Saving weights and logs of step 2500

events.out.tfevents.1638648947.t1v-n-6f5efcd5-w-0.432143.0.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86aeb8ca76bbad8ab38ced408e21d39e2938ea16b7a5d042c89500aa93598fc8
-size 3790461
+oid sha256:bd63af2e907b8815764a9103c1240351d060288a7522b1b9be7a7ec544ca7979
+size 3939601

events.out.tfevents.1638691720.t1v-n-6f5efcd5-w-0.475226.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a3a20f279201ec761155203bb3f8a0f9a8ecc84bb4e2d2f8f45c0bd7a749987
+size 40

events.out.tfevents.1638692589.t1v-n-6f5efcd5-w-0.476662.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c3fdb0680e32fffb8c7d2c008d7dc84b49c68aa9928f4f5b78e28a72f9bbe18
+size 334984

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:66afd8f2ae55e516f765eccba8ce4ec0a3a4eda7454cc4828a5aa1cbe594ce77
+oid sha256:e64af8866422a771b20518e009f04e8609c67cee915a485c9a85c7b4ae29b7b5
 size 498796983

run_ncc_like_oscar_10w.sh

@@ -0,0 +1,22 @@
+./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" \
+    --preprocessing_num_workers="10" \
+    --adafactor \
+    --push_to_hub

run_ncc_like_oscar_8w.sh

@@ -0,0 +1,22 @@
+./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" \
+    --preprocessing_num_workers="8" \
+    --adafactor \
+    --push_to_hub