chore(doc): clarify micro_batch_size (#1579) [skip ci]
Browse files- docs/config.qmd +1 -0
docs/config.qmd
CHANGED
|
@@ -268,6 +268,7 @@ torch_compile_backend: # Optional[str]
|
|
| 268 |
# If greater than 1, backpropagation will be skipped and the gradients will be accumulated for the given number of steps.
|
| 269 |
gradient_accumulation_steps: 1
|
| 270 |
# The number of samples to include in each batch. This is the number of samples sent to each GPU.
|
|
|
|
| 271 |
micro_batch_size: 2
|
| 272 |
eval_batch_size:
|
| 273 |
num_epochs: 4
|
|
|
|
| 268 |
# If greater than 1, backpropagation will be skipped and the gradients will be accumulated for the given number of steps.
|
| 269 |
gradient_accumulation_steps: 1
|
| 270 |
# The number of samples to include in each batch. This is the number of samples sent to each GPU.
|
| 271 |
+
# Batch size per gpu = micro_batch_size * gradient_accumulation_steps
|
| 272 |
micro_batch_size: 2
|
| 273 |
eval_batch_size:
|
| 274 |
num_epochs: 4
|