From 931cfaff00a385811ed5ee047782dd1aefcf03fc Mon Sep 17 00:00:00 2001 From: Hamza Date: Mon, 17 Jun 2024 15:09:06 +0200 Subject: [PATCH] add perplexity section --- FF2024/emails.html | 2 +- FF2024/exercise_session_1.html | 3 +- FF2024/perplexity.html | 357 +++++++++++++++++++++++++++++++++ 3 files changed, 359 insertions(+), 3 deletions(-) create mode 100644 FF2024/perplexity.html diff --git a/FF2024/emails.html b/FF2024/emails.html index fcbab5e..70ce129 100644 --- a/FF2024/emails.html +++ b/FF2024/emails.html @@ -83,7 +83,7 @@

Using LLMs to write emails

LLMs can help you write emails quickly and take on additional information to change the tone, length and translate the emails.

-

Here are some things that you can try out using the LLM you are using.

+

Here are some things that you can try out using the LLM you have access to.

Writing an email

To create an email you can give a simple prompt to the LLM. For example

diff --git a/FF2024/exercise_session_1.html b/FF2024/exercise_session_1.html index 48fde1e..fe752d4 100644 --- a/FF2024/exercise_session_1.html +++ b/FF2024/exercise_session_1.html @@ -53,8 +53,7 @@

Exercise session 1

Writing emails ->

-

Try text generation with citations

-

Try out Perplexity, another LLM that prides itself with accuracy and source citations. It is supposed to be better than the other popular LLMs for scientific work as it claims to provide references.

+

Try text generation with citations ->

Try an LLM fine-tuned with specific knowledge

diff --git a/FF2024/perplexity.html b/FF2024/perplexity.html new file mode 100644 index 0000000..1a9acfd --- /dev/null +++ b/FF2024/perplexity.html @@ -0,0 +1,357 @@ + + + + + + + + + +website + + + + + + + + + + + + + + + + + + + +
+ +
+ + + +

Exercises / Exercise session 1 /

+
+

Try text generation with citations

+

Try out Perplexity, another LLM that prides itself with accuracy and source citations. It is supposed to be better than the other popular LLMs for scientific work as it claims to provide references.

+

Here are some example prompts you can try out using the perplexity.

+
+ 
# Explain how olokizumab works.
+
+

Another prompt you can try

+
+ 
# In the context of RA, can serology help predict treatment outcomes among patients starting TNFi treatment?.
+
+ +
+

Chain of thought

+

Here is another example of extracting information and then answering questions with chain of thought shown by the LLM.

+ +

USMLE Question: A 27-year-old IV drug abuser female gave birth to a 7 lb 2 oz (3.2 kg) baby girl. The obstetrician is worried that the child may have been infected due to the mother’s habitual use of antiretroviral medications. Which of the following assays would hospital labs use to detect the genetic material of HIV if the child had been infected? +

    +
  • A) Enzyme-linked immunosorbent assay (ELISA)
    • +
  • B) Rapid HIV antibody test
    • +
  • C) Polymerase chain reaction (PCR)
    • +
  • D) Southern blot
    • +
+

+ + +
+ 
# 1.	Answer the question
+  2.	Ansewr the question with an added “think step by step”.
+
+ +
+ +
+ +
+ + +
+ + + + \ No newline at end of file