My understanding is that LLM prompt performance is linear with respect to the length of the prompt, so if you want fast responses you won't want to dump more into the prompt than necessary no matter how long it can be
LLMS
-
GPT-4 Code Interpreter Performance Degradation Discussed
By
–
Oh cool! I would say there was a time when code interpreter was so much better at coding than regular GPT-4 that I used it a lot, until it got nerfed, lol
-
Building Product Features with Haiku AI Model
By
–
And I am building every new feature in my product around Haiku, because it's just THAT good. Realistically, there is not a real difference there between GPT-4 and Opus, but for us working in the space, even a 2% increase in productivity means a lot.
-
Claude Opus Outperforms GPT-4 in Latest Benchmark Tests
By
–
It's official, Claude Opus > GPT-4. Even Haiku performs ridiculously well… 1/2 the cost of GPT-3.5-Turbo, yet beats Mistral-Large and GPT-4 (6/13).
-
Fine-Grained Human Feedback Training with Density Diversity Reward Functions
By
–
#MosiacAI researchers and @UW wanted to determine a better approach to Fine-Grained Human Feedback. See how they researched an approach that utilized training and learning from two different reward functions: density and diversity. https://
dbricks.co/49HyKOj -
LLM Developers Network Event in SF Bay Area
By
–
Network and learn from #LLM developers working on the most popular #opensource projects—all while taking in a view of the SF Bay. Click the link to see the talks and save your spot. Speakers include: @predibase @llama_index @guardrails_ai @tryolabs
. https://
pbase.ai/49etot4 -
Experimental LLM Rickrolls User Through Malicious Source Citations
By
–
This reminds me of an experimental LLM for HyperWrite that I trained a few weeks ago to better cite sources. @JasonKuperberg clicked on one of the links it provided, and it rickrolled him… I guess this is going to start happening more often now!
-
Clarification on text-davinci-002-render-sha model identifier
By
–
Made a mistake, I meant “text-davinci-002-render-sha”
-
Technical insight into frontend API model routing
By
–
The frontend is also checking an internal model slug coming from the api. By default it is “text-davinci-002-render-paid”
