GoldFinch: High Performance RWKV/Transformer Hybrid with Linear Pre-Fill and Extreme KV-Cache Compression

• URL: http://arxiv.org/abs/2407.12077v1
• Date: Tue, 16 Jul 2024 18:00:00 GMT
• Title: GoldFinch: High Performance RWKV/Transformer Hybrid with Linear Pre-Fill and Extreme KV-Cache Compression
• Abstract summary: GoldFinch is a hybrid Linear Attention/Transformer sequence model. It stacks our new GOLD transformer on top of an enhanced version of the Finch architecture. We train up to 1.5B parameter class models of the Finch, Llama, and GoldFinch architectures.
• Score: 0.5534140394498714
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We introduce GoldFinch, a hybrid Linear Attention/Transformer sequence model that uses a new technique to efficiently generate a highly compressed and reusable KV-Cache in linear time and space with respect to sequence length. GoldFinch stacks our new GOLD transformer on top of an enhanced version of the Finch (RWKV-6) architecture. We train up to 1.5B parameter class models of the Finch, Llama, and GoldFinch architectures, and find dramatically improved modeling performance relative to both Finch and Llama. Our cache size savings increase linearly with model layer count, ranging from 756-2550 times smaller than the traditional transformer cache for common sizes, enabling inference of extremely large context lengths even on limited hardware. Although autoregressive generation has O(n) time complexity per token because of attention, pre-fill computation of the entire initial cache state for a submitted context costs only O(1) time per token due to the use of a recurrent neural network (RNN) to generate this cache. We release our trained weights and training code under the Apache 2.0 license for community use.

Related papers

• Hidden Decoding at Scale: Latent Computation Scaling for Large Language Models [75.80275843320511]
  We study whether an existing backbone can keep improving by allocating more computation to each token while leaving the Transformer backbone fixed.<n>We propose Hidden Decoding, a sequence-length scaling method applied during continued pretraining.
  arXiv  Detail & Related papers  (2026-07-09T07:37:59Z)
• End-to-End Context Compression at Scale [81.70601323130997]
  Long-context language model inference is bottlenecked by memory, as the KV cache grows with context length.<n>Recent techniques to compress the KV cache fall short: they either degrade model quality substantially or require considerable time and compute to compress a single long prompt.<n>Existing approaches are not competitive with KV cache compression on the accuracy-efficiency frontier.
  arXiv  Detail & Related papers  (2026-06-08T15:43:16Z)
• Key-Value Means: Transformers with Expandable Block-Recurrent Compressed Memory [1.3714815350300447]
  Key-Value Means (KVM) is a block-recurrence for attention that can accommodate either fixed-size or growing state.<n>We show it performs competitively on long-context tests with only subquadratic prefill time and sublinear state growth.
  arXiv  Detail & Related papers  (2026-05-11T02:04:14Z)
• Long-Context Aware Upcycling: A New Frontier for Hybrid LLM Scaling [25.551309705184234]
  HyLo extends usable context length by up to $32times$ through efficient post-training.<n>HyLo delivers consistently strong short- and long-context performance.<n>At similar scale, HyLo-Qwen-1.7B trained on only 10B tokens significantly outperforms JetNemotron on GSM8K, Lm-Harness common sense reasoning and RULER-64K.
  arXiv  Detail & Related papers  (2026-04-27T17:23:37Z)
• Flow caching for autoregressive video generation [72.10021661412364]
  We present FlowCache, the first caching framework specifically designed for autoregressive video generation.<n>Our method achieves remarkable speedups of 2.38 times on MAGI-1 and 6.7 times on SkyReels-V2, with negligible quality degradation.
  arXiv  Detail & Related papers  (2026-02-11T13:11:04Z)
• Hybrid Linear Attention Done Right: Efficient Distillation and Effective Architectures for Extremely Long Contexts [27.8245634187787]
  We present HALO, a pipeline for distilling Transformer models into RNN-attention hybrid models.<n>We then present HypeNet, a hybrid architecture with superior length generalization enabled by a novel position encoding scheme.<n>The conversion requires just 2.3B tokens, less than 0.01% of their pre-training data.
  arXiv  Detail & Related papers  (2026-01-29T18:59:53Z)
• From Prefix Cache to Fusion RAG Cache: Accelerating LLM Inference in Retrieval-Augmented Generation [11.929816243824561]
  Retrieval-Augmented Generation enhances Large Language Models by integrating external knowledge.<n>Existing solutions aim to reuse the preprocessed KV cache of each retrieved chunk to accelerate RAG.<n>We propose FusionRAG, a novel inference framework that optimize both the preprocessing and reprocessing stages of RAG.
  arXiv  Detail & Related papers  (2026-01-19T09:59:39Z)
• PackCache: A Training-Free Acceleration Method for Unified Autoregressive Video Generation via Compact KV-Cache [61.57938553036056]
  We introduce PackCache, a training-free KV-cache management method that compacts the KV cache through three coordinated mechanisms.<n>In terms of efficiency, PackCache accelerates end-to-end generation by 1.7-2.2x on 48-frame long sequences.
  arXiv  Detail & Related papers  (2026-01-07T19:51:06Z)
• Artificial Hippocampus Networks for Efficient Long-Context Modeling [17.23148291364832]
  Long-sequence modeling faces a trade-off between the efficiency of compressive fixed-size memory in RNN-like models and the fidelity of growing memory in attention-based Transformers.<n>Inspired by the Multi-Store Model in cognitive science, we introduce a memory framework of artificial neural networks.<n>Experiments on long-context benchmarks LV-Eval and InfiniteBench demonstrate that AHN-augmented models consistently outperform sliding window baselines.
  arXiv  Detail & Related papers  (2025-10-08T17:59:55Z)
• LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models [52.56008278458534]
  LaCache is a training-free method for efficient and accurate generative inference of Large Language Models.<n>LaCache enables LLMs to address both of the critical challenges in long-range modeling: robust long-range capabilities and continuous generation without running out-of-memory.
  arXiv  Detail & Related papers  (2025-07-14T19:09:57Z)
• FastCache: Fast Caching for Diffusion Transformer Through Learnable Linear Approximation [46.57781555466333]
  Diffusion Transformers (DiT) are powerful generative models but remain computationally intensive due to their iterative structure and deep transformer stacks.<n>FastCache is a hidden-state-level caching and compression framework that accelerates DiT inference.<n> Empirical evaluations across multiple DiT variants demonstrate substantial reductions in latency and memory usage.
  arXiv  Detail & Related papers  (2025-05-26T05:58:49Z)
• QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation [84.91431271257437]
  Diffusion Transformers (DiTs) have emerged as a dominant architecture in video generation. DiTs come with significant drawbacks, including increased computational and memory costs. We propose QuantCache, a novel training-free inference acceleration framework.
  arXiv  Detail & Related papers  (2025-03-09T10:31:51Z)
• LinFusion: 1 GPU, 1 Minute, 16K Image [71.44735417472043]
  We introduce a low-rank approximation of a wide spectrum of popular linear token mixers. We find that the distilled model, termed LinFusion, achieves performance on par with or superior to the original SD. Experiments on SD-v1.5, SD-v2.1, and SD-XL demonstrate that LinFusion enables satisfactory and efficient zero-shot cross-resolution generation.
  arXiv  Detail & Related papers  (2024-09-03T17:54:39Z)
• ThinK: Thinner Key Cache by Query-Driven Pruning [63.13363917871414]
  Large Language Models (LLMs) have revolutionized the field of natural language processing, achieving unprecedented performance across a variety of applications. This paper focuses on the long-context scenario, addressing the inefficiencies in KV cache memory consumption during inference. We propose ThinK, a novel query-dependent KV cache pruning method designed to minimize attention weight loss while selectively pruning the least significant channels.
  arXiv  Detail & Related papers  (2024-07-30T17:59:08Z)
• Parallelizing Linear Transformers with the Delta Rule over Sequence Length [49.88826673324244]
  This work describes a hardware-efficient algorithm for training linear transformers with the delta rule. We train a 1.3B model for 100B tokens and find that it outperforms recent linear-time baselines.
  arXiv  Detail & Related papers  (2024-06-10T17:24:42Z)
• Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching [56.286064975443026]
  We make an interesting and somehow surprising observation: the computation of a large proportion of layers in the diffusion transformer, through a caching mechanism, can be readily removed even without updating the model parameters. We introduce a novel scheme, named Learningto-Cache (L2C), that learns to conduct caching in a dynamic manner for diffusion transformers. Experimental results show that L2C largely outperforms samplers such as DDIM and DPM-r, alongside prior cache-based methods at the same inference speed.
  arXiv  Detail & Related papers  (2024-06-03T18:49:57Z)
• Gated Linear Attention Transformers with Hardware-Efficient Training [60.670102007737476]
  This work describes a hardware-efficient algorithm for linear attention that trades off memory movement against parallelizability. We then generalize this algorithm to a more expressive variant of linear attention with data-dependent gates. When used as a replacement for the standard attention layer in Transformers, the resulting gated linear attention Transformer is found to perform competitively.
  arXiv  Detail & Related papers  (2023-12-11T18:51:59Z)
• DeepCache: Accelerating Diffusion Models for Free [65.02607075556742]
  DeepCache is a training-free paradigm that accelerates diffusion models from the perspective of model architecture. DeepCache capitalizes on the inherent temporal redundancy observed in the sequential denoising steps of diffusion models. Under the same throughput, DeepCache effectively achieves comparable or even marginally improved results with DDIM or PLMS.
  arXiv  Detail & Related papers  (2023-12-01T17:01:06Z)
• Kronecker-Factored Approximate Curvature for Modern Neural Network Architectures [85.76673783330334]
  Two different settings of linear weight-sharing layers motivate two flavours of Kronecker-Factored Approximate Curvature (K-FAC) We show they are exact for deep linear networks with weight-sharing in their respective setting. We observe little difference between these two K-FAC variations when using them to train both a graph neural network and a vision transformer.
  arXiv  Detail & Related papers  (2023-11-01T16:37:00Z)
• TransNormerLLM: A Faster and Better Large Language Model with Improved TransNormer [34.790081960470964]
  We present TransNormerLLM, the first linear attention-based Large Language Model (LLM) We make advanced modifications that include positional embedding, linear attention acceleration, gating mechanisms, tensor normalization, and inference acceleration and stabilization. We validate our model design through a series of ablations and train models with sizes of 385M, 1B, and 7B on our self-collected corpus.
  arXiv  Detail & Related papers  (2023-07-27T16:45:33Z)
• Fourier Transformer: Fast Long Range Modeling by Removing Sequence Redundancy with FFT Operator [24.690247474891958]
  Fourier Transformer is able to significantly reduce computational costs while retain the ability to inherit from various large pretrained models. Our model achieves state-of-the-art performances among all transformer-based models on the long-range modeling benchmark LRA. For generative seq-to-seq tasks including CNN/DailyMail and ELI5, by inheriting the BART weights our model outperforms the standard BART.
  arXiv  Detail & Related papers  (2023-05-24T12:33:06Z)
• Learned Queries for Efficient Local Attention [11.123272845092611]
  Self-attention mechanism in vision transformers suffers from high latency and inefficient memory utilization. We propose a new shift-invariant local attention layer, called query and attend (QnA), that aggregates the input locally in an overlapping manner. We show improvements in speed and memory complexity while achieving comparable accuracy with state-of-the-art models.
  arXiv  Detail & Related papers  (2021-12-21T18:52:33Z)
• FNet: Mixing Tokens with Fourier Transforms [0.578717214982749]
  We show that Transformer encoder architectures can be massively sped up with limited accuracy costs. We replace the self-attention sublayers with simple linear transformations that "mix" input tokens. The resulting model, which we name FNet, scales very efficiently to long inputs.
  arXiv  Detail & Related papers  (2021-05-09T03:32:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.