New in Predibase! Fine-tuning is now up to 5x faster! All users can fine-tune with A100s at no extra cost View and manage deployments in the UI
LLMS
-
Technical Knowledge Matters for LLM Understanding and Limitations
By
–
I am not sure the technical explanations are needed to use LLMs well (in fact I am pretty sure they aren’t) but they are useful for understanding their power and limitations.
-
LLMs Gullibility: The HR Truth Detection Challenge
By
–
A key skill in HR (and anything related to management) is being able to figure out when people are lying – and then piece together the ground truth LLMs are inherently gullible
-
Python Machine Learning MLOps CV NLP LLMs Daily Tutorials
By
–
If you are interested in: – Python – Machine Learning – MLOps – CV/NLP – LLMs Find me → @Sumanth_077 Everyday, I share tutorials on above topics! Like/RT the first tweet to help this reach more people!
-
ROBOSHOT: Optimizing Foundation Models Without Fine-Tuning
By
–
Coming up this Friday! PhD student Dyah Adila will explain how data scientists can get better performance out of foundation models without any additional fine-tuning. The technique, called ROBOSHOT, applies a kind of lens to model outputs Register now!
-
LLMs as task multipliers and the future of ACI
By
–
A very good confirmation that existing LLMs can deal with a high percentage of tasks we will be dealing with. A good ACI will be a multiplier
-
Mistral Partners with AWS to Expose Models on Amazon Bedrock
By
–
Very happy to partner with @awscloud to expose Mistral models on Amazon Bedrock, as we continue to bring our technology to every developer.
-
Gemini 1.5 Pro pricing makes Claude Haiku vastly superior
By
–
Gemini 1.5 Pro pricing is out. Claude Haiku is 28x cheaper. Very few reasons to use 1.5 Pro at this price.
-
RAG Remains Essential for Large Language Models
By
–
I can't imagine RAG becoming irrelevant. Even long context LLMs still need to have documents selectively added to that context to increase the chance of a correct answer, and I don't see any other way of answering questions using content that's only a few minutes/hours old
-
Anthropic Discovers Many-Shot Jailbreaking Technique for LLMs
By
–
Anthropic researchers have discovered a new "jailbreaking" technique called "many-shot jailbreaking". It can evade the safety guardrails of LLMs by exploiting expanded context windows. Pretty wild.