Saving weights and logs of step 1000
Browse files
config.json
CHANGED
@@ -19,7 +19,7 @@
|
|
19 |
"num_hidden_layers": 12,
|
20 |
"pad_token_id": 1,
|
21 |
"position_embedding_type": "absolute",
|
22 |
-
"transformers_version": "4.
|
23 |
"type_vocab_size": 1,
|
24 |
"use_cache": true,
|
25 |
"vocab_size": 50265
|
|
|
19 |
"num_hidden_layers": 12,
|
20 |
"pad_token_id": 1,
|
21 |
"position_embedding_type": "absolute",
|
22 |
+
"transformers_version": "4.16.0.dev0",
|
23 |
"type_vocab_size": 1,
|
24 |
"use_cache": true,
|
25 |
"vocab_size": 50265
|
events.out.tfevents.1644205555.t1v-n-79f0077b-w-0.401125.0.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e2af1fe635d7f82065c5bf878b782423969184c0166560393084266ea5901947
|
3 |
+
size 40
|
events.out.tfevents.1644212923.t1v-n-79f0077b-w-0.446891.0.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8111d41e059aec555485bed7138f0db88eecc16f5892eb6e19b5165841ec58db
|
3 |
+
size 40
|
events.out.tfevents.1644215479.t1v-n-79f0077b-w-0.450647.0.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:181bc17df02b41dbf683a685073aeb91c09fed6dec3b8ce74e09577c46de1971
|
3 |
+
size 147136
|
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 sha256:1c3043126169385088e6f98e40ad48982f05b396f4317b4be26034416ef2fcdc
|
3 |
size 498796983
|
run_128_scandinavian.sh
CHANGED
@@ -1,3 +1,7 @@
|
|
|
|
|
|
|
|
|
|
1 |
python run_mlm_flax.py \
|
2 |
--output_dir="./" \
|
3 |
--model_type="roberta" \
|
@@ -8,8 +12,8 @@ python run_mlm_flax.py \
|
|
8 |
--cache_dir="/mnt/disks/flaxdisk/cache/" \
|
9 |
--max_seq_length="128" \
|
10 |
--weight_decay="0.01" \
|
11 |
-
--per_device_train_batch_size="
|
12 |
-
--per_device_eval_batch_size="
|
13 |
--pad_to_max_length \
|
14 |
--learning_rate="3e-4" \
|
15 |
--warmup_steps="10000" \
|
|
|
1 |
+
# --per_device_train_batch_size="232" \
|
2 |
+
# --per_device_eval_batch_size="232" \
|
3 |
+
|
4 |
+
|
5 |
python run_mlm_flax.py \
|
6 |
--output_dir="./" \
|
7 |
--model_type="roberta" \
|
|
|
12 |
--cache_dir="/mnt/disks/flaxdisk/cache/" \
|
13 |
--max_seq_length="128" \
|
14 |
--weight_decay="0.01" \
|
15 |
+
--per_device_train_batch_size="116" \
|
16 |
+
--per_device_eval_batch_size="116" \
|
17 |
--pad_to_max_length \
|
18 |
--learning_rate="3e-4" \
|
19 |
--warmup_steps="10000" \
|