google Google Unveils LaMDA 2 to Improve Conversational AI Accuracy Alphabet CEO Sundar Pichai introduces LaMDA 2 at the Google I/O conference, aiming to make computers more accessible through advanced natural language processing. The updated model undergoes extensive internal testing to significantly reduce inaccurate and offensive responses.
openai OpenAI Unveils DALL-E 2 With Major Upgrades in AI Image Generation OpenAI introduces DALL-E 2, an advanced AI system that creates highly detailed, original images from text prompts. The upgraded model offers improved accuracy, diverse artistic styles, and built-in safety measures.
nvidia Nvidia Expands Voice AI Toolkit Amid Booming Enterprise Demand Nvidia releases Riva 2.0 to help enterprises build speech recognition and voice generation applications as the voice AI market rapidly expands.
bert BERT Model Transforms Natural Language Processing With Bidirectional Training Researchers introduce BERT, a bidirectional language model that achieves state-of-the-art results across eleven NLP tasks by pre-training on unlabeled text.
mixtral 8x7b French Startup Mistral AI Launches Powerful Open Source Mixtral 8x7B Model Mistral AI introduces Mixtral 8x7B, an advanced open source language model that brings high precision and versatility to natural language processing. The model stands out for its collaborative design and robust performance across various AI applications.
meta Meta Releases Llama 2, a Powerful Open-Source AI Language Model Meta unveils Llama 2, a family of open foundation and fine-tuned chat models that rival closed-source alternatives. The release provides researchers and developers with unprecedented access to high-performing large language models.
meta ai Meta Releases LLaMA to Democratize Large Language Model Research Meta introduces LLaMA, a foundational large language model available in sizes ranging from 7 billion to 65 billion parameters. The release aims to make advanced AI research more accessible to researchers who lack massive computing infrastructure.
llama New LLaMA AI Models Rival Giants Using Only Public Data Meta researchers introduce LLaMA, a group of open foundation language models ranging from 7 billion to 65 billion parameters. These highly efficient models match or outperform massive proprietary systems while training exclusively on publicly available datasets.
machine translation Meta's NLLB Project Breaks 200-Language Barrier in Machine Translation A new AI model called No Language Left Behind translates over 200 languages to help close the gap between high and low-resource languages. The system uses a sparsely gated mixture of experts architecture and undergoes rigorous safety evaluations to ensure high-quality, ethical results.
meta Meta's NLLB-200 Model Delivers Massive Leap in AI Translation Quality Meta releases NLLB-200, an open-source AI translation model that boosts translation accuracy by 44 percent across 200 languages. The breakthrough aims to connect people worldwide by overcoming language barriers for low-resource dialects.
large language models Researchers Introduce Massive Benchmark to Evaluate Language Model Capabilities A sweeping new study introduces the Beyond the Imitation Game benchmark to rigorously test the reasoning and knowledge of large language models. The project involves over 200 authors collaborating to push past traditional text generation metrics.
ai bias USC Researchers Discover Heavy Bias in Common Sense AI Databases A USC Information Sciences Institute team finds that 3.4% to 38.6% of facts in popular AI databases contain human biases, threatening the fairness of algorithms used by virtual assistants and chatbots.
meta Meta Releases Powerful Open-Source Language Model to Rival GPT-3 Meta challenges OpenAI by releasing a massive 175-billion-parameter language model for free. This open-source approach directly competes with OpenAI's proprietary GPT-3 system.
palm Google Unveils 540-Billion Parameter PaLM to Push AI Reasoning Boundaries Google introduces PaLM, a massive 540-billion parameter language model trained on thousands of specialized chips. The model achieves breakthrough performance on reasoning tasks and even surpasses average human scores on key benchmarks.
meta Meta Unveils OPT, an Open Source Alternative to GPT-3 Meta releases the OPT family of large language models to give researchers broader access to GPT-3-level technology. The release includes fully trained weights for models ranging from 125 million to 175 billion parameters.