File size: 831 Bytes
04eb95f | 1 2 3 4 5 6 7 8 9 10 11 | model,batch,window_tokens,hit_after_eod,hit_within_doc,frac_tokens_after,deficit
"8.1M active, coarse (6 of 64 experts)",16k,4,0.34035,0.38206,0.00232,0.04171
"8.1M active, coarse (6 of 64 experts)",16k,16,0.39419,0.38185,0.00928,-0.01234
"8.1M active, coarse (6 of 64 experts)",16k,64,0.40543,0.38106,0.03713,-0.02437
"3.9M active, fine-grained (18 of 192 experts)",16k,4,0.21820,0.34776,0.00232,0.12956
"3.9M active, fine-grained (18 of 192 experts)",16k,16,0.26722,0.34821,0.00928,0.08099
"3.9M active, fine-grained (18 of 192 experts)",16k,64,0.32678,0.34826,0.03713,0.02148
"38.0M active, coarse (6 of 64 experts)",50k,4,0.33883,0.30409,0.00440,-0.03474
"38.0M active, coarse (6 of 64 experts)",50k,16,0.36712,0.30312,0.01759,-0.06400
"38.0M active, coarse (6 of 64 experts)",50k,64,0.34460,0.30119,0.07035,-0.04341
|