It’s not best fit or regression lines since different model sizes/capabilities are on the graph. Why would you best fit to the average of the new 8B, 70B and 405B llama 3.1 models? These are just rough trends lines of the frontier
MARKET TRENDS
-
Global AI Summit Emphasizes Innovation-Friendly Regulatory Frameworks
By
–
During the recent Global IndiaAI Summit, the panel “Collaborative AI on Global Partnership” emphasized the need for regulatory frameworks that do not hinder innovation. Dignitaries from ministries and international corporations participated in the session, focusing on the future
-
Open-source Models Finally Beating Closed-source Models
By
–
One year ago, I created a video in which I said, hopefully open-source models are going to beat closed source models soon. I got this comment and a response that it was only possible in dreams. Today, I responded to the response 🙂
-
India’s Fintech Revolution: Digital Payment Innovation Ahead
By
–
Each time I visit India from Singapore, it feels like stepping into a futuristic fintech utopia. At payment counters, I feel like a dinosaur. While people flash all kinds of fancy payment options, I nervously take out my credit card, silently praying ‘please don’t judge me’!
-
Llama 3.1 Performance Benchmarks Compared to Sonnet and GPT-4o
By
–
El gráfico, aunque sólo se base MMLU, es consistente con las evaluaciones en el resto de benchmarks donde Llama 3.1 alcanza a Sonnet 3.5/GPT-4o, así que las conclusiones sí me parecen validas. Respecto al open source, efectivamente, estrictamente hablando, no lo es y encajaría
-
Meta’s Next Generation AI Models: Resources and Release Strategy
By
–
Todo depende de qué tan alto sea el escalón de la próxima generación de modelos. Lo que está claro es que Meta cuenta con el talento y los recursos para seguir la estela. Sólo falta que sigan conservando la voluntad de liberar modelos cada vez más potentes.
-
ChatGPT’s Rapid Impact: Less Than Two Years Later
By
–
Sorprende pensar que no han pasado ni 2 años desde la salida de ChatGPT.
-
OpenAI versus Llama: Can competitors keep pace with Meta’s open model?
By
–
¿Aguantarán, OpenAI y competidores, meses sin elevar las capacidades de sus modelos mientras la Llama sigue libre? ¿Encontraremos benchmark más difíciles que diferencien las capacidades de los modelos actuales y de los que están por venir? ¿Se dejará Zuckerberg barbita para
-
Generative AI Was More Enjoyable Before Recent Developments
By
–
Generative AI with all its imperfections was more fun a few years ago
-
GPT-4o-mini Rankings and Model Alignment Methods Evaluation
By
–
Knowing the great people behind lmsys, I don't think there is anything really fishy in the gpt-4o-mini ranking like some people have been saying Likely more of a symptom that as models (and alignement methods) are getting much better it's been harder and harder to tell models