multiple fixes

finetune_large_mt5_sentencefix.gin

@@ -7,12 +7,12 @@ from t5x import models
 from t5x import partitioning
 from t5x import utils
 
-include "t5x/examples/t5/mt5/small.gin"
+include "t5x/examples/t5/mt5/large.gin"
 include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

finetune_large_mt5_sentencefix_v4.gin

@@ -7,12 +7,12 @@ from t5x import models
 from t5x import partitioning
 from t5x import utils
 
-include "t5x/examples/t5/mt5/small.gin"
+include "t5x/examples/t5/mt5/large.gin"
 include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

finetune_mt5_sentencefix.gin

@@ -12,7 +12,7 @@ include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_050  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

finetune_mt5_sentencefix_v4.gin

@@ -12,7 +12,7 @@ include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_050  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

finetune_small_mt5_sentencefix.gin

@@ -12,7 +12,7 @@ include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

finetune_small_mt5_sentencefix_v4.gin

@@ -12,7 +12,7 @@ include "t5x/configs/runs/finetune.gin"
 
 MIXTURE_OR_TASK_NAME = "sentencefix"
 TASK_FEATURE_LENGTHS = {"inputs": 256, "targets": 256}
-TRAIN_STEPS = 1_100_000  # 1000000 pre-trained steps + 20000 fine-tuning steps.
+TRAIN_STEPS = 1_200_000  # 1000000 pre-trained steps + 200000 fine-tuning steps.
 USE_CACHED_TASKS = False
 DROPOUT_RATE = 0.0
 RANDOM_SEED = 0

tasks_v4.py

@@ -59,7 +59,7 @@ seqio.TaskRegistry.add(
       sentencefix_preprocessor,
       seqio.preprocessors.tokenize_and_append_eos,
     ],
-    #metric_fns=[t5.evaluation.metrics.bleu],
+    metric_fns=[metrics.bleu],
     output_features=DEFAULT_OUTPUT_FEATURES,
 )
 

train_large_v4.sh

@@ -1,6 +1,5 @@
 PROJECT_DIR=${HOME}"/models/multi-sentencefix-mt5"
 T5X_DIR="../../t5x"  # directory where the t5x is cloned.
-TFDS_DATA_DIR="gs://nb-t5x-us-central2/corpus_multi_sentencefix_mt5"
 MODEL_DIR="gs://nb-t5x-us-central2/large_model_multi_sentencefix_mt5"
 export PYTHONPATH=${PROJECT_DIR}
 

train_small_v4.sh

@@ -0,0 +1,11 @@
+PROJECT_DIR=${HOME}"/models/multi-sentencefix-mt5"
+T5X_DIR="../../t5x"  # directory where the t5x is cloned.
+MODEL_DIR="gs://nb-t5x-us-central2/small_model_multi_sentencefix_mt5"
+export PYTHONPATH=${PROJECT_DIR}
+
+python3 ${T5X_DIR}/t5x/train.py \
+  --gin_search_paths=${PROJECT_DIR} \
+  --gin_file="finetune_small_mt5_sentencefix_v4.gin" \
+  --gin.MODEL_DIR="'${MODEL_DIR}'" \
+  --tfds_data_dir=${TFDS_DATA_DIR}
+

train_v4.sh

@@ -1,6 +1,5 @@
 PROJECT_DIR=${HOME}"/models/multi-sentencefix-mt5"
 T5X_DIR="../../t5x"  # directory where the t5x is cloned.
-TFDS_DATA_DIR="gs://nb-t5x-us-central2/corpus_multi_sentencefix_mt5"
 MODEL_DIR="gs://nb-t5x-us-central2/model_multi_sentencefix_mt5"
 export PYTHONPATH=${PROJECT_DIR}