Saving weights and logs of step 2500

events.out.tfevents.1638828930.t1v-n-d4d6e0cd-w-0.721231.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f6dd60bd2a9f5218802933a8ad1e193b0fcfd889464d3d06d9084ad2dbc8ec9
+size 367912

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fff7bafec4ef805f528b34450ef3a3408cc179716b0089b610c4d29447b474ea
+oid sha256:ceac157b5b4dd77a8a109aa0a7acd9f4b3858da6fe44496eb76bba963c6248a9
 size 498796983

run_oscar_adafactor.sh

@@ -0,0 +1,22 @@
+./run_mlm_flax.py \
+    --output_dir="./" \
+    --model_type="roberta" \
+    --config_name="./" \
+    --tokenizer_name="./" \
+    --dataset_name="oscar" \
+    --dataset_config_name="unshuffled_deduplicated_no" \
+    --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