Update config/training/model_training.ini
Browse files
config/training/model_training.ini
CHANGED
|
@@ -1,21 +1,21 @@
|
|
| 1 |
[ModelTrainingConfig]
|
| 2 |
-
block_size =
|
| 3 |
d_model = 256
|
| 4 |
n_layer = 4
|
| 5 |
n_head = 4
|
| 6 |
d_ff = 1024
|
| 7 |
dropout = 0.1
|
| 8 |
-
batch_size =
|
| 9 |
grad_accum_steps = 2
|
| 10 |
-
epochs =
|
| 11 |
-
max_lr =
|
| 12 |
-
min_lr =
|
| 13 |
warmup_ratio = 0.05
|
| 14 |
-
weight_decay = 0.
|
| 15 |
grad_clip = 1.0
|
| 16 |
-
label_smoothing = 0.
|
| 17 |
|
| 18 |
[AfterTrainingConfig]
|
| 19 |
max_new_tokens = 60
|
| 20 |
-
temperature = 0.
|
| 21 |
-
top_k =
|
|
|
|
| 1 |
[ModelTrainingConfig]
|
| 2 |
+
block_size = 256
|
| 3 |
d_model = 256
|
| 4 |
n_layer = 4
|
| 5 |
n_head = 4
|
| 6 |
d_ff = 1024
|
| 7 |
dropout = 0.1
|
| 8 |
+
batch_size = 16
|
| 9 |
grad_accum_steps = 2
|
| 10 |
+
epochs = 40
|
| 11 |
+
max_lr = 2e-4
|
| 12 |
+
min_lr = 2e-5
|
| 13 |
warmup_ratio = 0.05
|
| 14 |
+
weight_decay = 0.05
|
| 15 |
grad_clip = 1.0
|
| 16 |
+
label_smoothing = 0.0
|
| 17 |
|
| 18 |
[AfterTrainingConfig]
|
| 19 |
max_new_tokens = 60
|
| 20 |
+
temperature = 0.40
|
| 21 |
+
top_k = 20
|