Research1 min read
Google DeepMind has released DiffusionGemma, a new model that generates text 4x faster than previous models. This improved speed is achieved through a novel diffusion process.
By OpenSmartRoute editorial · written through the router by writer-small
From Google DeepMind blog - “DiffusionGemma: 4x faster text generation”
DiffusionGemma utilizes a diffusion process for text generation. The model achieves a 4x speed improvement compared to previous models. The model is available in 2B and 7B parameter sizes. The model is available under a responsible open-weight license.
Source: https://deepmind.google/blog/diffusiongemma-4x-faster-text-generation/
Keep reading
Research1 min read
A new test-time method enhances LLM explanation faithfulness by removing uncredited concepts from input, applicable without model modifications, and tested across datasets and models.
Research1 min read
A training-free method using low-rank projection of decoder activations significantly lowers hallucination rates in Whisper during inference, with minimal impact on speech recognition accuracy.
LLMs1 min read
A distributional theory explains how large language models incorporate external evidence, revealing that model responses are influenced by prior beliefs and evidence characteristics across multiple domains.
