binary

batch_finetune_eu_jav_base_exp_binary.sh

@@ -0,0 +1,11 @@
+PROJECT_DIR=${HOME}"/models/eu-jav-categorisation"
+export PYTHONPATH=${PROJECT_DIR}
+INITIAL_CHECKPOINT_PATH=\"gs://t5-data/pretrained_models/t5x/mt5_base/checkpoint_1000000\"
+TRAIN_STEPS=1005000
+
+python3 ../../t5x/t5x/train.py --gin_search_paths="./" --gin.TRAIN_STEPS=${TRAIN_STEPS} --gin_file="finetune_classification_base.gin" --gin.INITIAL_CHECKPOINT_PATH=${INITIAL_CHECKPOINT_PATH}  --gin.MIXTURE_OR_TASK_NAME=\"classify_tweets_binary\" --gin.MODEL_DIR=\"gs://eu-jav-t5x/finetuned/italian_tweets/fulltext_words_classify_tweets_binary_base_v1\" &&
+python3 ../../t5x/t5x/train.py --gin_search_paths="./" --gin.TRAIN_STEPS=${TRAIN_STEPS} --gin_file="finetune_classification_base.gin" --gin.INITIAL_CHECKPOINT_PATH=${INITIAL_CHECKPOINT_PATH}  --gin.MIXTURE_OR_TASK_NAME=\"classify_tweets_binary\" --gin.MODEL_DIR=\"gs://eu-jav-t5x/finetuned/italian_tweets/fulltext_words_classify_tweets_binary_base_v2\" &&
+python3 ../../t5x/t5x/train.py --gin_search_paths="./" --gin.TRAIN_STEPS=${TRAIN_STEPS} --gin_file="finetune_classification_base.gin" --gin.INITIAL_CHECKPOINT_PATH=${INITIAL_CHECKPOINT_PATH}  --gin.MIXTURE_OR_TASK_NAME=\"classify_tweets_binary\" --gin.MODEL_DIR=\"gs://eu-jav-t5x/finetuned/italian_tweets/fulltext_words_classify_tweets_binary_base_v3\" &&
+python3 ../../t5x/t5x/train.py --gin_search_paths="./" --gin.TRAIN_STEPS=${TRAIN_STEPS} --gin_file="finetune_classification_base.gin" --gin.INITIAL_CHECKPOINT_PATH=${INITIAL_CHECKPOINT_PATH}  --gin.MIXTURE_OR_TASK_NAME=\"classify_tweets_binary\" --gin.MODEL_DIR=\"gs://eu-jav-t5x/finetuned/italian_tweets/fulltext_words_classify_tweets_binary_base_v4\" &&
+python3 ../../t5x/t5x/train.py --gin_search_paths="./" --gin.TRAIN_STEPS=${TRAIN_STEPS} --gin_file="finetune_classification_base.gin" --gin.INITIAL_CHECKPOINT_PATH=${INITIAL_CHECKPOINT_PATH}  --gin.MIXTURE_OR_TASK_NAME=\"classify_tweets_binary\" --gin.MODEL_DIR=\"gs://eu-jav-t5x/finetuned/italian_tweets/fulltext_words_classify_tweets_binary_base_v5\"
+

tasks.py

@@ -106,6 +106,37 @@ def categorise_fulltext_word_preprocessor(ds):
                 num_parallel_calls=tf.data.experimental.AUTOTUNE)
 
 
+
+def categorise_binary_preprocessor(ds):
+  def normalize_text(text):
+    """Lowercase and remove quotes from a TensorFlow string."""
+    text = tf.strings.regex_replace(text,"'(.*)'", r"\1")
+    return text
+
+  def fulltext(t):
+    if t=="0":
+        t="1"
+    elif t=="1":
+        t="1"
+    elif t=="2":
+        t="2"
+    return t
+
+  def to_inputs_and_targets(ex):
+    """Map {"source": ..., "source": ...}->{"target": ..., "target": ...}."""
+    return {
+        "inputs":
+             tf.strings.join(
+                 [normalize_text(ex["source"])]),
+        "targets": 
+	    tf.strings.join(
+                 [fulltext(normalize_text(ex["target"]))]),
+    }
+  return ds.map(to_inputs_and_targets, 
+                num_parallel_calls=tf.data.experimental.AUTOTUNE)
+
+
+
 seqio.TaskRegistry.add(
     "classify_tweets",
     source=seqio.TextLineDataSource(
@@ -140,6 +171,23 @@ seqio.TaskRegistry.add(
     output_features=DEFAULT_OUTPUT_FEATURES,
 )
 
+seqio.TaskRegistry.add(
+    "classify_tweets_binary",
+    source=seqio.TextLineDataSource(
+        split_to_filepattern=tsv_path,
+        #num_input_examples=num_nq_examples
+        ),
+    preprocessors=[
+      functools.partial(
+          t5.data.preprocessors.parse_tsv,
+          field_names=["annotator1","annotator2","annotator3","target","source","id"]),
+      categorise_fulltext_word_preprocessor,
+      seqio.preprocessors.tokenize_and_append_eos,
+    ],
+    metric_fns=[metrics.accuracy,my_metrics.f1_macro],
+    output_features=DEFAULT_OUTPUT_FEATURES,
+)
+
 seqio.TaskRegistry.add(
     "classify_tweets_fulltext_word",
     source=seqio.TextLineDataSource(