17 – Cosmian (
@CosmianOfficial
) presented by Hugo Rosenkranz-Costa, showcased how one can run LLMs with an data encryption layers, so user data remains encrypted and private Presentation: https://
docs.google.com/presentation/d
/1pAv1DIR7b1h2maAygRRcs6cNyCXvMG8XUZ3M7A6JaH4/edit#slide=id.g288027c00d1_0_4
… https://
cosmian.com
LLMS
-
Cosmian: Running LLMs with Encrypted Data Privacy Layers
By
–
-
NuMind Releases State-of-the-Art Multilingual Named Entity Recognition Models
By
–
12 – NuMind, presented by Sergei Bogdanov and team, just released state of the art multilingual named entity recognition models https://
numind.ai -
TitanML Inference Server: LLM Output Control and Performance
By
–
10 – TitanML presented by their CEO @MeryemArik9 and team, two demos of their takeoff Inference Server: 1. Controlling LLM outputs to fit a regex form 2. Performance of our rust inference server vs FastAPI website: https://
titanml.co profile: -
H4rmony: Aligning LLMs with Ecolinguistic Principles
By
–
7 – H4rmony – presented by Jorge Vallego, is a project to align LLMs with ecolinguistic principles Their presentation, with examples of the LLMs fine-tuned to take the environment into account: https://
slideshare.net/NeoV2/the-h4rm
ony-projectpptx
… And the dataset is open: -
Google PaLM Demonstrates Superior Multilingual Language Proficiency
By
–
Google’s PaLM beats all other contenders on non-English language proficiency, supporting the broadest range of languages of the online population effectively. This seems to have not been previously reported. (3/7)
-
Claude 2 Achieves Second Place in Reasoning Benchmarks
By
–
Claude 2 takes 2nd place on reasoning, encompassing tasks like solving math questions and programming challenges. (4/7)
-
Llama 2 70B Achieves Third Place in Writing Creativity Evaluation
By
–
Llama 2 70b manages 3rd place on writing and creativity, assessed in our expert human evaluation conducted with @SurgeAI
. (5/7) -
Top LLMs Compared: Reasoning, Writing, Creativity, Non-English
By
–
How do the top LLMs compare? We partnered with @SurgeAI for a systematic evaluation of each across four attributes: Reasoning, Writing, Creativity, and Non-English Fluency, and the results are in (1/7)
-
GPT-4 dominates AI model competition across all metrics
By
–
GPT4 wins the competition on all dimensions (not counting speed and price) but there are some notable callouts in 2nd place. (2/7)
-
Mistral AI Paper Generates Nearly 500 Community Implementations
By
–
Paper of the week by @MistralAI – almost 500 spaces based on it! You can ask questions about it to the authors here: https://
huggingface.co/papers/2310.06
825
…