- The AI Pulse
- Posts
- 🤖 OpenAI Partners With TIME
🤖 OpenAI Partners With TIME
PLUS: Hugging Face Updates Open LLM Leaderboard, Google’s New Gemma 2
James Roesser
June 28, 2024
Welcome back AI enthusiasts!
In today’s AI Report:
⏰OpenAI Partners With TIME
🏆Hugging Face Updates Open LLM Leaderboard
⚙️Google’s New Gemma 2
🛠5 Trending Tools
💰Venture Capital Updates
💼Who’s Hiring?
Read Time: 3 minutes
🗞RECENT NEWS
OPENAI
⏰OpenAI Partners With TIME
Image Source: Amir Cohen/Vox Screenshot
OpenAI partnered with TIME in a multi-year content deal and strategic partnership to bring TIME’s trusted journalism to OpenAI’s ChatGPT.
Key Details:
OpenAI will access current and historical content from TIME’s extensive archives covering the last 101 years to enhance ChatGPT’s historical training datasets.
When ChatGPT responds to a user’s query with TIME content, it’ll feature a citation and direct link back to the original source.
TIME COO Mark Howard explained: “This partnership with OpenAI advances our mission to expand access to trusted information globally.”
TIME will access OpenAI’s products and services to develop tailored AI offerings for its readers.
OpenAI COO Brad Lightcap expressed: “We’re partnering with TIME to support reputable journalism by providing proper attribution to original sources.”
Why It’s Important:
Eight established U.S. newspapers recently filed a lawsuit against OpenAI and Microsoft, accusing the technology giants of illegally leveraging copyrighted articles to train AI models.
OpenAI should change the company’s name to “PartnerAI.” They seem to be strategically partnering and acquiring publications to avoid legal disputes.
🩺 PULSE CHECK
Should conversational chatbots like ChatGPT have citations to original sources when generating outputs?Vote Below to View Live Results |
HUGGING FACE
🏆Hugging Face Updates Open LLM Leaderboard
Image Source: Canva AI Image Generator
Hugging Face introduced new benchmarks, metrics, and evaluation methods to measure the performance of Large Language Models (LLMs).
Key Details:
The Open LLM Leaderboard hosted over 2 million unique website visits and evaluated around 300,000 community member submissions and discussions.
This popularity, paired with LLM advancements, made it difficult for the current Open LLM Leaderboard to differentiate between AI models effectively.
For example, developers leveraged similar benchmark datasets to train their LLMs, leading to dataset contamination or “overfitting.”
This “overfitting” enabled LLMs to memorize specific problems from benchmarks instead of learning general problem-solving skills to apply concepts and perform well on unseen data.
Hugging Face added new benchmarks to address “overfitting” (e.g., MMLU-Pro, GPQA, MuSR, MATH, IFEval, and BBH) and new metrics for ranking LLMs using normalized scores.
Why It’s Important:
Measuring an LLM’s performance is difficult. LLMs are constantly evolving and approaching human-level performance on most tasks, which leads to benchmark saturation and metric contamination.
Updating the Open LLM Leaderboard creates a universal evaluation method with new benchmarks and metrics to rank LLMs accurately.
🚨Check out the Open LLM Leaderboard V2 here.
AI RESEARCH
⚙️Google’s New Gemma 2
Google unveiled Gemma 2, a series of open-source language models offering researchers and developers best-in-class performance across different hardware.
Gemma 2 was trained on a massive dataset of text data containing trillions of tokens:
The larger 27B parameter model was trained on a whopping 13 trillion tokens.
The smaller 9B parameter model was trained on a still-impressive 8 trillion tokens.
Tokens are the smallest units of data used by an AI model to process and generate text. Similarly, we break down sentences into words or characters. You can think of tokens as syllables.
However, tokens represent many components beyond just alphabetical characters, like punctuation or sentence boundaries. You can experiment with OpenAI’s tokenizer here.
Gemma 2 is designed to require fewer computing resources, which allows it to run on smaller devices and reduce deployment costs. It supports various tools and frameworks (e.g., TensorFlow, PyTorch, JAX, and Keras), giving developers flexibility.
🛠TRENDING TOOLS
💛ApyHub builds, tests, and documents API development.
💬Question Base automates repetitive questions in Slack.
🎨Gradient Generator creates beautiful gradients to match your design.
🕸Sider empowers you to chat, write, read, and translate on any webpage.
🎸Jamahook instantly finds the perfect musical elements for the song you’re working on.
🔮Browse our always Up-To-Date AI Tools Database.
💰VENTURE CAPITAL UPDATES
💼WHO’S HIRING?
Ventas (Chicago, IL): Software Engineering Intern, Summer 2025
Morgan Stanley (New York, NY): 2025 Technology Summer Analyst Program
Zoom (San Francisco, CA): AI/ML Speech Engineer, New Grad
Meta (Menlo Park, CA): Product Security Engineer, Entry-Level
Anyscale (Remote): Consulting Engineer
🤖PROMPT OF THE DAY
CUSTOMER SERVICE
🦾Chatbots for Customers
Research the use of chatbots for customer service in [Industry] for [Business] with [Product/Service]. Explore the benefits, such as improved response times, scalability and cost-effectiveness.
Include specific examples or case studies of businesses that’ve successfully integrated chatbots to enhance customer satisfaction and operational efficiency.
Industry = [Insert Here]
Business = [Insert Here]
Product/Service = [Insert Here]📒FINAL NOTE
If you found this useful, follow us on Twitter or provide honest feedback below. It helps us improve our content.
How was today’s newsletter?
❤️TAIP Review of the Day
“Very informative and up-to-date.”
REFER & EARN
🎉Your Friends Learn, You Earn!
You currently have 0 referrals, only 1 away from receiving ⚙️Ultimate Prompt Engineering Guide.
Refer 5 friends to enter 🎰July’s $200 Gift Card Giveaway.
Copy and paste this link to others: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER
Reply