Remove support for use_chars_concat

use_chars_lstm always performs better, and using ELMo is even better

src/main.py

@@ -63,7 +63,6 @@ def make_hparams():
         use_tags=False,
         use_words=False,
         use_chars_lstm=False,
-        use_chars_concat=False,
         use_elmo=False,
 
         d_char_emb=32, # A larger value may be better for use_chars_lstm

src/parse_nk.py

@@ -635,33 +635,21 @@ def __init__(
         self.use_tags = hparams.use_tags
 
         self.morpho_emb_dropout = None
-        if hparams.use_chars_lstm or hparams.use_chars_concat or hparams.use_elmo:
+        if hparams.use_chars_lstm or hparams.use_elmo:
             self.morpho_emb_dropout = hparams.morpho_emb_dropout
         else:
-            assert self.emb_types, "Need at least one of: use_tags, use_words, use_chars_lstm, use_chars_concat, use_elmo"
+            assert self.emb_types, "Need at least one of: use_tags, use_words, use_chars_lstm, use_elmo"
 
         self.char_encoder = None
-        self.char_embedding = None
         self.elmo = None
         if hparams.use_chars_lstm:
-            assert not hparams.use_chars_concat, "use_chars_lstm and use_chars_concat are mutually exclusive"
             assert not hparams.use_elmo, "use_chars_lstm and use_elmo are mutually exclusive"
             self.char_encoder = CharacterLSTM(
                 num_embeddings_map['chars'],
                 hparams.d_char_emb,
                 self.d_content,
                 char_dropout=hparams.char_lstm_input_dropout,
             )
-        elif hparams.use_chars_concat:
-            assert not hparams.use_elmo, "use_chars_concat and use_elmo are mutually exclusive"
-            self.num_chars_flat = self.d_content // hparams.d_char_emb
-            assert self.num_chars_flat >= 2, "incompatible settings of d_model/partitioned and d_char_emb"
-            assert self.num_chars_flat == (self.d_content / hparams.d_char_emb), "d_char_emb does not evenly divide model size"
-
-            self.char_embedding = nn.Embedding(
-                num_embeddings_map['chars'],
-                hparams.d_char_emb,
-                )
         elif hparams.use_elmo:
             self.elmo = get_elmo_class()(
                 options_file="data/elmo_2x4096_512_2048cnn_2xhighway_options.json",
@@ -724,6 +712,8 @@ def model(self):
     def from_spec(cls, spec, model):
         spec = spec.copy()
         hparams = spec['hparams']
+        if 'use_chars_concat' in hparams and hparams['use_chars_concat']:
+            raise NotImplementedError("Support for use_chars_concat has been removed")
         if 'sentence_max_len' not in hparams:
             hparams['sentence_max_len'] = 300
         if 'use_elmo' not in hparams:
@@ -833,30 +823,7 @@ def parse_batch(self, sentences, golds=None, return_label_scores_charts=False):
             assert i == packed_len
 
             extra_content_annotations = self.char_encoder(char_idxs_encoder, word_lens_encoder, batch_idxs)
-        elif self.char_embedding is not None:
-            char_idxs_encoder = np.zeros((packed_len, self.num_chars_flat), dtype=int)
-
-            i = 0
-            for snum, sentence in enumerate(sentences):
-                for wordnum, (tag, word) in enumerate([(START, START)] + sentence + [(STOP, STOP)]):
-                    if word == START:
-                        char_idxs_encoder[i, :] = self.char_vocab.index(CHAR_START_SENTENCE)
-                    elif word == STOP:
-                        char_idxs_encoder[i, :] = self.char_vocab.index(CHAR_STOP_SENTENCE)
-                    else:
-                        word_chars = (([self.char_vocab.index(CHAR_START_WORD)] * self.num_chars_flat)
-                            + [self.char_vocab.index_or_unk(char, CHAR_UNK) for char in word]
-                            + ([self.char_vocab.index(CHAR_STOP_WORD)] * self.num_chars_flat))
-                        char_idxs_encoder[i, :self.num_chars_flat//2] = word_chars[self.num_chars_flat:self.num_chars_flat + self.num_chars_flat//2]
-                        char_idxs_encoder[i, self.num_chars_flat//2:] = word_chars[::-1][self.num_chars_flat:self.num_chars_flat + self.num_chars_flat//2]
-                    i += 1
-            assert i == packed_len
-
-            char_idxs_encoder = from_numpy(char_idxs_encoder)
-
-            extra_content_annotations = self.char_embedding(char_idxs_encoder)
-            extra_content_annotations = extra_content_annotations.view(-1, self.num_chars_flat * self.char_embedding.embedding_dim)
-        if self.elmo is not None:
+        elif self.elmo is not None:
             # See https://github.com/allenai/allennlp/blob/c3c3549887a6b1fb0bc8abf77bc820a3ab97f788/allennlp/data/token_indexers/elmo_indexer.py#L61
             # ELMO_START_SENTENCE = 256
             # ELMO_STOP_SENTENCE = 257