pere commited on
Commit
a3573d6
1 Parent(s): 4fc43a3

Saving weights and logs of step 1000

Browse files
events.out.tfevents.1643733579.t1v-n-ccbf3e94-w-0.346519.3.v2 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dc59700ff46375d6e1d26811d0ce397af3dbfac5205427c19664138ac48f0f18
3
- size 14000468
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26741bf36a1c1e477fd7a061ab76ec7f959509f89a92becf5a05dcb413028085
3
+ size 14149680
events.out.tfevents.1643868791.t1v-n-ccbf3e94-w-0.486326.3.v2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14c379e6134826763d174838a8f244103ddda25c416ec68b877062025a0d1715
3
+ size 147136
flax_model.msgpack CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7f8088dbe7a16068e02a7b6ca3fc9450d8a987ff90231dc9b2b4a3feb5526db6
3
  size 498796983
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3abfd3e2f95c9249ed19c8a67ddef8d31a2ebaa14680c5f2cdbda8dec86107c0
3
  size 498796983
run_128_recover_8e.sh ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ python run_mlm_flax.py \
2
+ --output_dir="./" \
3
+ --model_type="roberta" \
4
+ --model_name_or_path="./" \
5
+ --config_name="roberta-base" \
6
+ --tokenizer_name="NbAiLab/nb-roberta-base" \
7
+ --dataset_name="NbAiLab/NCC" \
8
+ --cache_dir="/mnt/disks/flaxdisk/cache/" \
9
+ --max_seq_length="128" \
10
+ --weight_decay="0.01" \
11
+ --per_device_train_batch_size="232" \
12
+ --per_device_eval_batch_size="232" \
13
+ --pad_to_max_length \
14
+ --learning_rate="4.505741202365243e-08" \
15
+ --warmup_steps="0" \
16
+ --overwrite_output_dir \
17
+ --num_train_epochs="2" \
18
+ --adam_beta1="0.9" \
19
+ --adam_beta2="0.98" \
20
+ --adam_epsilon="1e-6" \
21
+ --logging_steps="1000" \
22
+ --save_steps="1000" \
23
+ --eval_steps="1000" \
24
+ --auth_token="True" \
25
+ --do_train \
26
+ --do_eval \
27
+ --dtype="bfloat16" \
28
+ --push_to_hub