6 ms·
14b model performs comparably with 32b size. the improvement is huge
by watsonmusic 1y ago
14b model performs comparably with 32b size. the improvement is huge
- 85392_school 1y agoare we only comparing them in terms of text completion accuracy? does it also improve perf on benchmarks?