← Back to Evaluation

stride-based eval

Evaluation
Used in
54 PRs
Best BPB
0.0214
Avg BPB
0.9794

Hyperparameters Across PRs

pr_numberparameters
162{"stride":64}
209{"stride":64,"eval_seq_len":2048}
212{"stride":64}
285{"stride":0}
316{"stride":1024}
348{"stride":32}
393{"stride":512}
410{"stride":32}
415{"stride":32}
436{"stride":512}
483{"stride":64}
526{"stride":64}
530{"EVAL_STRIDE":0,"description":"Standard evaluation, not sliding window, for fast iteration"}
533{"stride":32}
554{"stride":256}
588{"stride":64,"eval_batch_seqs":256}
598{"stride":64}
600{"stride":512}
601{"stride":128}
615{"stride":64}
626{"stride":64,"mode":"sliding"}
680{"stride":64}
686{"stride":64}
769{"stride":2048,"seq_len":2048}
776{"stride":64}
779{"stride":64}
785{"stride":64}
790{"stride":64}
808{"stride":64}
811{"stride":128}
824{"stride":64}
851{"stride":64}
868{"two_pass":true,"rescore_chunks":72,"order":12}
901{"stride":64}
912{"stride":64}
931{"stride":64}
948{"stride":64}
962{"stride":64}
965{"stride":64}
968{"stride":64}
1050{"stride":64}
1128{"stride":64}
1174{"stride":64}
1253{"stride":64}
1300{"stride":64}
1349{"stride":64}
1451{"chunk_size":256,"eval_seq_len":1024,"batch_size":64}
1452{"stride":384}
1454{"stride":384}
1478{"stride":1024}
1547{"chunk_size":256,"eval_seq_len":1024,"batch_size":64}
1605{"stride":96}
1630{"stride":64}
1655{"stride":76}