Elon is not even in the race here.
FAIR and Meta were working on LLM and other techniques before Elon even knew this was a thing.
We all know at Meta that LLMs are very useful, but not the be-all/end-all.
FAIR is working on the next generation of AI architectures beyond
RESEARCH
-
Meta FAIR Advances Beyond LLMs in AI Architecture Race
By
–
-
Training Data Mix and Domain-Specific Evals Drive AI Model Performance
By
–
My intuition is two things:
1. Carefully curated pre-training data & finding the right data mix 2. Domain/ task specific evals for downstream use-cases In the end training data is still king! – finding which combination to go for is where the real moneys at. In addition- Post -
Hopes for Advanced AI Agents Beyond Anthropic and OpenAI
By
–
I hope for better agents than Anthropic; OpenAI has certainly already finished GPT-5 and is certainly already working on o2. In this respect, I expect agents with good reasoning that can take on much better tasks 🙂
-
AGI breakthrough proposed by existing LLM architecture
By
–
AGI: "Does it count if the architecture breakthrough is proposed by en existing LLM?"
-
India AI CyberGuard Hackathon Tackles Digital Threats
By
–
Innovators, it’s time to tackle cyber threats! Join the #IndiaAICyberGuardAIHackathon and help secure Digital India. Apply Now at https://
indiaai.gov.in/article/ai-for
-citizen-safety-join-the-indiaai-cyberguard-ai-hackathon
… Stage 1: Virtual Hackathon – Show off your skills! Stage 2: Top 20 Teams get invited for an In-Person Hackathon -
IndiaAI CyberGuard AI Hackathon Launches Advanced Security Solutions
By
–
IndiaAI in partnership with I4C has launched the IndiaAI CyberGuard AI Hackathon. The Hackathon is a strategic effort to leverage advanced technologies such as Artificial Intelligence (AI) to improve cybersecurity and address the escalating threat of cyber fraud and crimes.
-
New Frameworks with 3.5 Sonnet Cross 50% Leaderboard Barrier
By
–
One day since this thread and we already have new frameworks powered by the new 3.5 Sonnet on the leaderboard. The 50% barrier has been crossed – you love to see it.
-
Contrastive RL Enables Emergent Skills Without Rewards
By
–
A Single Goal is All You Need: Skills and Exploration Emerge from Contrastive RL without Rewards, Demonstrations, or Subgoals https://
bit.ly/4gMQUlI
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
Oasis: Interactive World Model Using Diffusion Transformers
By
–
Oasis
— AK (@_akhaliq) 31 octobre 2024
A Universe in a Transformer
Oasis is an interactive world model developed by Decart and Etched. Based on diffusion transformers, Oasis takes in user keyboard input and generates gameplay in an autoregressive manner. pic.twitter.com/B28bcFsFggOasis A Universe in a Transformer Oasis is an interactive world model developed by Decart and Etched. Based on diffusion transformers, Oasis takes in user keyboard input and generates gameplay in an autoregressive manner.