Saving weights and logs of step 10000
Browse files
events.out.tfevents.1644662890.t1v-n-ccbf3e94-w-0.1307490.3.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:94f1dd0065038d65f57c3b0d41e411f83804552e6852555e2cd0ef5842c1be0b
|
3 |
+
size 1470136
|
flax_model.msgpack
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 498796983
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3f2523dae16815c083d982b608a29065d97c475203bab4289bcf0657159c4770
|
3 |
size 498796983
|
run_512.sh
CHANGED
@@ -11,16 +11,16 @@ python run_mlm_flax.py \
|
|
11 |
--per_device_train_batch_size="46" \
|
12 |
--per_device_eval_batch_size="46" \
|
13 |
--pad_to_max_length \
|
14 |
-
--learning_rate="0.
|
15 |
--warmup_steps="150000" \
|
16 |
--overwrite_output_dir \
|
17 |
--num_train_epochs="10" \
|
18 |
--adam_beta1="0.9" \
|
19 |
--adam_beta2="0.98" \
|
20 |
--adam_epsilon="1e-6" \
|
21 |
-
--logging_steps="
|
22 |
-
--save_steps="
|
23 |
-
--eval_steps="
|
24 |
--auth_token="True" \
|
25 |
--do_train \
|
26 |
--do_eval \
|
|
|
11 |
--per_device_train_batch_size="46" \
|
12 |
--per_device_eval_batch_size="46" \
|
13 |
--pad_to_max_length \
|
14 |
+
--learning_rate="0.00005" \
|
15 |
--warmup_steps="150000" \
|
16 |
--overwrite_output_dir \
|
17 |
--num_train_epochs="10" \
|
18 |
--adam_beta1="0.9" \
|
19 |
--adam_beta2="0.98" \
|
20 |
--adam_epsilon="1e-6" \
|
21 |
+
--logging_steps="10000" \
|
22 |
+
--save_steps="10000" \
|
23 |
+
--eval_steps="10000" \
|
24 |
--auth_token="True" \
|
25 |
--do_train \
|
26 |
--do_eval \
|