L23 cosmetic

lectures/459.bib

@@ -1273,7 +1273,7 @@ @inproceedings{coz
 }
 
 @article{gptforgood,
-title = {ChatGPT for good? On opportunities and challenges of large language models for education},
+title = {{ChatGPT} for good? On opportunities and challenges of large language models for education},
 journal = {Learning and Individual Differences},
 volume = {103},
 pages = {102274},

lectures/L23-slides.tex

@@ -256,7 +256,7 @@ \part{Password Cracking}
 
 You do not even need to make them yourself anymore (if you don't want) because you can download them on the internet... they are not hard to find. 
 
-They are large, yes, but in the 25 - 900 GB range. 
+They are large, yes, but in the 25--900 GB range. 
 
 
 \end{frame}
@@ -297,7 +297,7 @@ \part{Password Cracking}
 
 \begin{itemize}
 	\item Table generation on a GTX295 core for MD5 proceeds at around 430M links/sec.
-	\item Cracking a password 'K\#n\&r4Z': real: 1m51.962s, user: 1m4.740s. sys: 0m15.320s
+	\item Cracking a password 'K\#n\&r4Z': \\ \qquad real: 1m51.962s, user: 1m4.740s. sys: 0m15.320s
 \end{itemize}
 
 Yikes.
@@ -426,7 +426,7 @@ \part{Large Language Models}
 
 Such large language models have existed before, but ChatGPT ended up a hit because it's pretty good at being ``conversational''.
 
-This is referred to as Natural Language Processing (NLP).
+This is within the ambit of Natural Language Processing (NLP).
 
 \end{frame}
 
@@ -700,7 +700,7 @@ \part{Large Language Models}
 \begin{frame}
 \frametitle{Next Idea}
 
-We seem to be memory limited -- let's see what we can do?
+We seem to be memory limited---let's see what we can do?
 
 First idea: \alert{gradient accumulation}. 
 
@@ -731,7 +731,7 @@ \part{Large Language Models}
 \begin{frame}
 \frametitle{I got Suspicious}
 
-I got suspicious about the 128 dropoff in memory usage and it made me think about other indicators -- is it getting worse somehow? 
+I got suspicious about the 128 dropoff in memory usage and it made me think about other indicators---is it getting worse somehow? 
 
 The output talks about training loss...
 
@@ -791,7 +791,7 @@ \part{Large Language Models}
 \end{tabular}
 \end{center}
 
-Interesting results -- maybe a little concerning?
+Interesting results---maybe a little concerning?
 
 \end{frame}