[ASR] Confidence measure -> method renames (NVIDIA#7434)

* measure -> method

Signed-off-by: Aleksandr Laptev <alaptev@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Aleksandr Laptev <alaptev@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Signed-off-by: BestJuly <chntaoli@163.com>

nemo/collections/asr/metrics/rnnt_wer.py

@@ -100,16 +100,16 @@ class AbstractRNNTDecoding(ConfidenceMixin):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -140,7 +140,7 @@ class AbstractRNNTDecoding(ConfidenceMixin):
                     timestep during greedy decoding. Setting to larger values allows longer sentences
                     to be decoded, at the cost of increased execution time.
                 preserve_frame_confidence: Same as above, overrides above value.
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.
@@ -277,7 +277,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                         ),
                         preserve_alignments=self.preserve_alignments,
                         preserve_frame_confidence=self.preserve_frame_confidence,
-                        confidence_measure_cfg=self.confidence_measure_cfg,
+                        confidence_method_cfg=self.confidence_method_cfg,
                     )
                 else:
                     self.decoding = greedy_decode.GreedyTDTInfer(
@@ -291,7 +291,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                         ),
                         preserve_alignments=self.preserve_alignments,
                         preserve_frame_confidence=self.preserve_frame_confidence,
-                        confidence_measure_cfg=self.confidence_measure_cfg,
+                        confidence_method_cfg=self.confidence_method_cfg,
                     )
             else:
                 self.decoding = greedy_decode.GreedyMultiblankRNNTInfer(
@@ -304,7 +304,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                     ),
                     preserve_alignments=self.preserve_alignments,
                     preserve_frame_confidence=self.preserve_frame_confidence,
-                    confidence_measure_cfg=self.confidence_measure_cfg,
+                    confidence_method_cfg=self.confidence_method_cfg,
                 )
 
         elif self.cfg.strategy == 'greedy_batch':
@@ -320,7 +320,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                         ),
                         preserve_alignments=self.preserve_alignments,
                         preserve_frame_confidence=self.preserve_frame_confidence,
-                        confidence_measure_cfg=self.confidence_measure_cfg,
+                        confidence_method_cfg=self.confidence_method_cfg,
                     )
                 else:
                     self.decoding = greedy_decode.GreedyBatchedTDTInfer(
@@ -334,7 +334,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                         ),
                         preserve_alignments=self.preserve_alignments,
                         preserve_frame_confidence=self.preserve_frame_confidence,
-                        confidence_measure_cfg=self.confidence_measure_cfg,
+                        confidence_method_cfg=self.confidence_method_cfg,
                     )
 
             else:
@@ -348,7 +348,7 @@ def __init__(self, decoding_cfg, decoder, joint, blank_id: int):
                     ),
                     preserve_alignments=self.preserve_alignments,
                     preserve_frame_confidence=self.preserve_frame_confidence,
-                    confidence_measure_cfg=self.confidence_measure_cfg,
+                    confidence_method_cfg=self.confidence_method_cfg,
                 )
 
         elif self.cfg.strategy == 'beam':
@@ -1005,16 +1005,16 @@ class RNNTDecoding(AbstractRNNTDecoding):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -1047,7 +1047,7 @@ class RNNTDecoding(AbstractRNNTDecoding):
 
                 preserve_frame_confidence: Same as above, overrides above value.
 
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.

nemo/collections/asr/metrics/rnnt_wer_bpe.py

@@ -100,16 +100,16 @@ class RNNTBPEDecoding(AbstractRNNTDecoding):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -142,7 +142,7 @@ class RNNTBPEDecoding(AbstractRNNTDecoding):
 
                 preserve_frame_confidence: Same as above, overrides above value.
 
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.

nemo/collections/asr/metrics/wer.py

@@ -258,16 +258,16 @@ class AbstractCTCDecoding(ConfidenceMixin):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -300,7 +300,7 @@ class AbstractCTCDecoding(ConfidenceMixin):
                 preserve_alignments: Same as above, overrides above value.
                 compute_timestamps: Same as above, overrides above value.
                 preserve_frame_confidence: Same as above, overrides above value.
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.
@@ -389,7 +389,7 @@ def __init__(self, decoding_cfg, blank_id: int):
                 preserve_alignments=self.preserve_alignments,
                 compute_timestamps=self.compute_timestamps,
                 preserve_frame_confidence=self.preserve_frame_confidence,
-                confidence_measure_cfg=self.confidence_measure_cfg,
+                confidence_method_cfg=self.confidence_method_cfg,
             )
 
         elif self.cfg.strategy == 'beam':
@@ -1037,16 +1037,16 @@ class CTCDecoding(AbstractCTCDecoding):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -1079,7 +1079,7 @@ class CTCDecoding(AbstractCTCDecoding):
                 preserve_alignments: Same as above, overrides above value.
                 compute_timestamps: Same as above, overrides above value.
                 preserve_frame_confidence: Same as above, overrides above value.
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.

nemo/collections/asr/metrics/wer_bpe.py

@@ -74,16 +74,16 @@ class CTCBPEDecoding(AbstractCTCDecoding):
                     from the `token_confidence`.
                 aggregation: Which aggregation type to use for collapsing per-token confidence into per-word confidence.
                     Valid options are `mean`, `min`, `max`, `prod`.
-                measure_cfg: A dict-like object which contains the measure name and settings to compute per-frame
+                method_cfg: A dict-like object which contains the method name and settings to compute per-frame
                     confidence scores.
 
-                    name: The measure name (str).
+                    name: The method name (str).
                         Supported values:
                             - 'max_prob' for using the maximum token probability as a confidence.
                             - 'entropy' for using a normalized entropy of a log-likelihood vector.
 
                     entropy_type: Which type of entropy to use (str).
-                        Used if confidence_measure_cfg.name is set to `entropy`.
+                        Used if confidence_method_cfg.name is set to `entropy`.
                         Supported values:
                             - 'gibbs' for the (standard) Gibbs entropy. If the alpha (α) is provided,
                                 the formula is the following: H_α = -sum_i((p^α_i)*log(p^α_i)).
@@ -116,7 +116,7 @@ class CTCBPEDecoding(AbstractCTCDecoding):
                 preserve_alignments: Same as above, overrides above value.
                 compute_timestamps: Same as above, overrides above value.
                 preserve_frame_confidence: Same as above, overrides above value.
-                confidence_measure_cfg: Same as above, overrides confidence_cfg.measure_cfg.
+                confidence_method_cfg: Same as above, overrides confidence_cfg.method_cfg.
 
             "beam":
                 beam_size: int, defining the beam size for beam search. Must be >= 1.

nemo/collections/asr/models/confidence_ensemble.py

@@ -25,7 +25,7 @@
 from nemo.collections.asr.models.hybrid_rnnt_ctc_models import EncDecHybridRNNTCTCModel
 from nemo.collections.asr.parts.utils.asr_confidence_utils import (
     ConfidenceConfig,
-    ConfidenceMeasureConfig,
+    ConfidenceMethodConfig,
     get_confidence_aggregation_bank,
     get_confidence_measure_bank,
 )
@@ -61,7 +61,7 @@ def to_confidence_config(self) -> ConfidenceConfig:
         return ConfidenceConfig(
             exclude_blank=self.exclude_blank,
             aggregation=self.aggregation,
-            measure_cfg=ConfidenceMeasureConfig(
+            method_cfg=ConfidenceMethodConfig(
                 name=name, entropy_type=entropy_type, alpha=self.alpha, entropy_norm=entropy_norm,
             ),
         )
@@ -126,7 +126,7 @@ def compute_confidence(hypothesis: Hypothesis, confidence_cfg: ConfidenceConfig)
         hypothesis: generated hypothesis as returned from the transcribe
             method of the ASR model.
         confidence_cfg: confidence config specifying what kind of
-            measure/aggregation should be used.
+            method/aggregation should be used.
 
     Returns:
         float: confidence score.
@@ -135,12 +135,12 @@ def compute_confidence(hypothesis: Hypothesis, confidence_cfg: ConfidenceConfig)
     filtered_logprobs = get_filtered_logprobs(hypothesis, confidence_cfg.exclude_blank)
     vocab_size = filtered_logprobs.shape[1]
     aggr_func = get_confidence_aggregation_bank()[confidence_cfg.aggregation]
-    if confidence_cfg.measure_cfg.name == "max_prob":
+    if confidence_cfg.method_cfg.name == "max_prob":
         conf_type = "max_prob"
         alpha = 1.0
     else:
-        conf_type = f"entropy_{confidence_cfg.measure_cfg.entropy_type}_{confidence_cfg.measure_cfg.entropy_norm}"
-        alpha = confidence_cfg.measure_cfg.alpha
+        conf_type = f"entropy_{confidence_cfg.method_cfg.entropy_type}_{confidence_cfg.method_cfg.entropy_norm}"
+        alpha = confidence_cfg.method_cfg.alpha
     conf_func = get_confidence_measure_bank()[conf_type]
 
     conf_value = aggr_func(conf_func(filtered_logprobs, v=vocab_size, t=alpha)).cpu().item()