Saving weights and logs of step 5001
Browse files
events.out.tfevents.1628890854.t1v-n-1a0a7c50-w-0.49152.3.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 sha256:89e434da252b10ce4dfc920018496575bc6b801af99b371e768b9e187a43a5ab
|
3 |
+
size 10334199
|
events.out.tfevents.1629017411.t1v-n-1a0a7c50-w-0.25265.3.v2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7192563b41c3ca31475fe5aadf7e33543311cc6c518a7d8bbd50785a9e26d9e1
|
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 sha256:064f3047656d5a897e9e2c5899e14763ab4c38f1ade3d6b8d3729a1e9cd7c8ab
|
3 |
size 1100762015
|
recover_415_run_streaming.sh
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
./run_t5_mlm_flax_streaming.py \
|
2 |
+
--output_dir="./" \
|
3 |
+
--model_type="t5" \
|
4 |
+
--config_name="./" \
|
5 |
+
--tokenizer_name="./" \
|
6 |
+
--model_name_or_path="./" \
|
7 |
+
--dataset_name="pere/norwegian_colossal_corpus_v2_short100k" \
|
8 |
+
--max_seq_length="512" \
|
9 |
+
--weight_decay="0.01" \
|
10 |
+
--per_device_train_batch_size="32" \
|
11 |
+
--per_device_eval_batch_size="32" \
|
12 |
+
--learning_rate="4.707e-3" \
|
13 |
+
--warmup_steps="0" \
|
14 |
+
--overwrite_output_dir \
|
15 |
+
--cache_dir /mnt/disks/flaxdisk/cache/ \
|
16 |
+
--num_train_epochs="5" \
|
17 |
+
--adam_beta1="0.9" \
|
18 |
+
--adam_beta2="0.98" \
|
19 |
+
--logging_steps="500" \
|
20 |
+
--num_train_steps="585000" \
|
21 |
+
--num_eval_samples="5000" \
|
22 |
+
--save_steps="5000" \
|
23 |
+
--eval_steps="5000" \
|
24 |
+
--preprocessing_num_workers 96 \
|
25 |
+
--adafactor \
|
26 |
+
--push_to_hub
|
27 |
+
|