ASAP integrates an LLM agent over a pool of HPO tools and adds system-level optimizations (prefix-stable prompts, speculation parallelism, Self-Tuner) to improve end-to-end wall-clock performance on diverse HPO tasks.
hub
Fast Inference from Transformers via Speculative Decoding
34 Pith papers cite this work. Polarity classification is still indexing.
34
Pith papers citing it
abstract
Inference from large autoregressive models like Transformers is slow - decoding K tokens takes K serial runs of the model. In this work we introduce speculative decoding - an algorithm to sample from autoregressive models faster without any changes to the outputs, by computing several tokens in parallel. At the heart of our approach lie the observations that (1) hard language-modeling tasks often include easier subtasks that can be approximated well by more efficient models, and (2) using speculative execution and a novel sampling method, we can make exact decoding from the large models faster, by running them in parallel on the outputs of the approximation models, potentially generating several tokens concurrently, and without changing the distribution. Our method can accelerate existing off-the-shelf models without retraining or architecture changes. We demonstrate it on T5-XXL and show a 2X-3X acceleration compared to the standard T5X implementation, with identical outputs.
hub tools
citation-role summary
background 4
citation-polarity summary
roles
background 4polarities
background 4representative citing papers
Frontier is a new discrete-event simulator for disaggregated LLM serving that incorporates co-location, PDD, AFD, and optimizations, achieving under 4% throughput error and large reductions in latency prediction error versus prior simulators.
SlimSpec replaces the standard LM-head in draft models with a low-rank version to deliver 4-5x faster speculative decoding while preserving full vocabulary and competitive acceptance rates.
A training-free speculative decoding method for block-based autoregressive video diffusion uses a quality router on worst-frame ImageReward scores to accept drafter proposals, achieving up to 2.09x speedup at 95.7% quality retention.
Visual token pruning in MLLMs fails on complex reasoning due to Relevant Visual Information Shift during decoding, but the DSTP framework fixes it training-free across models.
Sequential KV compression via probabilistic language tries and predictive delta coding achieves 3.3-4.3 bits per token entropy, yielding up to 914x better ratios than TurboQuant even with large overhead.
MIST is a new simulator for heterogeneous multi-stage LLM inference that combines hardware traces with analytical models to explore configuration trade-offs in hybrid CPU-accelerator systems.
Medusa augments LLMs with multiple decoding heads and tree-based attention to predict and verify several tokens in parallel, yielding 2.2-3.6x inference speedup via two fine-tuning regimes.
Analytical bounds demonstrate edge-cloud DSD improves latency only in low-RTT regimes and primarily benefits multi-tenant throughput via draft offloading under client overlap.
HERALD enables near-lossless accuracy at 5-10% KV budget for block dLLMs by amortizing top-k selection across denoising steps and overlapping CPU-GPU retrieval, yielding up to 2.47x higher throughput than GPU-only inference.
STG generates deterministic testbenches 720x faster than iterative LLM flows with higher coverage and fewer false passes, while serving as an 11x faster data curation engine with 127x less energy.
RKSC delivers 3.008x mean speedup over baseline and 1.66x over vLLM prefix caching for multi-branch LLM reasoning via similarity-based KV sharing and confidence-gated early exit, with 0.37% error rate.
Sparrow uses a dynamic sparsity schedule keyed to the lower tail of sparse-to-dense actor-policy mismatch to enable stable and faster rollouts in long-context RL for LLMs.
SeProD is a plug-and-play self-prophetic decoding framework that combines pre- and post-training LVLM capabilities via probability-based sampling to improve coherent visual search and multi-step reasoning.
Orthrus unifies autoregressive LLMs and diffusion models via shared KV cache and consensus to enable up to 7.8x parallel token generation speedup with O(1) memory overhead and lossless results.
Ultra-compact 8-30M parameter models start contextually grounded responses on-device while cloud models seamlessly continue them, enabling responsive AI on power-constrained hardware.
Fused compressed-domain int4 attention on Apple Silicon delivers 48x speedup and 3.2x KV cache compression for 128K-context 70B models while matching FP16 token predictions.
DualDiffusion combines a lightweight drafter using approximations with a full verifier to reduce generation steps in masked diffusion models while keeping accuracy on MMLU and GSM8K.
PAS automates activation steering for LLMs using labeled data to improve behavior control on tasks like bias and alignment, with gains over ICL and SFT but limited effect on intelligence tasks.
Uptraining multi-head transformer checkpoints to grouped-query attention models achieves near multi-head quality at multi-query inference speeds using 5% additional compute.
The paper proposes a counterfactual definition of AI-native games, screens 53 examples, introduces a G/N taxonomy, and outlines a research roadmap for the field.
Push puppet networks learn a hierarchical penalty function during training to enable structured pruning of LLMs to arbitrary target sizes with reported speedups over SparseGPT and Wanda at high pruning ratios.
Cassandra is a self-speculative decoding system that builds a draft model via fine-grained data selection and optimized pruning/mantissa truncation, achieving up to 2.41x speedup over BF16 and 1.81x more tokens than Eagle-3 on Llama 3 8B without training.
A ReRAM-on-logic stacked chip delivers 14.08-135.69 tokens/s LLM inference with block-clustered compression and adaptive parallel speculative decoding, yielding 4.46-7.17x speedup over standard methods.
citing papers explorer
-
ASAP: Agent-System Co-Design for Wall-Clock-Centered Auto HPO Research for ML Experiments
ASAP integrates an LLM agent over a pool of HPO tools and adds system-level optimizations (prefix-stable prompts, speculation parallelism, Self-Tuner) to improve end-to-end wall-clock performance on diverse HPO tasks.
-
Frontier: Towards Comprehensive and Accurate LLM Inference Simulation
Frontier is a new discrete-event simulator for disaggregated LLM serving that incorporates co-location, PDD, AFD, and optimizations, achieving under 4% throughput error and large reductions in latency prediction error versus prior simulators.
-
SlimSpec: Low-Rank Draft LM-Head for Accelerated Speculative Decoding
SlimSpec replaces the standard LM-head in draft models with a low-rank version to deliver 4-5x faster speculative decoding while preserving full vocabulary and competitive acceptance rates.
-
Speculative Decoding for Autoregressive Video Generation
A training-free speculative decoding method for block-based autoregressive video diffusion uses a quality router on worst-frame ImageReward scores to accept drafter proposals, achieving up to 2.09x speedup at 95.7% quality retention.
-
Why and When Visual Token Pruning Fails? A Study on Relevant Visual Information Shift in MLLMs Decoding
Visual token pruning in MLLMs fails on complex reasoning due to Relevant Visual Information Shift during decoding, but the DSTP framework fixes it training-free across models.
-
Sequential KV Cache Compression via Probabilistic Language Tries: Beyond the Per-Vector Shannon Limit
Sequential KV compression via probabilistic language tries and predictive delta coding achieves 3.3-4.3 bits per token entropy, yielding up to 914x better ratios than TurboQuant even with large overhead.
-
MIST: A Co-Design Framework for Heterogeneous, Multi-Stage LLM Inference
MIST is a new simulator for heterogeneous multi-stage LLM inference that combines hardware traces with analytical models to explore configuration trade-offs in hybrid CPU-accelerator systems.
-
Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
Medusa augments LLMs with multiple decoding heads and tree-based attention to predict and verify several tokens in parallel, yielding 2.2-3.6x inference speedup via two fine-tuning regimes.
-
Speculation at a Distance: Where Edge-Cloud Speculative Decoding Actually Pays Off
Analytical bounds demonstrate edge-cloud DSD improves latency only in low-RTT regimes and primarily benefits multi-tenant throughput via draft offloading under client overlap.
-
HERALD: High-Throughput Block Diffusion LLM Serving via CPU-GPU Cooperative KV Cache Retrieval
HERALD enables near-lossless accuracy at 5-10% KV budget for block dLLMs by amortizing top-k selection across denoising steps and overlapping CPU-GPU retrieval, yielding up to 2.47x higher throughput than GPU-only inference.
-
Structured Testbench Generation for LLM-Driven HDL Design and Verification-Oriented Data Curation
STG generates deterministic testbenches 720x faster than iterative LLM flows with higher coverage and fewer false passes, while serving as an 11x faster data curation engine with 127x less energy.
-
RKSC: Reasoning-Aware KV Cache Sharing and Confident Early Exit for Multi-Step LLM Inference
RKSC delivers 3.008x mean speedup over baseline and 1.66x over vLLM prefix caching for multi-branch LLM reasoning via similarity-based KV sharing and confidence-gated early exit, with 0.37% error rate.
-
Sparrow: Sparse Rollout for Stable and Efficient Long-context RL of Large Language Models
Sparrow uses a dynamic sparsity schedule keyed to the lower tail of sparse-to-dense actor-policy mismatch to enable stable and faster rollouts in long-context RL for LLMs.
-
Self-Prophetic Decoding to Unlock Visual Search in LVLMs
SeProD is a plug-and-play self-prophetic decoding framework that combines pre- and post-training LVLM capabilities via probability-based sampling to improve coherent visual search and multi-step reasoning.
-
Orthrus: Memory-Efficient Parallel Token Generation via Dual-View Diffusion
Orthrus unifies autoregressive LLMs and diffusion models via shared KV cache and consensus to enable up to 7.8x parallel token generation speedup with O(1) memory overhead and lossless results.
-
Micro Language Models Enable Instant Responses
Ultra-compact 8-30M parameter models start contextually grounded responses on-device while cloud models seamlessly continue them, enabling responsive AI on power-constrained hardware.
-
Open-TQ-Metal: Fused Compressed-Domain Attention for Long-Context LLM Inference on Apple Silicon
Fused compressed-domain int4 attention on Apple Silicon delivers 48x speedup and 3.2x KV cache compression for 128K-context 70B models while matching FP16 token predictions.
-
DualDiffusion: A Speculative Decoding Strategy for Masked Diffusion Models
DualDiffusion combines a lightweight drafter using approximations with a full verifier to reduce generation steps in masked diffusion models while keeping accuracy on MMLU and GSM8K.
-
Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models
PAS automates activation steering for LLMs using labeled data to improve behavior control on tasks like bias and alignment, with gains over ICL and SFT but limited effect on intelligence tasks.
-
GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints
Uptraining multi-head transformer checkpoints to grouped-query attention models achieves near multi-head quality at multi-query inference speeds using 5% additional compute.
-
AI Native Games: A Survey and Roadmap
The paper proposes a counterfactual definition of AI-native games, screens 53 examples, introduces a G/N taxonomy, and outlines a research roadmap for the field.
-
Push Puppet Networks: Structured Bayesian Pruning Algorithm for Language Model Compression
Push puppet networks learn a hierarchical penalty function during training to enable structured pruning of LLMs to arbitrary target sizes with reported speedups over SparseGPT and Wanda at high pruning ratios.
-
Cassandra: Enabling Reasoning LLMs at Edge via Self-Speculative Decoding
Cassandra is a self-speculative decoding system that builds a draft model via fine-grained data selection and optimized pruning/mantissa truncation, achieving up to 2.41x speedup over BF16 and 1.81x more tokens than Eagle-3 on Llama 3 8B without training.
-
31.1 A 14.08-to-135.69Token/s ReRAM-on-Logic Stacked Outlier-Free Large-Language-Model Accelerator with Block-Clustered Weight-Compression and Adaptive Parallel-Speculative-Decoding
A ReRAM-on-logic stacked chip delivers 14.08-135.69 tokens/s LLM inference with block-clustered compression and adaptive parallel speculative decoding, yielding 4.46-7.17x speedup over standard methods.
-
Complexity Horizons of Compressed Models in Analog Circuit Analysis
Prerequisite graphs map compressed LLM performance boundaries in analog circuit analysis to allow selecting the smallest viable model for a given task complexity.
-
Understanding and Improving Communication Performance in Multi-node LLM Inference
Performance analysis of multi-node LLM inference identifies all-reduce bottlenecks and introduces NVRAR hierarchical all-reduce achieving 1.9-3.6x lower latency than NCCL and up to 1.72x end-to-end batch latency reduction for Llama 3.1 405B in decode-heavy tensor-parallel workloads.
-
Speculative Decoding Across Languages
N-gram draft models give larger and more consistent speed-ups for multilingual speculative decoding than fine-tuned neural drafts, despite lower acceptance rates, across translation and story generation.
-
EdgeFM: Efficient Edge Inference for Vision-Language Models
EdgeFM is an agent-driven VLM inference framework achieving up to 1.49x speedup over TensorRT-Edge-LLM on NVIDIA Orin and first end-to-end deployment on Horizon Journey platform.
-
Token-Operations-Oriented Inference Optimization Techniques for Large Models
The paper introduces a four-layer technical architecture for token-operations-oriented inference optimization in large models and reviews key technologies and industry status at each layer.
-
Mellum2 Technical Report
Mellum 2 is a 12B MoE model with 2.5B active parameters, trained on 10.6T tokens with MoE, GQA, SWA, and MTP, then post-trained into Instruct and Thinking variants, claimed competitive with 4B-14B models at 2.5B compute.
-
LLMOrbit: A Circular Taxonomy of Large Language Models -From Scaling Walls to Agentic AI Systems
A survey taxonomy of LLMs identifies three scaling crises and six efficiency paradigms while tracing the shift from generation to tool-using agents.
-
The Hitchhiker's Guide to Agentic AI: From Foundations to Systems
A comprehensive reference book organizing existing techniques for agentic AI systems across LLM substrate, reasoning, agent design patterns, inter-agent coordination, and production deployment.
- SOLARIS: Speculative Offloading of Latent-bAsed Representation for Inference Scaling
- Drift-AR: Single-Step Visual Autoregressive Generation via Anti-Symmetric Drifting