Saving weights and logs of step 10000
Browse files
events.out.tfevents.1640956682.t1v-n-358ff5d1-w-0.11974.3.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8b0fe1349611f01cd08d8a11929610ed03a8c20073c1fa070a7b7843955235bf
|
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:cc49f760bccb82672d93ae5c122aed1f410add84860109051c07d287d42a5517
|
3 |
size 498796983
|
run_step1.sh
CHANGED
@@ -7,12 +7,12 @@
|
|
7 |
--cache_dir="/mnt/disks/flaxdisk/cache/" \
|
8 |
--max_seq_length="128" \
|
9 |
--weight_decay="0.01" \
|
10 |
-
--per_device_train_batch_size="
|
11 |
-
--per_device_eval_batch_size="
|
12 |
-
--learning_rate="
|
13 |
--warmup_steps="10000" \
|
14 |
--overwrite_output_dir \
|
15 |
-
--num_train_epochs="
|
16 |
--adam_beta1="0.9" \
|
17 |
--adam_beta2="0.98" \
|
18 |
--adam_epsilon="1e-6" \
|
|
|
7 |
--cache_dir="/mnt/disks/flaxdisk/cache/" \
|
8 |
--max_seq_length="128" \
|
9 |
--weight_decay="0.01" \
|
10 |
+
--per_device_train_batch_size="200" \
|
11 |
+
--per_device_eval_batch_size="200" \
|
12 |
+
--learning_rate="4e-4" \
|
13 |
--warmup_steps="10000" \
|
14 |
--overwrite_output_dir \
|
15 |
+
--num_train_epochs="10" \
|
16 |
--adam_beta1="0.9" \
|
17 |
--adam_beta2="0.98" \
|
18 |
--adam_epsilon="1e-6" \
|