GPT, Generative AI and LLMs! A Year Earlier! @starconfs #BigData #Analytics #DataScience #AI #MachineLearning #NLProc #IoT #IIoT #Python #RStats #TensorFlow #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode
LLMS
-
Claude Projects Excel at FastHTML Questions Beyond Knowledge Cutoff
By
–
I've discovered that when I create an @AnthropicAI Claude project with one example .py file and a few of the markdown files, it does a great job of answering FastHTML questions, even although the lib predates Claude's knowledge cutoff!
-
Sonnet vs GPT-4: Performance and Speed Comparison Analysis
By
–
Sonnet isn't even setup as a default model, I had to add it. I don't think it's Sonnet 3.5 quality: for me it doesn't solve problems any more consistently than with GPT 4. Sonnet is quite a bit faster though, maybe it's UX/UI and no longer hitting limits or waiting as much…
-
Natural Language Programming with Generative Models
By
–
Writing significant software systems using natural language instead of computer-oriented programming languages is one of the most significant aspects of generative models. It opens up the ability to get computers to do what you want to many more people!
— Jeff Dean (@JeffDean) 25 août 2024
(Video is fun to watch!) https://t.co/a7QbUK0hNkWriting significant software systems using natural language instead of computer-oriented programming languages is one of the most significant aspects of generative models. It opens up the ability to get computers to do what you want to many more people!
(Video is fun to watch!) -
LLMs are a broken way to build a knowledge base
By
–
LLMs are a broken way to build a knowledge base.
-
Parameter Size vs Performance in AI Model Graphs
By
–
also, since Kevin Scott didn’t afaik specify what’s plotted in his hypothetical graph, maybe it’s literally just parameter size and not performance lol
-
Model Size Transparency Gap in Recent AI Releases
By
–
Do we have firm data on size for all the recent models? not all have been fully forthcoming, as I recall.
-
Thread Demonstrating k-shot Prompting Method
By
–
I made another thread to illustrate the k-shot prompting method mentioned above:
-
Repeated sampling and k-shot prompting affect creativity in instruction-tuned models
By
–
Repeated sampling from instruction-tuned models can yield low-diversity, stereotyped output. For creative tasks, k-shot prompting base models will improve creativity (measured e.g. as vocabulary size), though generally with worse adherence to the (now implicit) instructions.
-
Prompt design: using stop sequences to constrain model output
By
–
Note in the final image:
– Our prompt ends with “Output:” — constraining the model’s response to contain only the output we want
– After our desired response, the base model continues to produce more input/output pairs (only one shown) — avoiding this requires a stop sequence.