change run.sh
Browse files- src/run.sh +3 -3
src/run.sh
CHANGED
@@ -28,14 +28,14 @@ export LOGGING_STEPS=500
|
|
28 |
export EVAL_STEPS=2500
|
29 |
export SAVE_STEPS=2500
|
30 |
|
31 |
-
python src/
|
32 |
--output_dir="$OUTPUT_DIR" \
|
33 |
--model_type="$MODEL_TYPE" \
|
34 |
--config_name="$CONFIG_NAME" \
|
35 |
--tokenizer_name="$TOKENIZER_NAME" \
|
36 |
--dataset_name="$DATASET_NAME" \
|
37 |
--dataset_config_name="$DATASET_CONFIG_NAME" \
|
38 |
-
--
|
39 |
--per_device_train_batch_size=$PER_DEVICE_TRAIN_BATCH_SIZE \
|
40 |
--per_device_eval_batch_size=$PER_DEVICE_EVAL_BATCH_SIZE \
|
41 |
--num_train_epochs=$NUM_TRAIN_EPOCHS \
|
@@ -47,4 +47,4 @@ python src/run_clm.py \
|
|
47 |
--do_train \
|
48 |
--do_eval \
|
49 |
--overwrite_output_dir \
|
50 |
-
--push_to_hub
|
|
|
28 |
export EVAL_STEPS=2500
|
29 |
export SAVE_STEPS=2500
|
30 |
|
31 |
+
python src/run_clm_flax.py \
|
32 |
--output_dir="$OUTPUT_DIR" \
|
33 |
--model_type="$MODEL_TYPE" \
|
34 |
--config_name="$CONFIG_NAME" \
|
35 |
--tokenizer_name="$TOKENIZER_NAME" \
|
36 |
--dataset_name="$DATASET_NAME" \
|
37 |
--dataset_config_name="$DATASET_CONFIG_NAME" \
|
38 |
+
--block_size=$MAX_SEQUENCE_LENGTH \
|
39 |
--per_device_train_batch_size=$PER_DEVICE_TRAIN_BATCH_SIZE \
|
40 |
--per_device_eval_batch_size=$PER_DEVICE_EVAL_BATCH_SIZE \
|
41 |
--num_train_epochs=$NUM_TRAIN_EPOCHS \
|
|
|
47 |
--do_train \
|
48 |
--do_eval \
|
49 |
--overwrite_output_dir \
|
50 |
+
--push_to_hub
|