#model-efficiency

[ follow ]
#deep-learning #nanochat #small-language-models #open-source-ai #generative-ai #text-to-video
fromFortune
1 week ago

Want to build your own chatbot for $100? A glimpse into AI's small, cheap, DIY future | Fortune

Andrej Karpathy, a former OpenAI researcher and Tesla's former director of AI, calls his latest project the "best ChatGPT $100 can buy." Called "nanochat," the open-source project, released yesterday for his AI education startup EurekaAI, shows how anyone with a single GPU server and about $100 can build their own mini-ChatGPT that can answer simple questions and write stories and poems.
Tech industry
Environment
fromFuturism
4 weeks ago

Researchers Just Found Something Extremely Alarming About AI's Power Usage

Text-to-video generative AI consumes energy nonlinearly: doubling video length roughly quadruples energy use, greatly increasing hardware and environmental costs.
fromAbove the Law
1 month ago

It's A Small (Language Model) World After All - Above the Law

While OpenAI and Anthropic continue begging for more and more investor cash in the face of consistently lackluster earnings, some vendors delivering advanced AI to the legal industry dropped hints about growing interest in small models. It's not that large language models don't work - though they often don't - but they're overbloated science experiments that, as Goldman Sachs observed, require exponentially increased resources to achieve tiny linear gains.
Artificial intelligence
Artificial intelligence
fromHackernoon
1 year ago

Igniting Generative Power: Multi-Token LLMs for Advanced Text Summarization | HackerNoon

Comprehensive evaluation reveals that the 7B parameter models significantly improve summarization tasks when trained on vast amounts of natural language data.
Scala
fromHackernoon
55 years ago

How an 8B Open Model Sets New Standards for Safe and Efficient Vision-Language AI | HackerNoon

Idefics2 emerges as a state-of-the-art vision-language model, showcasing efficiency and performance improvements through systematic experimentation.
Data science
fromHackernoon
6 months ago

LightCap's Success on Nocaps: Limitations and Opportunities for Growth | HackerNoon

The proposed model framework shows efficient performance but has limitations regarding computational cost and training data.
[ Load more ]