Related papers: Non-Monotonic Latency in Apple MPS Decoding: KV Ca…

Large language models are increasingly applied to multi-document and long-form inputs, yet long-context inference remains memory- and noise-inefficient. Key-value (KV) caching scales linearly with context length, while external retrieval…

Large language models (LLMs) have recently shown remarkable performance across a wide range of tasks. However, the substantial number of parameters in LLMs contributes to significant latency during model inference. This is particularly…

Steering large language models (LLMs) is usually done by either instruction prompting or activation steering. Prompting often gives strong control, but caches guidance tokens at every layer and can clutter long interactions; activation…

Key-Value (KV) cache remains a major bottleneck for deploying Large Language Models (LLMs) in long-generation tasks. Prior work often applies uniform compression across both prefill and decoding caches, but compressing the prefill cache…

KV cache quantization reduces the memory cost of long-context LLM inference, but introduces approximation error that is typically validated only empirically. Existing systems rely on average-case robustness, with no mechanism to detect or…

Transformers process tokens in parallel but are temporally shallow: at position $t$, each layer attends to key-value pairs computed based on the previous layer, yielding a depth capped by the number of layers. Recurrent models offer…

Transformers have become central to natural language processing and large language models, but their deployment at scale faces three major challenges. First, the attention mechanism requires massive matrix multiplications and frequent…

Existing captioning models often adopt the encoder-decoder architecture, where the decoder uses autoregressive decoding to generate captions, such that each token is generated sequentially given the preceding generated tokens. However,…

Decoding with autoregressive large language models (LLMs) traditionally occurs sequentially, generating one token after another. An emerging line of work explored parallel decoding by identifying and simultaneously generating semantically…

Large language models (LLMs) rely on key-value (KV) caches for efficient autoregressive decoding; however, cache size grows linearly with context length and model depth, becoming a major bottleneck in long-context inference. Prior KV cache…

End-to-end image/video codecs are getting competitive compared to traditional compression techniques that have been developed through decades of manual engineering efforts. These trainable codecs have many advantages over traditional…

Deploying large language models (LLMs) on mobile devices increasingly relies on heterogeneous execution, yet no prior study has systematically characterized NPU effectiveness at the operator and pipeline level. We present the first…

Masked autoregressive (MAR) models unify the strengths of masked and autoregressive generation by predicting tokens in a fixed order using bidirectional attention for image generation. While effective, MAR models suffer from significant…

In real-world Tool-Integrated Reasoning (TIR) scenarios, where LLMs interleave reasoning with external tool calls, a major source of inefficiency is that the toolcalls create pauses between LLM requests and cause KV-Cache eviction, forcing…

Key-Value (KV) Caching has become an essential technique for accelerating the inference speed and throughput of generative Large Language Models~(LLMs). However, the memory footprint of the KV cache poses a critical bottleneck in LLM…

Autoregressive (AR) language models generate text one token at a time, even when consecutive tokens are highly predictable given earlier context. We introduce MARS (Mask AutoRegreSsion), a lightweight fine-tuning method that teaches an…

As the field of Large Language Models (LLMs) continues to evolve, the context length in inference is steadily growing. Key-Value Cache (KVCache), the intermediate representations of tokens within LLM inference, has now become the primary…

Transformers evaluated in a single, fixed-depth pass are provably limited in expressive power to the constant-depth circuit class TC0. Running a Transformer autoregressively removes that ceiling -- first in next-token prediction and, more…

LLM decoding is bottlenecked for large batches and long contexts by loading the key-value (KV) cache from high-bandwidth memory, which inflates per-token latency, while the sequential nature of decoding limits parallelism. We analyze the…

When key-value (KV) stores use SSDs for storing a large number of items, oftentimes they also require large in-memory data structures including indices and caches to be traversed to reduce IOs. This paper considers offloading most of such…