From 7cc3f2c2943d51ffe464fc63608a226ed7831c30 Mon Sep 17 00:00:00 2001
From: Maja Jablonska <majajjablonska@gmail.com>
Date: Mon, 16 Oct 2023 15:51:00 +1100
Subject: [PATCH] Add prediction to main.py
---
combo/data/vocabulary.py | 2 +-
combo/main.py | 20 +++-----------------
2 files changed, 4 insertions(+), 18 deletions(-)
diff --git a/combo/data/vocabulary.py b/combo/data/vocabulary.py
index 3cf5fed..1c0a384 100644
--- a/combo/data/vocabulary.py
+++ b/combo/data/vocabulary.py
@@ -300,7 +300,6 @@ class Vocabulary(FromParameters):
filename = os.path.join(directory, namespace_filename)
vocab.set_from_file(filename, is_padded, namespace=namespace, oov_token=oov_token)
- get_slices_if_not_provided(vocab)
vocab.constructed_from = 'from_files'
return vocab
@@ -758,6 +757,7 @@ class Vocabulary(FromParameters):
def get_slices_if_not_provided(vocab: Vocabulary):
if hasattr(vocab, "slices"):
return vocab.slices
+ print("Getting slices...")
if "feats_labels" in vocab.get_namespaces():
idx2token = vocab.get_index_to_token_vocabulary("feats_labels")
diff --git a/combo/main.py b/combo/main.py
index a426b10..aeb1dc8 100755
--- a/combo/main.py
+++ b/combo/main.py
@@ -92,25 +92,11 @@ flags.DEFINE_enum(name="predictor_name", default="combo-lambo",
help="Use predictor with whitespace, spacy or lambo (recommended) tokenizer.")
-def get_saved_model(parameters) -> ComboModel:
- return ComboModel.load(os.path.join(FLAGS.model_path),
- config=parameters,
- weights_file=os.path.join(FLAGS.model_path, 'best.th'),
- cuda_device=FLAGS.cuda_device)
-
-
def get_predictor() -> COMBO:
- # Check for GPU
- # allen_checks.check_for_gpu(FLAGS.cuda_device)
checks.file_exists(FLAGS.model_path)
- with open(os.path.join(FLAGS.model_path, 'params.json'), 'r') as f:
- serialized = json.load(f)
- model = get_saved_model(serialized)
- if 'dataset_reader' in serialized:
- dataset_reader = resolve(serialized['dataset_reader'])
- else:
- dataset_reader = default_ud_dataset_reader()
- return COMBO(model, dataset_reader)
+ arch = load_archive(FLAGS.model_path)
+ dataset_reader = default_ud_dataset_reader()
+ return COMBO(arch.model, dataset_reader)
def run(_):
--
GitLab