Claude 3 Opus, the powerhouse AI model from Anthropic, is now available on Amazon Bedrock.

Opus can tackle highly complex tasks with remarkable fluency, and has the ability to navigate open-ended scenarios with human-like understanding. Bedrock is the easiest and most secure…

Today, we're announcing Claude 3, our next generation of AI models.

The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.

Amazon has publicly released RefChecker, a combination tool and dataset that detects hallucinations in #LLMs . To characterize factual claims, RefChecker uses knowledge triplets rather than natural language, enabling finer-grained judgments. #GenerativeAI amazon.science/blog/new-tool-…

The first article of a series at the Netflix technology blog: 'Rebuilding Netflix Video Processing Pipeline with Microservices' netflixtechblog.com/rebuilding-net…

The History of #AI
by Parmida Beigi

#BigData #ArtificialIntelligence #ML #MI #DataScience

cc: Pascal Bornet @yvesmulkers Kohei Kurihara - Privacy for all together 🌍

❓Will AI be the future? Unmask the AI with Parmida Beigi, a true Artificial Intelligence and Machine Learning visionary, researcher behind #amazon Alexa, and a speaker at the upcoming #TEDxSFU Conference!

Join us in welcoming Parmida Beigi to the TEDxSFU stage! Parmida Beigi

The emergence of open source LLMs led to the potential generation of toxic outputs. Amazon Bedrock, the latest step in the company’s ongoing effort to democratize ML, uses Titan FM to help customers detect and remove harmful content in inputs and filter model outputs. #GenAI

First fine-tuned model outperforming GPT-4 💥

I've been eagerly awaiting what comes next🔻after 15B WizardCoder: surpassing Claude-Plus and Bard on HumanEval.

And 𝗷𝘂𝘀𝘁 𝘁𝗼𝗱𝗮𝘆, 34B model is released, 𝘀𝘂𝗿𝗽𝗮𝘀𝘀𝗶𝗻𝗴 𝗚𝗣𝗧-𝟰 (March model)!

huggingface.co/WizardLM/Wizar…

Lighter and faster LLMs with 🤗 Transformers 🤝 AutoGPTQ!

1. Low GPU memory usage
2. Fast inference speed

Results of GPTQ quantization. 🧮

Try it out ⬇️

🔗 Blog post: huggingface.co/blog/gptq-inte…
📒 Google colab: colab.research.google.com/drive/1_TIrmuK…

Six techniques for data privacy include encrypted data analysis, obscuring individual data with noise, controlled access to encrypted text, sharing analysis insights, concealing information through proofs, and collective analysis without full data exposure.

Microblog Antonio Grasso

Streamline your LLM training with Amazon SageMaker! 👷‍♂️Thrilled to share a new blog post on how to leverage Amazon Web Services SageMaker to efficiently train LLMs (Falcon 40B) using QLoRA and Hugging Face Transformers. 🧑🏻‍💻

👉 philschmid.de/sagemaker-falc…

philschmid.de/sagemaker-falc…

'Hallucinations' is not the right term when large language models like ChatGPT output fiction as fact; we should use 'confabulations,' a neuropsychiatric disorder wherein a patient generates false memories without the intention of deceit. ~ G. Hinton at ACL today. #ai #ChatGPT

A group of four NASA volunteers has embarked on a 378-day mission in which they will be locked in a ground-based simulation of the planet Mars. The mission, which began on Sunday, is the first of three year-long Mars surface simulations, according to NASA. During the mission,…

Congrats to the 8 university teams who have advanced to the #AlexaPrize TaskBot Challenge 2 semifinals. Learn more about this phase of the challenge, and how the teams will compete for the $500k first-place prize. #ConvAI

LIMA: Meta’s new LLM, based on the 65B LLAMA

Interestingly, it has been trained using a mere 1,000 samples, yet it shows comparable results to the SOTA LLMs.

Language models are not exactly 'few-shot' learners, don’t you agree?

🔖 arxiv.org/pdf/2305.11206… #NLProc @MetaAI

Should LLM-base models be open or closed? Meta advocates for openness (eg. LLaMA), while OpenAI & Google lean towards proprietary.

They say openness is risky (or are they safeguarding their own interests?)
Closeness though poses far greater dangers. #ai

nytimes.com/2023/05/18/tec…

Geoffrey Hinton, an AI pioneer, quit his job at Google, where he has worked for more than decade, so he can freely speak out about the risks posed by AI.

“It is hard to see how you can prevent the bad actors from using it for bad things,” he said.
nyti.ms/3Ht2Adg