maltehb commited on
Commit
fd91202
1 Parent(s): 92e9efd
continue_run_mlm_flax_stream.sh CHANGED
@@ -1,22 +1,21 @@
1
- export MODEL_DIR=/home/Z6HJB/roberta-base-danish/roberta-base-danish/
2
-
3
  source /home/Z6HJB/test/bin/activate
4
 
5
- python3 ./src/run_mlm_flax_stream.py \
6
  --model_name_or_path="${MODEL_DIR}" \
7
  --output_dir="${MODEL_DIR}" \
8
  --tokenizer_name="${MODEL_DIR}" \
9
  --dataset_name="mc4" \
10
  --dataset_config_name="unshuffled_deduplicated_en" \
11
- --max_seq_length="514" \
12
- --per_device_train_batch_size="32" \
13
- --per_device_eval_batch_size="32" \
14
- --learning_rate="3e-4" \
15
  --warmup_steps="1000" \
16
  --overwrite_output_dir \
17
  --adam_beta1="0.9" \
18
  --adam_beta2="0.98" \
19
- --num_train_steps="100000" \
20
  --num_eval_samples="5000" \
21
  --save_steps="1000" \
22
  --logging_steps="250" \
1
+ export MODEL_DIR=/home/Z6HJB/train-roberta-base-danish/roberta-base-danish/
 
2
  source /home/Z6HJB/test/bin/activate
3
 
4
+ python3 ./src/danish_run_mlm_flax_stream.py \
5
  --model_name_or_path="${MODEL_DIR}" \
6
  --output_dir="${MODEL_DIR}" \
7
  --tokenizer_name="${MODEL_DIR}" \
8
  --dataset_name="mc4" \
9
  --dataset_config_name="unshuffled_deduplicated_en" \
10
+ --max_seq_length="128" \
11
+ --per_device_train_batch_size="128" \
12
+ --per_device_eval_batch_size="128" \
13
+ --learning_rate="1.087e-6" \
14
  --warmup_steps="1000" \
15
  --overwrite_output_dir \
16
  --adam_beta1="0.9" \
17
  --adam_beta2="0.98" \
18
+ --num_train_steps="200000" \
19
  --num_eval_samples="5000" \
20
  --save_steps="1000" \
21
  --logging_steps="250" \
events.out.tfevents.1625999454.t1v-n-ba9840ed-w-0.129483.3.v2 DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c57d5be8414b153a10ee9b8ca9d21bb2232fb9129ac5d521e5f5c94af5f9ae1f
3
- 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 RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:24ed25853743a37a86892767785e9a9a30f82ef00ddb5250a1c3f7aa00fe9e0b
3
- size 4412262
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e3b0ba1460852ed0ad9c3a92cadbf858e7229561676e976bbbdbe4bcb10a7f49
3
+ size 17561398
flax_model.msgpack CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c221eb636b60017750c5a40abee0d1578ad3425f7b8460a9d80ee59e0ae13b8b
3
  size 498796983
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5ff54ea0454079142eb2dd56d289cb9aa37a29d9b631821342518bb4cf9a292
3
  size 498796983