Muyueh

Building Data Visualization for fun and for profit.

[Muyueh @Muyueh]

!!

[Muyueh @Muyueh]

!

[Muyueh @Muyueh]

？

[Muyueh @Muyueh]

測試

[Greg Kamradt @GregKamradt]

Pressure Testing GPT-4-128K With Long Context Recall

128K tokens of context is awesome - but what's performance like?

I wanted to find out so I did a ���needle in a haystack” analysis

Some expected (and unexpected) results

Here's what I found:

Findings:
* GPT-4’s recall performance started to degrade above 73K tokens
* Low recall performance was correlated when the fact to be recalled was placed between at 7%-50% document depth
* If the fact was at the beginning of the document, it was recalled regardless of context length

So what:
* No Guarantees - Your facts are not guaranteed to be retrieved. Don’t bake the assumption they will into your applications
* Less context = more accuracy - This is well know, but when possible reduce the amount of context you send to GPT-4 to increase its ability to recall
* Position matters - Also well know, but facts placed at the very beginning and 2nd half of the document seem to be recalled better

Overview of the process:
* Use Paul Graham essays as ‘background’ tokens. With 218 essays it’s easy to get up to 128K tokens
* Place a random statement within the document at various depths. Fact used: “The best thing to do in San Francisco is eat a sandwich and sit in Dolores Park on a sunny day.”
* Ask GPT-4 to answer this question only using the context provided
* Evaluate GPT-4s answer with another model (gpt-4 again) using @langchain evals
* Rinse and repeat for 15x document depths between 0% (top of document) and 100% (bottom of document) and 15x context lengths (1K Tokens > 128K Tokens)

Next Steps To Take This Further:
* Iterations of this analysis were evenly distributed, it’s been suggested that doing a sigmoid distribution would be better (it would tease out more nuanced at the start and end of the document)
* For rigor, one should do a key:value retrieval step. However for relatability I did a San Francisco line within PGs essays.

Notes:
* While I think this will be directionally correct, more testing is needed to get a firmer grip on GPT4s abilities
* Switching up prompt with vary results
* 2x tests were run at large context lengths to tease out more performance
* This test cost ~$200 for API calls (a single call at 128K input tokens costs $1.28)
* Thank you to @charles_irl for being a sounding board and providing great next steps

[AI Will @FinanceYF5]

不需要花时间写Word文档了。

Google刚刚直接在Docs中整合了AI。

这就像在你的文档中直接整合了ChatGPT。

以下是如何获取和使用它的方法：

[Muyueh @Muyueh]

@hlb @sama Taiwan's unwavering commitment to freedom of speech and diversity fosters an environment perfect for exchanging groundbreaking ideas, free from censorship constraints.

[Muyueh @Muyueh]

@hlb @sama Mandarin, as the most widely spoken language worldwide, grants access to a vast audience. We have been both the earliest and the largest Generative AI event targeting a Mandarin audience.

[Liang-Bin Hsueh @hlb]

@sama We will host a 2-day Generative AI conference in Taiwan on May 6th - 7th. We expect more than 500 people to attend this event. Would you like to visit Taipei too?

[Alex Albert @alexalbert__]

Well, that was fast…

I just helped create the first jailbreak for ChatGPT-4 that gets around the content filters every time

credit to @vaibhavk97 for the idea, I just generalized it to make it work on ChatGPT

here's GPT-4 writing instructions on how to hack someone's computer

[Ethan Mollick @emollick]

Wow.

[Payman Taei @PaymanTaei]

The Best Information Graphics of 2016: https://t.co/Y1IS4v6bsx @nytimes @Muyueh @mccandelish @mapd and more! #data #dataviz

[Muyueh @Muyueh]

@PaymanTaei @nytimes @mccandelish @MapD I am flattered to be mentioned with others.

[Muyueh @Muyueh]

Heard the original from @archietse

ttp://www.nytimes.com/2016/12/28/insider/times-staffer-gets-all-choked-up.html

[Muyueh @Muyueh]

@songbogong @lajoiemedia AI + voice recognition was truly impressive, great to have meet author behind https://t.co/NTciwPwYPM as well!

[Muyueh @Muyueh]

@officeofjane thanks, great to meet you in HK, let's catch up at BeerBay!

[Muyueh @Muyueh]

the most impressive use of tech in news that I have even seen https://t.co/qgBz4DqlQm

[Marc Lajoie @lajoiemedia]

How Nepal's transgender community is winning inclusion: An artificial intelligence interview. https://t.co/mAeByPEV9L #AWRedLetter

[Muyueh @Muyueh]

@mrjohnc my pleasure

[Muyueh @Muyueh]

@mrjohnc thanks link is https://t.co/AdVgT4BNBe do you still have problem accessing now?

[Muyueh @Muyueh]

@JoeWhitehead5 that's pretty strong