Unfortunately, I have to agree with this. Especially w/ longer context. I would bet that this is either extremely quantized, or significantly smaller than older GPT-4s. It has moments of genius, for sure, and the speed is amazing. But it definitely loses some of its magic.
LLMS
-
GPT-4o developed since 2022 by OpenAI
By
–
The development of GPT-4o commenced in 2022(!) They are so far ahead of what they release.
-
LLMs Transform Retail: GenAI Unlocks $660 Billion Potential
By
–
Discover how Large Language Models (LLMs) are transforming retail! From enhanced customer service to streamlined logistics, GenAI could unlock $660 billion for retailers. Our experts explore how to use advanced RAG pipelines and function calling for AI agents in retail. Delve
-
Comparative Analysis of GPT-4o Live Translation Capabilities
By
–
Current 4o is already pretty good at live translation. New model from the demo is not there yet
-
GPT-4o is far less vague than previous GPT-4
By
–
Ciertamente algo que se puede decir de GPT-4o es que es muuuucho menos vago que lo que era GPT-4 hace unos meses. Le preguntas una duda sobre un paper, te lo explica en detalle y además te suelta una implementación en pseudo-código
-
GPT-4o capability in detecting annotations
By
–
GPT-4o seem to be able to easily detect these annotations
-
ChatGPT iOS update allows photo annotation
By
–
ICYMI: Now you can annotate photos uploaded to ChatGPT on iOS.
-
GPT-4 All Tools: Web Browsing, DALLE3, and Autonomous Agents
By
–
gpt-４all tools = gpt4 + web browsing + dalle3 + ada ada-v2 + voice engine + search engine + autonomous agents =
-
Making Inference Performance Faster: Key Discussion Points
By
–
I also discussed a number of issues around how to make inference performance faster and why this was important.
-
Gemini Model Training: Challenges and Recent Advances
By
–
I gave an overview of some of the challenges and advances in our recent Gemini model training runs.