For data transmutation I use function calling https://
openai.com/blog/function-
calling-and-other-api-updates
…
LLMS
-
Function Calling for Data Transmutation with OpenAI API
By
–
-
GPT-4 excels at unstructured to structured data conversion
By
–
GPT-4 is one of the best unstructured to structured data adapters on the planet.
-
AI Trade-offs: Safety Boundaries vs Hallucinations
By
–
Actually this is working as intended. There are always trade-offs. Right now it is prone to gaslighting and hallucinations, which are clearly very bad. The flipside is that its also very boundaried on safety topics. We'll nail gaslighting and hallucinations soon…
-
LLMs Sound Smart But Lack Real Intelligence Planning Memory Reasoning
By
–
LLMs are designed to *sound* smart, which is not quite the same thing as actually being smart we all know enough bullshitters to know the difference planning, memory, reasoning, self-reflection… all key steps to get real intelligence (to be clear, smartest ppl working on it)
-
Pathways Models and PaLM-E Are Not Open-Source
By
–
All Pathways models including PaLM-E are not open-source as far as I know.
-
Emotion Modulation in Second Language Learning Tasks
By
–
Also, I wonder how emotion might be modulated when performing a similar task with second language learners. If it's a similar effect, it might be more about linguistic ability than age?
-
The Transformer Blueprint: A Holistic Guide to Neural Networks
By
–
A NEW ARTICLE The first article of Deep Learning Revision Research Blog(introduced recently) is out: "The Transformer Blueprint: A Holistic Guide to the Transformer Neural Network Architecture". In the article, we discuss the core mechanics of transformer neural
-
Google’s Bard AI Now Supports Voice and Visual Recognition
By
–
Now Google’s Bard AI chatbot can talk and respond to visual prompts https://
bit.ly/3OoklhV #AI #MachineLearning #DeepLearning #LLMs #DataScience -
Open-Source Model Fine-tuning Methods and Training Flexibility
By
–
Yes, any finetuning method is allowed, afaik. Technically, you could also just further (pre)train the model. Anything that runs <24 h on open source code and openly available data.