Benchmarking Text Generation Inference

Introducing multi-backends (TRT-LLM, vLLM) support for Text Generation Inference

Faster Text Generation with Self-Speculative Decoding

Lost-in-the-Middle in Long-Text Generation: Synthetic Dataset, Evaluation Framework, and Mitigation

arXiv:2503.06868v2 Announce Type: replace-cross Abstract: Existing long-text generation methods produce lengthy outputs from short inputs, leaving long-input-to-long-output generation underexplored....

Assisted Generation: a new direction toward low-latency text generation

The Effect of Text Chunk Size on Retrieval-Augmented Generation Performance

arXiv:2607. 24767v1 Announce Type: cross Abstract: Retrieval-Augmented Generation (RAG) systems have emerged as a powerful process for allowing large language models (LLMs) to retrieve relevant information to use as source material during text generation.

FLEET: From Logits Entropy to Enhanced Trajectories in Text Generation

FLEET is a new method for text generation that adds a memory mechanism to large language models. It represents each generation as a sparse trajectory of high‑entropy states and uses these trajectories to compute per‑token utility scores that adjust the logits. Benchmarks show that FLEET matches the accuracy of repeated sampling while being three times faster and improving accuracy on complex coding tasks, all with minimal changes to existing pipelines.