nkululeko 0.95.9__py3-none-any.whl → 0.96.1__py3-none-any.whl

nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.95.9"
+VERSION="0.96.1"
 SAMPLING_RATE = 16000

nkululeko/feat_extract/feats_bert.py

@@ -0,0 +1,105 @@
+import os
+
+import pandas as pd
+from tqdm import tqdm
+import transformers
+import torch 
+from transformers import BertTokenizer, BertModel
+
+from nkululeko.feat_extract.featureset import Featureset
+import nkululeko.glob_conf as glob_conf
+
+
+class Bert(Featureset):
+    """Class to extract bert embeddings"""
+
+    def __init__(self, name, data_df, feat_type):
+        """Constructor.
+
+        If_train is needed to distinguish from test/dev sets,
+        because they use the codebook from the training
+        """
+        super().__init__(name, data_df, feat_type)
+        cuda = "cuda" if torch.cuda.is_available() else "cpu"
+        self.device = self.util.config_val("MODEL", "device", cuda)
+        self.model_initialized = False
+        if feat_type == "bert":
+            self.feat_type = "bert-base-uncased"
+        else:
+            self.feat_type = feat_type
+
+    def init_model(self):
+        # load model
+        self.util.debug(f"loading {self.feat_type} model...")
+        model_path = self.util.config_val(
+            "FEATS", "bert.model", f"google-bert/{self.feat_type}"
+        )
+        config = transformers.AutoConfig.from_pretrained(model_path)
+        layer_num = config.num_hidden_layers
+        hidden_layer = int(self.util.config_val("FEATS", "bert.layer", "0"))
+        config.num_hidden_layers = layer_num - hidden_layer
+        self.util.debug(f"using hidden layer #{config.num_hidden_layers}")
+
+        self.tokenizer = BertTokenizer.from_pretrained(model_path)
+        self.model = BertModel.from_pretrained(model_path, config=config).to(
+            self.device
+        )
+        print(f"initialized {self.feat_type} model on {self.device}")
+        self.model.eval()
+        self.model_initialized = True
+
+    def extract(self):
+        """Extract the features or load them from disk if present."""
+        store = self.util.get_path("store")
+        storage = os.path.join(store, f"{self.name}.pkl")
+        extract = self.util.config_val("FEATS", "needs_feature_extraction", False)
+        no_reuse = eval(self.util.config_val("FEATS", "no_reuse", "False"))
+        if extract or no_reuse or not os.path.isfile(storage):
+            if not self.model_initialized:
+                self.init_model()
+            self.util.debug(
+                f"extracting {self.feat_type} embeddings, this might take a while..."
+            )
+            emb_series = pd.Series(index=self.data_df.index, dtype=object)
+            for idx, row in tqdm(self.data_df.iterrows(), total=len(self.data_df)):
+                file = idx[0]
+                text = row['text']
+                emb = self.get_embeddings(text, file)
+                emb_series[idx] = emb
+            # print(f"emb_series shape: {emb_series.shape}")
+            self.df = pd.DataFrame(emb_series.values.tolist(), index=self.data_df.index)
+            # print(f"df shape: {self.df.shape}")
+            self.df.to_pickle(storage)
+            try:
+                glob_conf.config["DATA"]["needs_feature_extraction"] = "false"
+            except KeyError:
+                pass
+        else:
+            self.util.debug(f"reusing extracted {self.feat_type} embeddings")
+            self.df = pd.read_pickle(storage)
+            if self.df.isnull().values.any():
+                self.util.error(
+                    f"got nan: {self.df.shape} {self.df.isnull().sum().sum()}"
+                )
+
+    def get_embeddings(self, text, file):
+        r"""Extract embeddings from raw audio signal."""
+        try:
+            with torch.no_grad():
+                inputs = self.tokenizer(text, return_tensors="pt")
+                outputs = self.model(**inputs)
+                # mean pooling 
+                y = torch.mean(outputs[0], dim=1)
+                y = y.ravel()
+        except RuntimeError as re:
+            print(str(re))
+            self.util.error(f"couldn't extract file: {file}")
+            y = None
+        if y is None:
+            return None
+        return y.detach().cpu().numpy()
+
+    def extract_sample(self, text):
+        self.init_model()
+        feats = self.get_embeddings(text, "no file")
+        return feats

nkululeko/feature_extractor.py

@@ -80,7 +80,7 @@ class FeatureExtractor:
             return MLD_set
 
         elif feats_type.startswith(
-            ("wav2vec2", "hubert", "wavlm", "spkrec", "whisper", "ast", "emotion2vec")
+            ("bert", "wav2vec2", "hubert", "wavlm", "spkrec", "whisper", "ast", "emotion2vec")
         ):
             return self._get_feat_extractor_by_prefix(feats_type)
 

nkululeko/reporting/reporter.py

@@ -198,7 +198,9 @@ class Reporter:
             )
 
     def plot_proba_conf(self):
-        uncertainty_threshold = self.util.config_val("PLOT", "uncertainty_threshold", False)
+        uncertainty_threshold = self.util.config_val(
+            "PLOT", "uncertainty_threshold", False
+        )
         if uncertainty_threshold:
             uncertainty_threshold = float(uncertainty_threshold)
             old_size = self.probas.shape[0]
@@ -210,9 +212,13 @@ class Reporter:
             )
             truths = df["truth"].values
             preds = df["predicted"].values
-            self._plot_confmat(truths, preds, f"uncertainty_less_than_{uncertainty_threshold}_cnf",
-                            epoch=None, test_result=None)
-
+            self._plot_confmat(
+                truths,
+                preds,
+                f"uncertainty_less_than_{uncertainty_threshold}_cnf",
+                epoch=None,
+                test_result=None,
+            )
 
     def set_id(self, run, epoch):
         """Make the report identifiable with run and epoch index."""
@@ -434,7 +440,10 @@ class Reporter:
             self.util.debug(f"####->{file_name}<-####")
             file_name = f"{res_dir}{file_name}{self.filenameadd}.txt"
         if self.util.exp_is_classification():
-            labels = glob_conf.labels
+            if glob_conf.label_encoder is not None:
+                labels = glob_conf.label_encoder.classes_
+            else:
+                labels = glob_conf.labels
             try:
                 rpt = classification_report(
                     self.truths,
@@ -451,9 +460,7 @@ class Reporter:
                     target_names=s_labels,
                     digits=4,
                 )
-                self.util.debug(
-                    f"\n {class_report_str}"
-                )
+                self.util.debug(f"\n {class_report_str}")
             except ValueError as e:
                 self.util.debug(
                     "Reporter: caught a ValueError when trying to get"

{nkululeko-0.95.9.dist-info → nkululeko-0.96.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.95.9
+Version: 0.96.1
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.95.9.dist-info → nkululeko-0.96.1.dist-info}/RECORD

@@ -4,7 +4,7 @@ nkululeko/aug_train.py,sha256=wpiHCJ7zsW38kumg3ypwXZe2HQrhUblAnv7P2QeJnAc,3525
 nkululeko/augment.py,sha256=3RzaxB3gRxovgJVjHXi0glprW01J7RaHhUkqotW2T3U,2955
 nkululeko/balance.py,sha256=r7opXbrqAipm2euPPaOmLlA5J10p2bHQgO5kWk2x9ro,8702
 nkululeko/cacheddataset.py,sha256=XFpWZmbJRg0pvhnIgYf0TkclxllD-Fctu-Ol0PF_00c,969
-nkululeko/constants.py,sha256=t_C_hQqVC1idXJB6HHr1m7ZtCYC5JVvqhYrVLRhzwIw,39
+nkululeko/constants.py,sha256=SGFx3gfUvHqg9Qr8X6a1NxL8ovndS4SAEy6Seh65maE,39
 nkululeko/demo-ft.py,sha256=iD9Pzp9QjyAv31q1cDZ75vPez7Ve8A4Cfukv5yfZdrQ,770
 nkululeko/demo.py,sha256=tu7Al2l5MCLVegkDC-NE2wcuc_YE7NRbgOlPW3yhGEs,4940
 nkululeko/demo_feats.py,sha256=BvZjeNFTlERIRlq34OHM4Z96jdDQAhB01BGQAUcX9dM,2026
@@ -13,7 +13,7 @@ nkululeko/ensemble.py,sha256=71V-rre61H3J4sh7lu-OTo4I2_g7mm_rQxwW1ARDHgY,12782
 nkululeko/experiment.py,sha256=TG9G9kSETT_R8d92aRKMMsb0HRGyM_GBFHBsU9A6ppw,38633
 nkululeko/explore.py,sha256=PjNcLuPdvWqCqYXUvGhd0hBijIhzdyi3ED1RF6o5Gjk,4212
 nkululeko/export.py,sha256=U-V4acxtuL6qKt6oAsVcM5TTeWogYUJ3GU-lA6rq6d4,4336
-nkululeko/feature_extractor.py,sha256=CsKmBoxwNClRGu20ox_eCxMG4u_1OH8Y83FYw7GfUwA,4230
+nkululeko/feature_extractor.py,sha256=d3G42OOh315Aho-yLaFT739U0UI8otiB1I4ZksK8kfg,4238
 nkululeko/file_checker.py,sha256=xJY0Q6w47pnmgJVK5rcAKPYBrCpV7eBT4_3YBzTx-H8,3454
 nkululeko/filter_data.py,sha256=4sGrKvMZ_hLnJPrHm_CqjDPKIRV8REWoT7nfSYGXbwo,7305
 nkululeko/fixedsegment.py,sha256=Tb92QiuiyMsOO3WRWwuGjZGibS8hbHHCrcWAXGk7g04,2868
@@ -69,6 +69,7 @@ nkululeko/feat_extract/feats_analyser.py,sha256=lodim7qQ8M7c3iMeJ5bHQ-nCy9Cehx1X
 nkululeko/feat_extract/feats_ast.py,sha256=w62xEoLiFtU-rj6SXkqXAktmoFaXcAcAWpUyEjp8JWo,4652
 nkululeko/feat_extract/feats_auddim.py,sha256=CGLp_aYhudfwoU5522vjrvjPxfZcyw593A8xLjYefV8,3134
 nkululeko/feat_extract/feats_audmodel.py,sha256=OsZyB1rdcG0Fai2gAwBlbuubmWor1_-P4IDkZLqgPKE,3161
+nkululeko/feat_extract/feats_bert.py,sha256=KgWLYLA11e86ubY1KtUk74QGOrZaiUongn-2LKWyf1M,4114
 nkululeko/feat_extract/feats_clap.py,sha256=1tttpfm2SJmQgYm2u8eUVpDiDOpWdKqFChpY3ZZokNs,3395
 nkululeko/feat_extract/feats_emotion2vec.py,sha256=LnV8xEg7L7HIDqz0ulqUNoaAHBU0d5gyQPb2_32T_18,9694
 nkululeko/feat_extract/feats_hubert.py,sha256=F3vrPCkx8EimJjFWYCZ7Yg9uo1G3NjYt4UKrGIUev8k,5172
@@ -122,7 +123,7 @@ nkululeko/reporting/defines.py,sha256=0vh-Tlx4fAPpk1o6mP_4x3EkIoqzYMr38IZnj-JM5z
 nkululeko/reporting/latex_writer.py,sha256=NGwSIfd4nfslDkNUOSZSdqY_VDLA8634thyhe-vj1bY,1824
 nkululeko/reporting/report.py,sha256=B5eoIKMz46VKDBsi7M9u_iegzAD-E3eGCmolzSFjZ3c,1118
 nkululeko/reporting/report_item.py,sha256=drkknsyFhGviaPJNmPQtCXJmRhTSSfjNcJt0Bls6JAA,533
-nkululeko/reporting/reporter.py,sha256=ITxM5O9Hoe_1z_59g-GF4b9vciR4shokZxeFzCrDaag,21869
+nkululeko/reporting/reporter.py,sha256=usmc2GsqGua40p8AbV08oAcZdvoelP72vsG1xS1NzOQ,22051
 nkululeko/reporting/result.py,sha256=G63a2tHCwHhM6NBJgYzsWKWJm4Yu3r4hsCHA2Km7eHU,1073
 nkululeko/segmenting/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nkululeko/segmenting/seg_inaspeechsegmenter.py,sha256=b3t0zdpJYofKWMyKRMtMMX91xeR-k8d5pbnNaQHcsOE,1902
@@ -136,9 +137,9 @@ nkululeko/utils/files.py,sha256=SrrYaU7AB80MZHiV1jcB0h_zigvYLYgSVNTXV4ao38g,4593
 nkululeko/utils/stats.py,sha256=3Fyx8q8BSKYmiufT6OkRug9RATWmGrr9BaX_y8jziWo,3074
 nkululeko/utils/unzip.py,sha256=G68f5120TjwACZC3bQcneMniddnwubPbBdMc2L5KBOo,1206
 nkululeko/utils/util.py,sha256=s7Hd7Ju1r3_WCw8gLD9YK4O6k3S_WhFcN2-XZBSctSM,18705
-nkululeko-0.95.9.dist-info/licenses/LICENSE,sha256=0zGP5B_W35yAcGfHPS18Q2B8UhvLRY3dQq1MhpsJU_U,1076
-nkululeko-0.95.9.dist-info/METADATA,sha256=WhITXnJHYD5GhyATjEb7kJhmMecWRu-BeMBw7pSWNdc,21998
-nkululeko-0.95.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nkululeko-0.95.9.dist-info/entry_points.txt,sha256=lNTkFEdh6Kjo5o95ZAWf_0Lq-4ztGoAoMVSDuPtuyS0,442
-nkululeko-0.95.9.dist-info/top_level.txt,sha256=bf1k1YKkqcXemNX_cUgoyKqQ3_GVErPqAY-53J36jkM,19
-nkululeko-0.95.9.dist-info/RECORD,,
+nkululeko-0.96.1.dist-info/licenses/LICENSE,sha256=0zGP5B_W35yAcGfHPS18Q2B8UhvLRY3dQq1MhpsJU_U,1076
+nkululeko-0.96.1.dist-info/METADATA,sha256=y6Yr1jPBXvOEyS8VZiFa_as56oS8_h0Z-fYAkoBdky8,21998
+nkululeko-0.96.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nkululeko-0.96.1.dist-info/entry_points.txt,sha256=lNTkFEdh6Kjo5o95ZAWf_0Lq-4ztGoAoMVSDuPtuyS0,442
+nkululeko-0.96.1.dist-info/top_level.txt,sha256=bf1k1YKkqcXemNX_cUgoyKqQ3_GVErPqAY-53J36jkM,19
+nkululeko-0.96.1.dist-info/RECORD,,