111001

continue_run_mlm_flax_stream.sh

@@ -1,22 +1,21 @@
-export MODEL_DIR=/home/Z6HJB/roberta-base-danish/roberta-base-danish/
-
+export MODEL_DIR=/home/Z6HJB/train-roberta-base-danish/roberta-base-danish/
 source /home/Z6HJB/test/bin/activate
 
-python3 ./src/run_mlm_flax_stream.py \
+python3 ./src/danish_run_mlm_flax_stream.py \
     --model_name_or_path="${MODEL_DIR}" \
     --output_dir="${MODEL_DIR}" \
     --tokenizer_name="${MODEL_DIR}" \
     --dataset_name="mc4" \
     --dataset_config_name="unshuffled_deduplicated_en" \
-    --max_seq_length="514" \
-    --per_device_train_batch_size="32" \
-    --per_device_eval_batch_size="32" \
-    --learning_rate="3e-4" \
+    --max_seq_length="128" \
+    --per_device_train_batch_size="128" \
+    --per_device_eval_batch_size="128" \
+    --learning_rate="1.087e-6" \
     --warmup_steps="1000" \
     --overwrite_output_dir \
     --adam_beta1="0.9" \
     --adam_beta2="0.98" \
-    --num_train_steps="100000" \
+    --num_train_steps="200000" \
     --num_eval_samples="5000" \
     --save_steps="1000" \
     --logging_steps="250" \

events.out.tfevents.1625999454.t1v-n-ba9840ed-w-0.129483.3.v2

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c57d5be8414b153a10ee9b8ca9d21bb2232fb9129ac5d521e5f5c94af5f9ae1f
-size 10389142

events.out.tfevents.1625976092.t1v-n-ba9840ed-w-0.105113.3.v2 → events.out.tfevents.1626175938.t1v-n-ba9840ed-w-0.311007.3.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:24ed25853743a37a86892767785e9a9a30f82ef00ddb5250a1c3f7aa00fe9e0b
-size 4412262
+oid sha256:e3b0ba1460852ed0ad9c3a92cadbf858e7229561676e976bbbdbe4bcb10a7f49
+size 17561398

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c221eb636b60017750c5a40abee0d1578ad3425f7b8460a9d80ee59e0ae13b8b
+oid sha256:e5ff54ea0454079142eb2dd56d289cb9aa37a29d9b631821342518bb4cf9a292
 size 498796983