Saving weights and logs of step 2500

events.out.tfevents.1638794992.t1v-n-1a0a7c50-w-0.374880.0.v2 → events.out.tfevents.1638824108.t1v-n-1a0a7c50-w-0.405640.0.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2953ad96f1accba06f3677a12c8fe0c56680e3cf8ec9da0d8a8c05a16575928c
-size 3790461
+oid sha256:05f68ea6223dcbc422c09ea4d0b42e02eaef9bc78922c7428007ca98432c3132
+size 334984

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:14bf609c747e6a2c7f052ea5a27a70190ec5a3f0f91c43ff4a88931b95d02aaf
+oid sha256:f10c94c754dfef687b45c3aef74f6c2a54bb9306847d078892e0826ea5f5bd58
 size 498796983

run_ncc_like_oscar_adafactor.sh

@@ -0,0 +1,21 @@
+./run_mlm_flax.py \
+    --output_dir="./" \
+    --model_type="roberta" \
+    --config_name="./" \
+    --tokenizer_name="./" \
+    --dataset_name="NbAiLab/NCC_small" \
+    --max_seq_length="128" \
+    --weight_decay="0.01" \
+    --per_device_train_batch_size="128" \
+    --per_device_eval_batch_size="128" \
+    --learning_rate="3e-4" \
+    --warmup_steps="1000" \
+    --overwrite_output_dir \
+    --num_train_epochs="18" \
+    --adam_beta1="0.9" \
+    --adam_beta2="0.98" \
+    --logging_steps="500" \
+    --save_steps="2500" \
+    --eval_steps="2500" \
+    --adafactor \
+    --push_to_hub