Self-training: F(x,y) is a filtering/ranking function, eg., what we call a reward/return. The input x may be chosen by humans, but the model generates the y’s and F ranks and selects for further rounds of self-training. F can be explicit or implicit (human in the loop as in RLHF)
PROMPT ENGINEERING
-
AI-Powered Long-Form Content and Podcast Summarization
By
–
I want to try it with a realllly long ramble. Would also love to have it listen to podcast convos and write the summary for me.
-
LangChain Tools for Data Loading and AI Chat Applications
By
–
AirByte Loader: https://
langchain.readthedocs.io/en/latest/modu
les/document_loaders/examples/airbyte_json.html
… ChatVectorDBChain Notebook: https://
langchain.readthedocs.io/en/latest/modu
les/chains/combine_docs_examples/chat_vector_db.html
… Chat-Your-Data Challenge: https://
blog.langchain.dev/chat-your-data
-challenge/
… -
Gen-1 Customization Mode Unlocks Higher Fidelity AI Results
By
–
Mode 05: Customization
Unleash the full power of Gen-1 by customizing the model for even higher fidelity results. -
Gary Marcus counterexamples: ChatGPT fails, Bing succeeds
By
–
Contraejemplos de Gary Marcus, parte 3. Ahora por petición de @antor en inglés. ChatGPT SÍ falla en ambos ejemplos y Bing aguanta el tipo sin problemas.
-
Tokenizer Word Reversal Issue: Model Perception and Prompt Solutions
By
–
Escribir palabras al revés: Mismo fallo en ambos. Como dato curioso, esto ocurre por la forma en la que se subdividen las palabras por el tokenizador y cómo el modelo las percibe. Se puede solucionar fácil con instrucciones intermedias. (sig. tweet)
-
Failed ChatGPT Prompts: Sharing Examples for Comparison
By
–
Can you share with me prompts that have failed for you with ChatGPT? I want to do some comparisons I'll be posting examples in this thread
-
Bing Chat and Education: Learning Beyond Homework Assistance
By
–
BING CHAT y EDUCACIÓN Sí, estos chatbots pueden hacer los deberes. Pero no perdamos el foco también en lo útiles que son de cara al aprendizaje del alumno. Aquí en conversación con Bing le pido que me evalúe sobre los conocimientos de Deep Learning que me ha explicado.
-
Skepticism about prompt vs human-written policy in ChatGPT RLHF
By
–
Thought about this more: It doesn’t make sense it’s the prompt, because inference costs. Still feels human-written — but why would you RLHF into reciting all policy verbatim? ChatGPT falls for the same trick and the text it recites is just parameters.