Extractive Question Answering With Span Prediction

via “answer span extraction and evaluation metrics for reading comprehension”

95K trivia questions requiring cross-document reasoning.

Unique: Provides multiple valid answer spans per question and ground-truth span annotations within evidence documents, enabling training of span-based extractive QA models with proper handling of answer paraphrasing. The span-level annotations allow fine-grained evaluation of reading comprehension beyond simple answer matching.

vs others: More flexible than SQuAD (which has single answer spans) by allowing multiple valid spans, and more realistic than curated datasets by including noisy documents where answer spans may be paraphrased or implicit

via “span-based answer annotation with character-level indexing”

150K reading comprehension questions including unanswerable ones.

Unique: Uses character-level span indexing rather than token-level, making answers independent of tokenization choices. This enables fair comparison across models with different tokenizers and avoids off-by-one errors from token boundaries.

vs others: More precise than free-form answer generation (which requires BLEU/ROUGE metrics) and more tokenizer-agnostic than token-level span prediction, enabling reproducible evaluation across different model architectures.

via “question-answering via extractive span selection from context”

fill-mask model by undefined. 11,20,072 downloads.

Unique: Implements extractive QA via dual classification heads predicting start/end token positions, leveraging bidirectional context from 24-layer transformer to disambiguate answer boundaries without generating new text, enabling interpretable and hallucination-free answers directly traceable to source passages

vs others: More efficient and interpretable than generative QA models (T5, GPT) for document-based QA, with lower latency and no hallucination risk, but limited to questions answerable by span extraction and requires fine-tuning on QA datasets for competitive performance

via “extractive question-answering with span prediction”

question-answering model by undefined. 2,87,434 downloads.

Unique: Fine-tuned on SQuAD 2.0 with whole-word masking (masking entire words rather than subword tokens during pre-training), improving robustness to morphological variations and reducing spurious attention to subword boundaries. This contrasts with standard BERT which uses subword masking.

vs others: Faster and more interpretable than generative QA models (GPT-based) because it predicts token spans rather than generating sequences, enabling real-time inference on CPU and guaranteed source attribution without hallucination.

via “extractive question-answering with span selection”

question-answering model by undefined. 6,23,377 downloads.

Unique: Fine-tuned specifically on SQuAD v2 dataset which includes unanswerable questions, enabling the model to recognize when no valid answer exists in the context rather than hallucinating answers — a critical distinction from v1-only models that always force an answer

vs others: Outperforms BERT-base on SQuAD v2 benchmarks due to RoBERTa's improved pretraining (robustness to input perturbations, larger batch sizes), while remaining lightweight enough for CPU inference unlike larger models like ELECTRA or DeBERTa

via “token-level span prediction with logit output”

question-answering model by undefined. 8,99,590 downloads.

Unique: Exposes raw transformer logits for both start and end positions without post-processing, allowing consumers to implement custom decoding strategies (e.g., constrained span selection, confidence thresholding, ensemble voting) rather than forcing a single argmax decoding path.

vs others: Provides more flexibility than models that return only the top-1 answer span, enabling advanced inference patterns like beam search or confidence-based filtering, but requires more sophisticated downstream handling compared to models that return pre-selected answers.

via “extractive question-answering with span prediction”

question-answering model by undefined. 2,25,087 downloads.

Unique: Uses knowledge distillation from BERT-base to achieve 40% parameter reduction while maintaining 97% performance on SQuAD, enabling sub-100ms inference on CPU. Implements dual-head token classification (start/end logits) rather than sequence-to-sequence generation, making answers deterministic and directly grounded in source text.

vs others: Faster and more memory-efficient than full BERT-base QA models (66M vs 110M parameters) while maintaining accuracy, and more reliable than generative QA models because answers are always extractive spans from the source material

via “extractive question-answering with whole-word masking”

question-answering model by undefined. 1,93,069 downloads.

Unique: Whole-word masking pretraining strategy masks all subword tokens of a word together (vs. standard BERT's random subword masking), forcing the model to learn stronger semantic representations and improving performance on span-based tasks like QA where token boundaries matter

vs others: Outperforms standard BERT-large on SQuAD v2 by 1-2 F1 points due to whole-word masking; smaller inference footprint than dense retrieval + generation pipelines (single forward pass vs. retrieval + LLM generation)

via “extractive question-answering with span prediction”

question-answering model by undefined. 1,16,670 downloads.

Unique: Distilled from BERT-base using knowledge distillation (40% parameter reduction, 60% speedup) while maintaining 97% of original accuracy on SQuAD v1.1, achieved through layer-wise distillation and attention transfer — not just pruning or quantization

vs others: 40% faster inference than BERT-base with minimal accuracy loss, and 3-5x smaller model size than full BERT, making it practical for production QA systems where latency and memory are constraints

via “extractive question-answering with span selection”

question-answering model by undefined. 1,45,572 downloads.

Unique: Trained on SQuAD 2.0 which includes unanswerable questions, enabling the model to output null answers when questions cannot be answered from context — a critical distinction from SQuAD 1.1 models that assume all questions are answerable

vs others: Smaller and faster than full-scale QA models (BERT-base, ELECTRA) while maintaining competitive accuracy on SQuAD benchmarks, making it ideal for resource-constrained deployments and real-time inference scenarios

via “extractive question-answering with span prediction”

question-answering model by undefined. 3,19,759 downloads.

Unique: Fine-tuned specifically on SQuAD v2 which includes 30% unanswerable questions, enabling the model to output null/no-answer predictions with confidence scores rather than forcing spurious answers — a critical distinction from v1-only models that always predict an answer span

vs others: More reliable than BERT-base QA models due to RoBERTa's improved pretraining (dynamic masking, larger batches) and outperforms smaller extractive models on SQuAD v2 by 3-5 F1 points while remaining deployable on modest hardware

via “span-based answer extraction with confidence scoring”

question-answering model by undefined. 1,61,301 downloads.

Unique: Uses independent start/end token classification with softmax scoring over sequence positions, enabling efficient O(n²) span enumeration and confidence-based ranking; confidence computed as product of start/end probabilities rather than joint span probability, making it computationally efficient but potentially miscalibrated

vs others: Faster than generative QA models (no autoregressive decoding); more interpretable than black-box span selection; enables confidence-based filtering unlike models without probability outputs; simpler than pointer networks but less flexible for non-contiguous answers

via “multilingual extractive question-answering with span prediction”

question-answering model by undefined. 1,90,899 downloads.

Unique: Uses DeBERTa-v3's disentangled attention (separate content and position attention heads) instead of standard multi-head attention, improving efficiency and cross-lingual generalization; multilingual training on 100+ languages via mBERT-style token embeddings enables zero-shot transfer without language-specific fine-tuning

vs others: Outperforms mBERT and XLM-RoBERTa on SQuAD 2.0 multilingual benchmarks while using 40% fewer parameters than XLM-R-large, making it faster for edge deployment while maintaining cross-lingual accuracy

via “token-level span extraction with confidence scoring”

question-answering model by undefined. 1,24,380 downloads.

Unique: Outputs token-level logits for both start and end positions, enabling fine-grained analysis and custom span ranking logic vs black-box APIs that return only top-1 answer

vs others: Provides interpretability and flexibility for downstream ranking/filtering vs fixed single-answer output, at the cost of requiring more complex post-processing

via “extractive question-answering on korean text”

question-answering model by undefined. 78,274 downloads.

Unique: Uses ELECTRA discriminator architecture (efficient token classification via replaced-token detection pretraining) fine-tuned on KorQuAD, enabling faster inference than BERT-based Korean QA models while maintaining competitive accuracy on Korean-specific linguistic phenomena like agglutination and complex morphology

vs others: Faster inference and smaller model size than mBERT or XLM-RoBERTa Korean QA variants while achieving higher accuracy on KorQuAD benchmark due to ELECTRA's discriminative pretraining approach

via “extractive question-answering with span prediction”

question-answering model by undefined. 40,750 downloads.

Unique: Fine-tuned on SQuAD 2.0 with whole-word masking pre-training strategy (masks complete words rather than subword tokens), improving semantic understanding compared to standard BERT. Uses cased tokenization preserving capitalization information, beneficial for named entity recognition within answers.

vs others: Faster inference than generative QA models (BART, T5) with lower memory footprint, but cannot answer unanswerable questions or synthesize information like SQuAD 2.0-aware models; more accurate on SQuAD benchmarks than smaller DistilBERT variants due to larger 24-layer architecture.

via “vietnamese extractive question-answering with span prediction”

question-answering model by undefined. 1,09,840 downloads.

Unique: RoBERTa-large backbone fine-tuned specifically on Vietnamese SQuAD data, combining English pre-training knowledge with Vietnamese-specific downstream task adaptation; uses token-level span prediction rather than generative decoding, enabling deterministic answer extraction directly from source passages

vs others: Outperforms monolingual Vietnamese models and English-only QA systems on Vietnamese benchmarks due to large pre-trained encoder, while remaining faster and more interpretable than generative Vietnamese QA models that require autoregressive decoding

via “extractive question-answering on passages with span prediction”

question-answering model by undefined. 32,657 downloads.

Unique: MobileBERT uses bottleneck layer architecture with knowledge distillation from BERT-large, achieving 4.3x smaller model size (25MB) and 5.5x faster inference than BERT-base while maintaining 95%+ accuracy on SQuAD v2. This is achieved through inverted bottleneck blocks (wide intermediate layers, narrow hidden states) and aggressive parameter sharing, not just pruning.

vs others: Significantly faster and smaller than BERT-base QA models (25MB vs 110MB, 5.5x speedup) with minimal accuracy loss, making it the preferred choice for mobile/edge deployment; slower but more accurate than DistilBERT for QA tasks due to superior architecture design.

via “passage-level answer span extraction with position tracking”

question-answering model by undefined. 48,782 downloads.

Unique: Predicts token-level start/end positions which are converted to character offsets via the tokenizer's offset_mapping, enabling precise answer localization without post-hoc string matching; supports both token and character-level indexing for flexibility

vs others: More precise than regex-based answer extraction (handles tokenization edge cases); token-level prediction is more efficient than character-level models; offset tracking enables direct document highlighting without string search

via “extractive question-answering on document passages”

question-answering model by undefined. 66,453 downloads.

Unique: Fine-tuned on SQuAD 2.0 which includes 20% unanswerable questions, enabling the model to predict when no valid answer exists in a passage rather than forcing an incorrect extraction — a critical capability for production QA systems handling adversarial or out-of-scope queries

vs others: More reliable than generic BERT-base on unanswerable questions and achieves higher F1 on SQuAD 2.0 than models trained only on SQuAD 1.1, making it production-ready for real-world FAQ systems where not all queries have answers