2 ms·
These are still FP16/32 models, almost certainly a few times slower and larger than the latest N bit quantized GGMLs.
by moffkalast 3y ago
These are still FP16/32 models, almost certainly a few times slower and larger than the latest N bit quantized GGMLs.