Documentation and small fixes

experimaestro · Dec 7, 2023 · 3322f5f · 3322f5f
1 parent f5f4cb1
commit 3322f5f
Show file tree

Hide file tree

Showing 14 changed files with 5 additions and 78 deletions.
diff --git a/docs/source/data/index.rst b/docs/source/data/index.rst
@@ -7,7 +7,5 @@ as well as its own datasets (e.g. hard negatives datasets).
 .. toctree::
    :maxdepth: 2
 
-   types
-   irds
    xpmir
    adapters
diff --git a/docs/source/data/irds.rst b/docs/source/data/irds.rst
diff --git a/docs/source/data/types.rst b/docs/source/data/types.rst
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -55,7 +55,6 @@ Table of Contents
    text/index
    papers/index
    pretrained
-   misc
 
 
 Indices and tables

diff --git a/docs/source/learning/index.rst b/docs/source/learning/index.rst
@@ -31,7 +31,7 @@ Trainers
 Trainers are responsible for defining the the way to train
 a learnable scorer.
 
-.. autoxpmconfig:: xpmir.letor.trainers.Trainer
+.. autoxpmconfig:: xpmir.learning.trainers.Trainer
 .. autoxpmconfig:: xpmir.learning.trainers.multiple.MultipleTrainer
 
 .. autoxpmconfig:: xpmir.letor.trainers.LossTrainer

diff --git a/docs/source/letor/batchwise.rst b/docs/source/letor/batchwise.rst
@@ -20,4 +20,5 @@ Samplers
 
 
 .. autoxpmconfig:: BatchwiseSampler
-.. autoxpmconfig:: xpmir.documents.samplers.BatchwiseRandomSpanSampler
+
+.. autoxpmconfig:: xpmir.documents.samplers.RandomSpanSampler
diff --git a/docs/source/letor/index.rst b/docs/source/letor/index.rst
@@ -45,7 +45,6 @@ Samplers
 
 Samplers provide samples in the form of *records*. They all inherit from:
 
-.. autoxpmconfig:: Sampler
 .. autoclass:: SerializableIterator
 
 

diff --git a/docs/source/letor/pairwise.rst b/docs/source/letor/pairwise.rst
@@ -4,7 +4,6 @@ Pairwise
 Trainers are responsible for defining the the way to train
 a learnable scorer.
 
-.. autoxpmconfig:: xpmir.letor.trainers.Trainer
 .. autoxpmconfig:: xpmir.learning.trainers.multiple.MultipleTrainer
 
 .. autoxpmconfig:: xpmir.letor.trainers.LossTrainer
@@ -45,6 +44,5 @@ Pairwise
 .. autoxpmconfig:: PairwiseSampleDatasetFromTSV
 .. autoxpmconfig:: PairwiseSamplerFromTSV
 .. autoxpmconfig:: ModelBasedHardNegativeSampler
-.. autoxpmconfig:: TripletBasedInBatchNegativeSampler
 
 .. autoxpmconfig:: xpmir.letor.samplers.hydrators.PairwiseTransformAdapter
diff --git a/docs/source/letor/pointwise.rst b/docs/source/letor/pointwise.rst
@@ -4,7 +4,6 @@ Pointwise
 Trainers are responsible for defining the the way to train
 a learnable scorer.
 
-.. autoxpmconfig:: xpmir.letor.trainers.Trainer
 .. autoxpmconfig:: xpmir.learning.trainers.multiple.MultipleTrainer
 
 .. autoxpmconfig:: xpmir.letor.trainers.LossTrainer

diff --git a/docs/source/misc.rst b/docs/source/misc.rst
diff --git a/docs/source/retrieval.rst b/docs/source/retrieval.rst
@@ -69,7 +69,6 @@ Anserini
 --------
 
 .. autoxpmconfig:: xpmir.index.anserini.Index
-.. autoxpmconfig:: xpmir.interfaces.anserini.Index
 .. autoxpmconfig:: xpmir.interfaces.anserini.AnseriniRetriever
 .. autoxpmconfig:: xpmir.interfaces.anserini.IndexCollection
 .. autoxpmconfig:: xpmir.interfaces.anserini.SearchCollection

diff --git a/requirements.txt b/requirements.txt
@@ -1,6 +1,6 @@
 # Experimaestro
 
-experimaestro>=1.2.1
+experimaestro>=1.3.4
 datamaestro>=0.8.13
 datamaestro_text>=2023.11.22
 ir_datasets

diff --git a/src/xpmir/letor/samplers/__init__.py b/src/xpmir/letor/samplers/__init__.py
@@ -1,4 +1,3 @@
-import random
 from pathlib import Path
 from typing import Iterator, List, Tuple, Dict, Any
 import numpy as np
@@ -344,48 +343,6 @@ def iter(random):
         return RandomSerializableIterator(self.random, iter)
 
 
-class TripletBasedInBatchNegativeSampler(PairwiseSampler):
-    """An in-batch negative sampler which generate the triplets,
-    which use the postives of the other in batch as the negatives"""
-
-    sampler: Param[PairwiseSampler]
-    """The base pairwise sampler"""
-
-    batch_size: Param[int]
-    """How many triplets to be used for building the ibn"""
-
-    def initialize(self, random):
-        super().initialize(random)
-        self.sampler.initialize(random)
-
-    def pairwise_iter(self) -> SerializableIterator[PairwiseRecord, Any]:
-        def iter(pair_iter):
-            while True:
-                topics = []
-                positives = []
-                for _, record in zip(range(self.batch_size), pair_iter):
-                    topics.append(record.query)
-                    positives.append(record.positive)
-                all_qry = [
-                    topic for topic in topics for _ in range(self.batch_size - 1)
-                ]
-                all_pos = [pos for pos in positives for _ in range(self.batch_size - 1)]
-                pos_as_neg = positives * self.batch_size
-                pos_index = [(self.batch_size + 1) * i for i in range(self.batch_size)]
-                all_neg = [
-                    doc for i, doc in enumerate(pos_as_neg) if i not in pos_index
-                ]
-
-                # randomize, to make the same document not gather too close
-                mapping = list(zip(all_qry, all_pos, all_neg))
-                for _ in range(30000):
-                    random.shuffle(mapping)
-                    for (topic, positive, negative) in mapping:
-                        yield PairwiseRecord(topic, positive, negative)
-
-        return SerializableIteratorAdapter(self.sampler.pairwise_iter(), iter)
-
-
 class PairwiseInBatchNegativesSampler(BatchwiseSampler):
     """An in-batch negative sampler constructured from a pairwise one"""
 

diff --git a/src/xpmir/test/test_documented.py b/src/xpmir/test/test_documented.py
@@ -9,4 +9,4 @@ def test_documented():
 
     analyzer.analyze()
     analyzer.report()
-    analyzer.assert_no_undocumented()
+    analyzer.assert_valid_documentation()