Merge branch 'main' into prompt_relation_component

.pre-commit-config.yaml

@@ -7,23 +7,23 @@ repos:
       - id: ssort
 
   - repo: https://github.com/asottile/add-trailing-comma
-    rev: v2.4.0
+    rev: v3.0.1
     hooks:
       - id: add-trailing-comma
 
   - repo: https://github.com/PyCQA/docformatter
-    rev: v1.5.1
+    rev: v1.7.5
     hooks:
       - id: docformatter
         args: [--in-place]
 
   - repo: https://github.com/psf/black
-    rev: 23.1.0
+    rev: 23.7.0
     hooks:
       - id: black
         args: [--line-length, "88"]
 
-  - repo: https://github.com/charliermarsh/ruff-pre-commit
-    rev: v0.0.254
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.0.284
     hooks:
       - id: ruff

CHANGELOG.md

@@ -2,6 +2,26 @@
 
 <!--next-version-placeholder-->
 
+## v0.5.0 (2023-03-31)
+### Feature
+* Added gold data to package data ([`b892aaa`](https://github.com/centre-for-humanities-computing/conspiracies/commit/b892aaaa83a73e2074485ec5cd40df08d95b9c65))
+* Fixed spacy dep. to allow for the visualizer ([`375b3bf`](https://github.com/centre-for-humanities-computing/conspiracies/commit/375b3bf96fa4dd1704fe2fb079541b0cbbce05f3))
+* Added triplet data classes along with seraliazation functions ([`0325c63`](https://github.com/centre-for-humanities-computing/conspiracies/commit/0325c63c1c6c8dfb36c7ba21edda87ab575f6a04))
+
+### Fix
+* Fixed circular import ([`bfc5595`](https://github.com/centre-for-humanities-computing/conspiracies/commit/bfc5595ead67bdb7ac3bbb8c3fa585dc552efcec))
+* Prevented circular import ([`2af3559`](https://github.com/centre-for-humanities-computing/conspiracies/commit/2af3559f337e38eb8b07b6dc04d8e264df634b72))
+
+### Documentation
+* Changed req. for docs ([`dcf2a1c`](https://github.com/centre-for-humanities-computing/conspiracies/commit/dcf2a1c8e46c4687e7e7b47a5bf8b40d393e8644))
+* Updated tutorial based on reivew ([`0f29203`](https://github.com/centre-for-humanities-computing/conspiracies/commit/0f292036efa842f5f729ff31f152671ecdd8522f))
+* Updated docstring based on review comments ([`b669555`](https://github.com/centre-for-humanities-computing/conspiracies/commit/b6695557426dd47a4fb10433de586d210478c47d))
+* Updated tutorial to reflect changes ([`833ceae`](https://github.com/centre-for-humanities-computing/conspiracies/commit/833ceae49c83cbc44ecedd97a7cdd4262609c1a0))
+* Increased execution time for tutorial ([`4fee404`](https://github.com/centre-for-humanities-computing/conspiracies/commit/4fee4042d6a7704b426b6f563aeee51b0b79ae57))
+* Updated notebook ([`330fc22`](https://github.com/centre-for-humanities-computing/conspiracies/commit/330fc22efba02f05958cd51993ff404f14ba5036))
+* Updated to properly read data ([`3e3e62d`](https://github.com/centre-for-humanities-computing/conspiracies/commit/3e3e62d7ed22b44492661447cb00f7c516893e9d))
+* Added reading gold as an example ([`b6d83b6`](https://github.com/centre-for-humanities-computing/conspiracies/commit/b6d83b61119c0f644abdd7659396f552cb7ffded))
+
 ## v0.4.0 (2023-03-10)
 ### Feature
 * Add meaningful default task descriptions for templates ([#20](https://github.com/centre-for-humanities-computing/conspiracies/issues/20)) ([`17436ee`](https://github.com/centre-for-humanities-computing/conspiracies/commit/17436eefd5a0bb83f223cb8f67c6db44105ffdda))

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "conspiracies"
-version = "0.4.0"
+version = "0.5.0"
 authors = [
     {name = '"Kenneth Enevoldsen'},
     {name = "Lasse Hansen"},

src/conspiracies/relationextraction/model.py

@@ -17,8 +17,10 @@ def __init__(self, arg_layer, n_layers):
         predicates. It uses ArgExtractorLayer as a base block and repeat the
         block N('n_layers') times.
 
-        :param arg_layer: an instance of the ArgExtractorLayer() class (required)
-        :param n_layers: the number of sub-layers in the ArgModule (required).
+        :param arg_layer: an instance of the ArgExtractorLayer() class
+            (required)
+        :param n_layers: the number of sub-layers in the ArgModule
+            (required).
         """
         super(ArgModule, self).__init__()
         self.layers = _get_clones(arg_layer, n_layers)
@@ -66,11 +68,14 @@ def __init__(
         attention. (only encoder-decoder multi-head attention followed by feed-
         forward layers)
 
-        :param d_model: model dimensionality (default=768 from BERT-base)
+        :param d_model: model dimensionality (default=768 from BERT-
+            base)
         :param n_heads: number of heads in multi-head attention layer
-        :param d_feedforward: dimensionality of point-wise feed-forward layer
+        :param d_feedforward: dimensionality of point-wise feed-forward
+            layer
         :param dropout: drop rate of all layers
-        :param activation: activation function after first feed-forward layer
+        :param activation: activation function after first feed-forward
+            layer
         """
         super(ArgExtractorLayer, self).__init__()
         self.multihead_attn = nn.MultiheadAttention(d_model, n_heads, dropout=dropout)
@@ -89,7 +94,8 @@ def forward(self, target, source, key_mask=None):
 
         :param target: a tensor which takes a role as a query
         :param source: a tensor which takes a role as a key & value
-        :param key_mask: key mask tensor with the shape of (batch_size, sequence_length)
+        :param key_mask: key mask tensor with the shape of (batch_size,
+            sequence_length)
         """
         # Multi-head attention layer (+ add & norm)
         attended = self.multihead_attn(

src/conspiracies/relationextraction/other/bio.py

@@ -76,10 +76,11 @@ def filter_arg_tags(arg_tags, pred_tags, tokens):
     """Same as the description of @filter_pred_tags().
 
     :param arg_tags: argument tags with the shape of (B, L).
-    :param pred_tags: predicate tags with the same shape.
-        It is used to force predicate position to be allocated the 'Outside' tag.
-    :param tokens: list of string tokens with the length of L.
-        It is used to force special tokens like [CLS] to be allocated the 'Outside' tag.
+    :param pred_tags: predicate tags with the same shape. It is used to
+        force predicate position to be allocated the 'Outside' tag.
+    :param tokens: list of string tokens with the length of L. It is
+        used to force special tokens like [CLS] to be allocated the
+        'Outside' tag.
     :return: tensor of filtered argument tags with the same shape.
     """
     # filter by tokens ([CLS], [SEP], [PAD] tokens should be allocated as 'O')
@@ -120,9 +121,9 @@ def get_max_prob_args(arg_tags, arg_probs):
     labels.
 
     :param arg_tags: argument tags with the shape of (B, L).
-    :param arg_probs: argument softmax probabilities with the shape of (B, L, T),
-        where B is the batch size, L is the sequence length, and T is the # of tag
-            labels.
+    :param arg_probs: argument softmax probabilities with the shape of
+        (B, L, T), where B is the batch size, L is the sequence length,
+        and T is the # of tag labels.
     :return: tensor of filtered argument tags with the same shape.
     """
     for cur_arg_tag, cur_probs in zip(arg_tags, arg_probs):
@@ -292,7 +293,7 @@ def _find_begins(idxs):
 
 
 def get_confidence_score(pred_probs, arg_probs, extraction_idxs):
-    """get the confidence score of each extraction for drawing PR-curve.
+    """Get the confidence score of each extraction for drawing PR-curve.
 
     :param pred_probs: (sequence length, # of predicate labels)
     :param arg_probs: (# of predicates, sequence length, # of argument labels)

src/conspiracies/relationextraction/wrap_model_spacy.py

@@ -18,7 +18,7 @@
 
 
 class SpacyRelationExtractor(TrainablePipe):
-    """spaCy pipeline component that adds a multilingual relation-extraction
+    """SpaCy pipeline component that adds a multilingual relation-extraction
     component. The extractions are saved in the doc._.relation_triplets,
 
     ._.relation_head,