definitely, 1.5b's speed & size is a good balance for mobile. worth having downloaded too
LLMS
-
Humanity’s Last Exam: No AI model scores above 10%
By
–
Examples from Humanity’s Last Exam, designed by CAIS & Scale to be the *final* broad-subject, closed-end Q&A exam for LLMs written by humans. Today, no model scores above 10%. This is how hard a test has to be for frontier AI to score that low — in Classics, Ecology, and C.S.:
-
DeepSeek Model Efficiency and Industry Impact
By
–
Deepseek, le modèle chinois a rebattu les cartes de l’IA. Apparement c’est la panique chez Meta. 10 millions pour le développement contre plusieurs milliard chez les autres, Son modèle distillé en 1.5B tourne sur un rasberry en local en temps réel. Hâte de voir la suite.
-
More Bits Enable Greater Precision in Larger AI Models
By
–
more bits = more decimal places = more precision = bigger model
-
Tiny AI Models Struggle with Coding Tasks Compared to Reasoning
By
–
It will be a while before tiny models can code well, programming needs much more mental horsepower than reasoning
-
Citations Now Supported for Claude 3.5 Sonnet and Haiku
By
–
Citations is currently supported for the new Claude 3.5 Sonnet and Claude 3.5 Haiku. Learn more in our docs:
-
Citations Feature Outperforms Custom Methods in Internal Evaluations
By
–
In our internal evals, we found Citations performance exceeds custom citation methods for generating accurate/consistent citations. It also requires significantly fewer tokens than having to implement your own citations prompt since Claude already knows the format to use.
-
Claude Citations Feature for PDFs and Text Documents
By
–
With Citations enabled, Claude can cite chunks of content from pdfs, plain text docs, and text chunks. The returned citations are easy to parse, and Claude makes sure to only highlight the part of the response that is applicable to the citation.
-
Claude Citations Feature Now Available for Developers via API
By
–
Under the hood, Claude is trained to cite sources. With Citations, we are exposing this ability to devs. To use Citations, users can pass a new "citations: {enabled:true}" parameter on any document type they send through the API.
-
Anthropic API Citations Feature Now Available for Grounded Responses
By
–
We've rolled out Citations in the Anthropic API. Citations allows Claude to ground its answers in user-provided information and provide precise references to the sentences and passages used in its responses. Here's how it works: