Saving weights and logs of epoch 1
Browse files
events.out.tfevents.1625848981.t1v-n-b95d739e-w-0.195447.3.v2 → events.out.tfevents.1625926631.t1v-n-b95d739e-w-0.271163.3.v2
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:058091f2aa628751ffce3a9b123432d04c86420a4355fec8763decfd8a17d9be
|
3 |
+
size 3581817
|
flax_model.msgpack
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 711588089
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8678c0f46f5db25de579fa7beaba06e1c3b2cf73dfa0635f07e5822b760dce16
|
3 |
size 711588089
|
nohup.out
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:355b4eaf707e807efbb0fc5c51bfc2a4a766c6e846b5f6c6e8484f7016d03a01
|
3 |
+
size 2037187
|
run.sh
CHANGED
@@ -8,17 +8,13 @@ python3 run_mlm_flax.py \
|
|
8 |
--dataset_name="oscar" \
|
9 |
--dataset_config_name="unshuffled_deduplicated_id" \
|
10 |
--max_seq_length="128" \
|
11 |
-
--weight_decay="0.01" \
|
12 |
--preprocessing_num_workers="64" \
|
13 |
--per_device_train_batch_size="128" \
|
14 |
--per_device_eval_batch_size="128" \
|
15 |
-
--learning_rate="
|
16 |
--warmup_steps="1000" \
|
17 |
--overwrite_output_dir \
|
18 |
--num_train_epochs="10" \
|
19 |
-
--logging_steps="500" \
|
20 |
-
--save_steps="10000" \
|
21 |
-
--eval_steps="10000" \
|
22 |
--dtype="bfloat16" \
|
23 |
--push_to_hub
|
24 |
|
|
|
8 |
--dataset_name="oscar" \
|
9 |
--dataset_config_name="unshuffled_deduplicated_id" \
|
10 |
--max_seq_length="128" \
|
|
|
11 |
--preprocessing_num_workers="64" \
|
12 |
--per_device_train_batch_size="128" \
|
13 |
--per_device_eval_batch_size="128" \
|
14 |
+
--learning_rate="7e-5" \
|
15 |
--warmup_steps="1000" \
|
16 |
--overwrite_output_dir \
|
17 |
--num_train_epochs="10" \
|
|
|
|
|
|
|
18 |
--dtype="bfloat16" \
|
19 |
--push_to_hub
|
20 |
|