AI lost a monumental scientist in Doug Lenat.
His work on Cyc deeply inspires me till today — not so much because he worked on symbolic AI but because of how far he'd taken it, through a multi-decade dedication.
I hope the propreitary Cyc work (ontologies, etc.) get open-sourced
OPEN SOURCE
-
Doug Lenat’s Cyc Legacy: Symbolic AI Pioneer’s Ontological Work
By
–
-
Train KerasNLP Models Faster with Data Parallel Training
By
–
Learn how to train KerasNLP models faster with data parallel training: https://
keras.io/examples/nlp/d
ata_parallel_training_with_keras_nlp/
… -
Multilingual AI Dataset Benchmark on HuggingFace Platform
By
–
That’s a really cool datasets and benchmark! Super nice to push progress on multilinguality. Would you be maybe interested in hosting there http://
HuggingFace.co/facebook/ and/or think about having an automatically updated community leaderboard around it -
Open-Source AI Technologies Reshaping Global Power Dynamics
By
–
"This is an age when the most powerful technologies are open-sourced in months, when millions have access to the cutting edge, and that cutting edge is the greatest force amplifier ever seen. This is the greatest reshuffling of power in history…"
-
Belebele: Multilingual Benchmark for High and Low-Resource Languages
By
–
Belebele is composed of carefully crafted multiple-choice questions & answers based on FLORES-200 passages. This work enables the evaluation of NLP systems and large language models in high and low-resource languages. More details in the paper
-
Llama 2 Open Source Model for Flexible Bot Development
By
–
If you're creating a bot, this option gives you more flexibility in case the other models don't meet your needs. And since Llama 2 is open source, you don't have to worry about your bot changing behavior or the base model going away in the long term. (2/2)
-
Condor Galaxy 1 Powers BTLM and Jais Open-Source LLMs
By
–
Condor Galaxy 1 (CG-1) was used to train BTLM, the top-performing open-source 3B parameter LLM. Now, CG-1 has been used to train Jais, the best open-source Arabic model. CG-1 is moving the open-source community forward. Contact us to train on CG-1 http://
cerebras.net/contact-us -
CodeLlama PR merged with 4-bit quantization inference benchmarks
By
–
The CodeLlama PR just got merged: https://
github.com/Lightning-AI/l
it-gpt/pull/472
… When I tried it with bnb's 4-bit Normal Float quantization, the 34B Instruct and Python variants used about 20 Gb for inference: -
Create Your Own AutoTrain Advanced Space and Train Models
By
–
Create your own AutoTrain Advanced Space and start training models!!!
-
JAIS Arabic Language Model Released with Paper and Demo
By
–
Paper, sandbox, and model here: https://
arxiv.org/abs/2308.16149
#new_tab
… https://
arabic-gpt.ai https://
huggingface.co/inception-mbzu
ai/jais-13ba
…