Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions PyTorch/LanguageModeling/BERT/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -802,9 +802,9 @@ Our results were obtained by running the `scripts run_pretraining.sh` training s

| GPUs | Batch size / GPU (TF32 and FP16) | Accumulation steps (TF32 and FP16) | Sequence length | Throughput - TF32(sequences/sec) | Throughput - mixed precision(sequences/sec) | Throughput speedup (TF32 - mixed precision) | Weak scaling - TF32 | Weak scaling - mixed precision
|------------------|----------------------|----------------------|-------------------|-----------------------------------------------|------------------------------------|---------------------------------|----------------------|----------------------------------------------
|1 | 65232 and 65536 | 1208 and 512| 128| 234 |415 |1.77 |1.00 | 1.00
|4 | 16308 and 16384 | 302 and 128| 128| 910 |1618 | 1.77| 3.89| 3.90
|8 | 8154 and 8192 | 151 and 64| 128| 1777 |3231 | 1.81| 7.59| 7.79
|1 | 65232 and 65536 | 1208 and 1024| 128| 234 |415 |1.77 |1.00 | 1.00
|4 | 16308 and 16384 | 302 and 256| 128| 910 |1618 | 1.77| 3.89| 3.90
|8 | 8154 and 8192 | 151 and 128| 128| 1777 |3231 | 1.81| 7.59| 7.79
|1 | 32768 and 32768| 4096 and 2048| 512| 41 |78 |1.90 |1.00 | 1.00
|4 | 8192 and 8192| 1024 and 512| 512| 159 |308 | 1.93| 3.88| 3.95
| 8| 4096 and 4096| 512 and 256| 512| 318 |620 | 1.94| 7.95| 7.76
Expand Down