How to generate text: using different decoding methods for language generation with Transformers

Generating Human-level Text with Contrastive Search in Transformers 🤗

Guiding Text Generation with Constrained Beam Search in 🤗 Transformers

Faster Text Generation with Self-Speculative Decoding

Universal Assisted Generation: Faster Decoding with Any Assistant Model

How to train a new language model from scratch using Transformers and Tokenizers

Train and Fine-Tune Sentence Transformers Models

Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs

The paper demonstrates that Large Language Models, despite their non‑linear components, exhibit a fundamental linearity property: when inputs from two distinct text streams are linearly combined, the model outputs a superposition of the individual next‑token distributions. This "Superposition Linearity Hypothesis" appears to be an intrinsic feature of the Transformer architecture, tends to weaken during pretraining, but can be largely restored with lightweight fine‑tuning. The authors also present a guided decoding method that separates the superposed outputs, allowing two coherent continuations to be generated from a single forward pass.

Fine-Tune Whisper For Multilingual ASR with 🤗 Transformers

On the quantitative analysis of decoder-based generative models

Transformer-based Encoder-Decoder Models

Training a language model with 🤗 Transformers using TensorFlow and TPUs

Tokenization in Transformers v5: Simpler, Clearer, and More Modular