Saving weights and logs of step 5001
Browse files
events.out.tfevents.1627322413.t1v-n-358ff5d1-w-0.279972.3.v2
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 sha256:6ae095ce195c7b2c0098000b3fb8dd4987fb0b185b434ac9c30a660382fdc005
|
3 |
+
size 3622325
|
events.out.tfevents.1627408712.t1v-n-358ff5d1-w-0.356869.3.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ac4f12c4cb39a097afa86e70d7c9722a7e4da9c1da90a62c3c2c93e7e2aabae6
|
3 |
+
size 735757
|
flax_model.msgpack
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1100762015
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3e11efa6140818e1689b29e00c7c52aa5db87ea7638cf2ac4c245bc49951ebba
|
3 |
size 1100762015
|
run_streaming.sh
CHANGED
@@ -9,17 +9,17 @@
|
|
9 |
--per_device_train_batch_size="32" \
|
10 |
--per_device_eval_batch_size="32" \
|
11 |
--learning_rate="8e-3" \
|
12 |
-
--warmup_steps="
|
13 |
--overwrite_output_dir \
|
14 |
--cache_dir /mnt/disks/flaxdisk/cache/ \
|
15 |
--num_train_epochs="5" \
|
16 |
--adam_beta1="0.9" \
|
17 |
--adam_beta2="0.98" \
|
18 |
--logging_steps="500" \
|
19 |
-
--num_train_steps="
|
20 |
--num_eval_samples="5000" \
|
21 |
-
--save_steps="
|
22 |
-
--eval_steps="
|
23 |
--preprocessing_num_workers 96 \
|
24 |
--adafactor \
|
25 |
--push_to_hub
|
|
|
9 |
--per_device_train_batch_size="32" \
|
10 |
--per_device_eval_batch_size="32" \
|
11 |
--learning_rate="8e-3" \
|
12 |
+
--warmup_steps="5000" \
|
13 |
--overwrite_output_dir \
|
14 |
--cache_dir /mnt/disks/flaxdisk/cache/ \
|
15 |
--num_train_epochs="5" \
|
16 |
--adam_beta1="0.9" \
|
17 |
--adam_beta2="0.98" \
|
18 |
--logging_steps="500" \
|
19 |
+
--num_train_steps="500000" \
|
20 |
--num_eval_samples="5000" \
|
21 |
+
--save_steps="5000" \
|
22 |
+
--eval_steps="5000" \
|
23 |
--preprocessing_num_workers 96 \
|
24 |
--adafactor \
|
25 |
--push_to_hub
|