- Published on July 23, 2024
- In Global Tech
Meta’s Llama 3.1 is Starting to Look a Lot Like OpenAI’s GPT-4o
Rumours suggest that Meta has already begun training Llama 4, which is expected to be multimodal with audio features and integrated into the Meta Ray-Ban glasses.
- By Siddharth Jindal
Mark Zuckerberg seems to be on the right track as Meta prepares to unveil the next iteration of Llama 3.1, expected to be released today. The new version will come in three sizes: 8B, 70B, and 405B, with a context length of 128K.
However, even before Meta could officially release the model, its benchmark card was leaked, and it is doing rounds on social media.
https://twitter.com/maximelabonne/status/1815677783311339832
According to the leaked information, Llama 3.1 has been trained on over 15 trillion tokens sourced from publicly available datasets. Their fine-tuning data comprises publicly available instruction-tuning datasets, along with an additional 15 million synthetic samples.
The models are explicitly advertised as multilingual, offering support for French
📣 Want to advertise in AIM? Book here
Siddharth Jindal
Siddharth is a media graduate who loves to explore tech through journalism and putting forward ideas worth pondering about in the era of artificial intelligence.
Related Posts
Fighting Deepfakes May Not Be a Technology Problem
Defenders must be active at all times, while attackers need only one opportunity.
India’s Data Centre Expansion Is Decentralising
Without compute buildup beyond metros, the next wave of digital adoption will be constrained
How Mumbai Keeps Winning India’s Data-Centre Race
Land prices are among the highest in the country, but total build economics remain competitive by global standards.
From Shortages to Scale, io.net’s Approach to Rewriting AI Compute Access
A decentralised GPU marketplace may scale AI compute faster than traditional clouds, as GPU demand towers over supply
Why Deloitte Built a Tax AI That Knows When to Say ‘I Don’t Know’
The company has launched an agentic AI platform for tax research that’s targeting something radical in a conservative profession.
Can India’s AI Copyright Plan Survive Legal and Technical Scrutiny?
India’s ambitious proposal for a single mandatory AI training licence faces feasibility, legal and innovation concerns.
2026 Could be India’s Year in AI, But Only the Resilient Will Survive
“This isn’t a freeze. It’s a filter… By 2026, the real signal will be resilience, not rhetoric.”
Enterprises Won’t Choose Sovereign Models For Patriotism’s Sake
The first set of sovereign models is aimed at Indic languages and national datasets. Their value lies in cultural nuance.
Download the easiest way to
stay informed
stay informed
