Prompt engineering is ChatGPT training you.
LLMS
-
ChatGPT Needs Abductive Inference to Understand Text
By
–
ChatGPT, if you want to understand text you need to learn to do abductive inference.
-
ChatGPT’s Width Versus Depth Compared to Eliza
By
–
Eliza was an inch wide and an inch deep. ChatGPT is a mile wide and still an inch deep.
-
Discussion on the internal name ‘Assistant’ for OpenAI’s large language model
By
–
That's not true — I was one of the people who found the prompt the first few days and it's always begun with the phrase "Assistant is a large language model trained by OpenAI". ChatGPT avoids using any name, but its name has always been "Assistant" if you ask it.
-
Factual Hallucination in ChatGPT and Claude Discussed
By
–
Factual hallucination is still an issue for both ChatGPT and Claude. When asked, "Who won the Super Bowl in the year Justin Bieber was born?", Claude is usually wrong and ChatGPT is usually right. But even when right, ChatGPT is prone to including wrong, nonsensical details:
-
LangChain Notebook Replicates ChatGPT Functionality
By
–
@cyberfilmai we even have a notebook replicating ChatGPT! https://
langchain.readthedocs.io/en/latest/modu
les/memory/examples/chatgpt_clone.html
… -
AI Capability Evaluation: Moving Beyond Expiring Failure Cases
By
–
It may strengthen your argument if you provided examples where AI fails consistently over time, rather than failures cases that expire in such a short time, as DL improves. So far all the fails you used to prove GPT-2's lack of language understanding in 2020 are now easily solved
-
ChatGPT’s killer app: generating unread research papers
By
–
The killer app for ChatGPT is writing that no one reads. E.g., research papers.
-
LLM Tokenization Trick for EOT Emission
By
–
It’s not willing to produce what it thinks is an EOT token. You have to trick it by making it emit a sequence of tokens that re-tokenize to the EOT token after being turned into a string.