocol in the LINE messenger. A TLS man-in-the-middle attacker or malicious server can replay, reorder and block messages undetected, and colluding users can forge message authorship in one-to-one and group chats. Stickers and URL previews leak plaintext. The attacks were verified on the real iOS app and an independent implementation.
LINEiOS
Letter SealingLetter Sealing v2Diego F. Aranha+1
Mobile Security, Offensive SecurityΒ· Published and added 26 Oct 2026
ZMAC+ is a tweakable-block-cipher MAC (List and Nandi, ToSC 2017) with claimed beyond-birthday-bound security. This ePrint paper (2026/2647) shows a tag-swap attack: for two-block tags, a distinguisher and a fresh-message forgery with constant success probability using O(2^{n/2}) queries in the ideal-TBC model. It also corrects a counting error in the published Lemma 4 and proposes a repaired variant with an adaptive PRF bound of order q^2/2^{n+min(n,t)}.
A Memory-efficient Implementation of Wiedemann XL is an ePrint paper (2026/2641) giving the first efficient on-the-fly implementation of XL with block Wiedemann for the multivariate quadratic problem. It builds on the CCNY SSE implementation, removes column-index tables, and computes column indices with O(n^2) instead of O(n^3) look-ups. It runs in comparable BW1 time with up to five times less memory, and an AVX2 port covers F256, the main field of UOV.
Provable Security of SHA-256-Based LMS Revisited in the Ideal-Cipher Model is an October 2026 ePrint paper by Chun Guo, Hao Cheng, Bart Preneel and Meiqin Wang. It shows forgery attacks on LMS_SHA256_M32 and LMS_SHA256_M24 that exploit Merkle-DamgΓ₯rd structure, so the Katz and Fluhrer random-oracle bounds do not apply. It proves ideal-cipher-model bounds: about 256 minus log2 of the block count for M32, and a minimum of two terms for M24.
Preprint (IACR ePrint 2026/2611) proposing a federated learning defense that replaces gradient detection with zero-knowledge-proof sample-level audits of client datasets. A staking mechanism forces attackers to train on discovered poisoned samples with flipped labels, producing counter-gradients that remove the backdoor without knowing the trigger. ZKP commitments lock the training data, which blocks adaptive attacks and gradient tampering. On CIFAR-10 under Dir(0.1) non-IID, it holds attack success rate below 6% against distributed backdoor and edge-case attacks.
MPQ (Message-Passing Quantization) is a post-training quantization method that treats row-wise PTQ as posterior inference over discrete codebooks. It keeps soft posteriors across weight blocks using vector approximate message passing, then refines a hard-decision initialization with exact block coordinate descent. Tested at 2-bit lattice and 3- and 4-bit scalar quantization on Llama, Qwen and Mistral models, it improves perplexity and zero-shot accuracy over GPTQ-style baselines.
GPTQLDLQvector approximate message passing+5
arXiv
research, inferenceΒ· Added 26 Oct 2026, published 23 Oct 2026
Decode-time provenance binding is a defense for Model Context Protocol agent pipelines against cross-modal trust laundering, where instructions hidden in an email-signature image via LSB, DCT or metadata steganography get decoded by a tool. It combines HMAC-authenticated trust labels, a provenance ledger and context-taint tracking to block critical tool calls. On a real MCP testbed with 24 adaptive payloads, attack success fell from 100% undefended (62.5% for keyword and regex filters) to 0%.
TTR-Compressor is a sequence compression method that treats input tokens as a test-time regression problem. The regression is learned end-to-end by backpropagation, and its coefficients are solved in closed form in the forward pass to give the compressed representation. As a block-wise sparse attention compressor it improves RULER over attention-based compressors by a large margin. As a visual token merger for multimodal LLMs it beats the MLP merger and matches it with 55.6% fewer visual tokens.
TransformerLLMvision-language model
TTR-CompressorRULERarXiv
research, multimodalΒ· Added 24 Oct 2026, published 22 Oct 2026
SARIA (Selective Admission and Reload for Inference Acceleration) is a KV caching system that jointly manages DRAM and SSD for multi-turn LLM agents. It persists only blocks likely to be reused beyond DRAM residency, promotes SSD-hit blocks into DRAM for same-session reuse, and paces DRAM reads with a transfer coordinator. Built on LMCache MP, it cuts SSD writes by 57.1 to 58.9% versus full duplicate caching and mean TTFT by 42.3 to 79.7% versus DRAM-only caching.
LMCache MPLLMSSD+1
SARIALMCacheTraceLab+2
inference, researchΒ· Added 23 Oct 2026, published 22 Oct 2026
SoK on design-level backdoors in block ciphers. It places four published designs (Rijmen and Preneel, Paterson, Bannier and Filiol, Peyrin and Wang) in one taxonomy and gives a sound algorithm that recovers a partition trapdoor from S-box tables. It attacks KG-DES, BEA-1 and LowMC-M: KG-DES key recovery takes minutes, and the LowMC-M MALICIOUS backdoor yields the key from 35 to 409 queries. Trapdoors hidden in linear layers leave no trace, so the authors recommend deriving constants verifiably from a public seed.
Sign Here is an arXiv paper (October 2026) testing whether building packages from source, instead of pulling from registries, prevents malware. It aggregates 47,066 malicious npm and PyPI packages from five datasets and indexes 3.9 million npm and 764 thousand PyPI packages. Checking for an attributed source blocks 87.7 to 99.9 percent of malware depending on vetting level, but only 47.4 percent of benign packages attribute a buildable source.
npmPyPI
Shai-HuludAxios
Supply ChainΒ· Added 24 Oct 2026, published 20 Oct 2026
Lafa (Large flow language model) is a blockwise continuous flow language model scaled to 8B parameters, with causal attention on the prefix and bidirectional, asynchronous flow within each block. It is adapted from pretrained autoregressive models and trained at 0.5B, 7B and 8B with under 30B tokens. A self-speculative decoding scheme verifies flow drafts autoregressively in one model, giving 2.54x the throughput of an AR model at comparable quality. Flow maps allow one- and few-step generation.
flow matchingdiffusion modelTransformer+1
Lafa
research, modelsΒ· Added 22 Oct 2026, published 20 Oct 2026
Preprint by Jiseung Kim, Minkyu Kim, Changmin Lee and Hochang Lee on the Regular Syndrome Decoding (RSD) problem. It presents a hybrid-algebraic attack that balances zero-coordinate guesses and syndrome-based eliminations across regular blocks. Parameter sets in Ferret, Wolverine and committed VOLE fall below 128-bit targets by up to 67 bits, and the revised Liu et al. (EUROCRYPT 2024) PCG sets cost under 2^111 field operations.
Preprint 2026/2563 presents a refined hybrid-algebraic attack on Regular Syndrome Decoding (RSD) that balances zero-coordinate guesses and syndrome-based eliminations across all regular blocks. It puts at least one parameter set each from Ferret (CCS 2020), Wolverine (IEEE S&P 2021) and committed VOLE (CCS 2025) below the 128-bit target, with losses up to 67 bits. PCG attack costs improve by up to 29.77 bits (non-binary) and 25.91 bits (binary), and the revised Liu et al. (EUROCRYPT 2024) 128-bit sets fall below 2^111 field operations.
Tandem is a runtime for serving hybrid linear-attention GUI agents on mobile NPUs. It saves recurrent state where consecutive prompts diverge, keeps most prefill on the matrix unit, uses a verification kernel that reads each weight block once, and releases NPU power votes between bursts. On a Snapdragon 8 Elite it cuts per-step latency 30% for text screen descriptions and 16% for screenshots, cuts energy per text step by 33%, and lowers loaded-model standby power from 0.76 to 0.22 W.
SnapdragonNPUhybrid linear attention
TandemSnapdragon 8 ElitearXiv
inference, agents, researchΒ· Added 21 Oct 2026, published 19 Oct 2026
SpanGuard is an indirect prompt injection (IPI) guardrail that splits detection into two steps. Lightweight unsupervised segmentation first locates candidate instruction blocks in tool outputs, then a small LLM judges whether each block is authorized by the user's request. The authors report a 0.32% attack success rate with negligible benign utility loss, and the guardrail still works with substantially smaller models.
LLM
indirect prompt injectionprompt injection
AI SecurityΒ· Added 21 Oct 2026, published 19 Oct 2026
IISFA (Ineffective Instruction-Skip Fault Attack) targets bitsliced block cipher implementations protected by redundancy countermeasures. Skipping one bitslice instruction leaves the result unchanged only when all m S-box inputs fall in an ineffective set, so each released ciphertext restricts all m subkeys. The authors give known- and chosen-plaintext key-recovery procedures and expected injection counts, and recover keys in every trial on GIFT, SKINNY and PIPO implementations on an 8-bit AVR microcontroller using clock glitches.
ePrint 2026/2527 by Paola de Perthuis and Filip TrenkiΔ analyses module-BKZ attacks on NTRU+, the South Korean KEM standard. It introduces relative canonical and coefficient embeddings and shows they agree up to scaling exactly when rad(c)/rad(β) is 1 or 2. Simulation-based estimates show the best module attack saves 4 to 10 blocksizes, costing NTRU+ keys at most three bits of Core-SVP security. NTRU+ parameters stay safe in the overstretched regime.
This ePrint paper (2026/2525, October 2026) by Brendan Funk and Jiafeng Xie presents the first hardware accelerator for FAEST, an MPC-in-the-head post-quantum signature in NIST's Round 3 additional-signature process. It analyses FAEST's compute-heavy operations, builds them into hardware blocks, and adds an L1-EM-fast variant. The authors report at least 98% lower equivalent area-delay product than the recent complete SDitH design at the same security level.
FAESTFPGA
NISTSDitHMPC-in-the-head+2
Hardware & RFΒ· Added 19 Oct 2026, published 18 Oct 2026
Cryptanalysis of the Eevee forkcipher-based AEAD modes Umbreon and Espeon, which claim online AE security under nonce reuse. Associated-data correlations give probability-one non-empty-message forgeries with three encryption queries and one verification query. The paper adds an affine treatment of partial final blocks (n-r+3 queries, or n-r+2 with adaptive choices). It also checks the published Umbreon attack with the official ForkSkinny primitive and reports a two-query attack accepted by the unmodified public C modes.
FuTransformer is a CUDA implementation of a Llama-style Transformer block for NVIDIA Hopper GPUs. It uses FP8 prefill with fused GEMM epilogues and INT4 weights plus an INT4 KV-cache copy for decode, with a block-diagonal Hadamard transform to cut cache quantization error. On one H100, prefill is up to 1.48x faster than compiled PyTorch FP8, and INT4 decode is 1.3-2.4x faster than the faster PyTorch baseline. Llama-3.1-8B decode takes 2.8 ms per step at batch one, versus 8.1 ms for compiled transformers. WikiText-2 perplexity rises from 6.24 to 7.23 with round-to-nearest INT4 weights, or 6.99 with GPTQ.
CUDAPyTorchTransformer+2
NVIDIACODALlama-3.1-8B+5
inference, researchΒ· Added 20 Oct 2026, published 16 Oct 2026
DLite is a history-free speculative decoding framework for agentic LLM workloads. It reuses the target model's contextual hidden states and predicts locally without keeping the draft model's historical KV cache, which avoids the up to 80% acceptance-length drop seen with long, multi-turn contexts. A lightweight sequential head restores intra-block causal dependencies, and a two-stage training strategy is used. It reaches 2-6 acceptance length and beats DSpark by up to 70%.
LLM
DLiteDSparkarXiv
inference, researchΒ· Added 20 Oct 2026, published 16 Oct 2026
ThetreStageAI is a retrieval-augmented framework for structured Bengali theatre production planning. It keeps separate scene, actor-blocking and lighting views, embeds Bengali text with BGE-M3, indexes it in view-specific Qdrant collections, and has Gemini generate schema-constrained JSON, with validation and at most one repair attempt. The authors build a synthetic dataset of 500 records (1,500 retrieval documents) and compare single-combined retrieval with multi-view retrieval. Code and data are public.
RAGGeminiBGE-M3+1
ThetreStageAI
llm-apps, researchΒ· Added 20 Oct 2026, published 16 Oct 2026
Velox is a training-free attention router for chunked video-diffusion world models. It uses camera frustum slabs with depth intervals from streaming geometry to pick which historical token blocks to attend to, instead of letting the KV cache grow with rollout length. Frustum overlap softly modulates each layer's model-derived block importance, and uncovered regions keep the model's own routing. The fixed block granularity suits FlashAttention-style tiled execution, and slab construction runs asynchronously with denoising on the same H200. At chunk 48 it speeds up attention 4.36Γ and denoising 2.43Γ, and on the 158-case WBench navigation split it stays within 0.15 average points of dense LingBot-Fast-Cam.
diffusion modelFlashAttentionCUDA
VeloxLingBot-Fast-CamWBench+1
research, multimodal, inferenceΒ· Added 17 Oct 2026, published 15 Oct 2026
arxiv.orgpaperπ§ AI
