jillbore

4 ms·

1. 1.▲
   by jillbore 3y ago

   This is about seeing inference benefits from unstructured sparsity.
2. 2.▲
   by jillbore 3y ago

   That is the accuracy of the non-finetuned model i.e. mpt-7b. Once finetuned, it can get around 30% accuracy on 0-shot. Not sure what 8-shot performance is for the finetuned model. https://arxiv.org/pdf/2310.06927.pdf
3. 3.▲
   MLPerf: Neural Magic Sparsity, Nvidia Hopper, and Alibaba Network Introduced(zdnet.com)

   4 points by jillbore 4y ago|0 comments
4. 4.▲
   Neural Magic raises $15M seed round to accelerate AI inference on CPUs(techcrunch.com)

   6 points by jillbore 7y ago|1 comments