Add batch_size flag to lm_salience_demo and fix minor bugs.

PiperOrigin-RevId: 623274231

Author: bdu91

lit_nlp/examples/lm_salience_demo.py

@@ -8,7 +8,7 @@
 To run with the default configuration (Gemma on TensorFlow via Keras):
 
   blaze run -c opt examples:lm_salience_demo -- \
-    --models=gemma_instruct_2b_en:gemma_instruct_2b_en \
+    --models=gemma_1.1_instruct_2b_en:gemma_1.1_instruct_2b_en \
     --port=8890 --alsologtostderr
 
 MODELS:
@@ -64,7 +64,7 @@
 
 _MODELS = flags.DEFINE_list(
     "models",
-    ["gemma_instruct_2b_en:gemma_instruct_2b_en"],
+    ["gemma_1.1_instruct_2b_en:gemma_1.1_instruct_2b_en"],
     "Models to load, as <name>:<path>. Path can be a URL, a local file path, or"
     " the name of a preset for the configured Deep Learning framework (either"
     " KerasNLP or HuggingFace Transformers; see --dl_framework for more). This"
@@ -91,6 +91,10 @@
     ),
 )
 
+_BATCH_SIZE = flags.DEFINE_integer(
+    "batch_size", 4, "The number of examples to process per batch.",
+)
+
 _DL_BACKEND = flags.DEFINE_enum(
     "dl_backend",
     "tensorflow",
@@ -278,18 +282,17 @@ def main(argv: Sequence[str]) -> Optional[dev_server.LitServerType]:
     path = file_cache.cached_path(
         path,
         extract_compressed_file=path.endswith(".tar.gz"),
-        copy_directories=True,
     )
 
-    if _DL_FRAMEWORK.value == "keras":
+    if _DL_FRAMEWORK.value == "kerasnlp":
       # pylint: disable=g-import-not-at-top
       from keras_nlp import models as keras_models
       from lit_nlp.examples.models import instrumented_keras_lms as lit_keras
       # pylint: enable=g-import-not-at-top
       # Load the weights once for the underlying Keras model.
       model = keras_models.CausalLM.from_preset(path)
       models |= lit_keras.initialize_model_group_for_salience(
-          model_name, model, max_length=512, batch_size=4
+          model_name, model, max_length=512, batch_size=_BATCH_SIZE.value
       )
       # Disable embeddings from the generation model.
       # TODO(lit-dev): re-enable embeddings if we can figure out why UMAP was
@@ -301,7 +304,11 @@ def main(argv: Sequence[str]) -> Optional[dev_server.LitServerType]:
       # Assuming a valid decoder model name supported by
       # `transformers.AutoModelForCausalLM` is provided to "path".
       models |= pretrained_lms.initialize_model_group_for_salience(
-          model_name, path, framework=_DL_BACKEND.value, max_new_tokens=512
+          model_name,
+          path,
+          batch_size=_BATCH_SIZE.value,
+          framework=_DL_BACKEND.value,
+          max_new_tokens=512,
       )
 
   for name in datasets: