Multi Task Learning With Shared Representations And Task Specific Heads

via “multi-task dataset enabling transfer learning across detection, segmentation, captioning, and pose tasks”

330K images with object detection, segmentation, and captions.

Unique: Single dataset with annotations for 7+ vision tasks enables multi-task learning and transfer learning; shared image set allows models to learn task-agnostic visual representations and transfer knowledge across tasks

vs others: More comprehensive than single-task datasets; enables multi-task learning unlike separate datasets for each task; shared image set ensures fair comparison across tasks unlike different image distributions

via “cross-task knowledge transfer through shared representations”

Microsoft's unified model for diverse vision tasks.

Unique: Achieves knowledge transfer across 6+ vision tasks through a single unified seq2seq architecture, where shared visual encoding and decoder parameters enable cross-task learning without task-specific branches or ensemble methods

vs others: Outperforms task-specific models on low-data scenarios through knowledge transfer, though with 5-10% lower peak performance on high-data tasks compared to specialized models

via “multi-task learning with shared representations and task-specific heads”

PyTorch NLP framework with contextual embeddings.

Unique: Implements multi-task learning through a unified architecture where a shared BiLSTM encoder feeds into task-specific output heads (CRF for tagging, softmax for classification), enabling flexible combinations of different task types; supports dynamic task weighting during training to balance task contributions

vs others: More efficient than training separate models for each task while maintaining task-specific output constraints; enables knowledge transfer between related tasks, improving performance on low-resource tasks; simpler to implement than complex multi-task architectures with task-specific encoders

via “multi-task learning and auxiliary objective training”

fill-mask model by undefined. 1,90,34,963 downloads.

Unique: RoBERTa's improved pretraining produces representations with stronger task-agnostic semantic content, enabling more effective multi-task learning with less task interference compared to BERT — auxiliary tasks improve primary task performance by 1-3% absolute on average

vs others: More effective for multi-task learning than single-task fine-tuning due to stronger base representations; requires more careful tuning than task-specific models but provides better generalization and inference efficiency than ensemble approaches

via “task-conditioned-query-generation”

image-segmentation model by undefined. 90,906 downloads.

Unique: Implements task conditioning via learnable query tokens (e.g., 100 queries for panoptic, 150 for semantic) that are concatenated with positional encodings and processed through the same transformer decoder stack. This differs from multi-head approaches (separate decoder heads per task) by forcing shared feature representations while allowing task-specific query distributions.

vs others: Reduces model parameters by 25-30% vs separate task-specific decoders while maintaining within 0.5 mIoU of task-specific models, enabling efficient multi-task deployment. However, task-specific models can be independently optimized, potentially achieving 1-2 mIoU higher performance if model size is not constrained.

via “task-conditioned-prediction-head-with-dynamic-routing”

image-segmentation model by undefined. 54,407 downloads.

Unique: Implements task-conditioned routing where the task token modulates both which prediction branches execute and how intermediate features are processed through learned gating mechanisms. Unlike multi-head approaches that always compute all heads, this design conditionally activates branches based on task requirements.

vs others: Reduces inference latency by 15-20% compared to always-active multi-head decoders when only semantic segmentation is needed, while maintaining the flexibility to switch to instance/panoptic tasks without model reloading.

via “multi-task learning with panoptic and instance segmentation heads”

OpenMMLab Detection Toolbox and Benchmark

Unique: Implements panoptic segmentation by combining instance predictions (from detection head) with semantic segmentation predictions (from semantic head) in a unified framework, where task-specific losses are weighted and summed, enabling end-to-end training of multiple related tasks with shared backbone

vs others: More integrated than combining separate instance and semantic segmentation models because it shares backbone features and enables joint optimization; more flexible than Detectron2's panoptic segmentation because it supports arbitrary combinations of detection, instance, and semantic heads

via “multi-task-learning-with-shared-representations”

A very simple framework for state-of-the-art NLP

Unique: Flair's multi-task learning framework uses shared embedding and encoder layers with task-specific output heads, enabling efficient knowledge transfer while maintaining task-specific prediction heads. This architecture allows fine-grained control over task weighting and loss functions, supporting both hard parameter sharing and soft parameter sharing strategies.

vs others: Flair's multi-task learning is more flexible than single-task pipelines (supports arbitrary task combinations) and more interpretable than end-to-end multi-task transformers, with explicit control over task weighting and loss functions.

via “multi-task learning and transfer learning dataset composition”

Dataset by nyu-mll. 3,97,160 downloads.

Unique: Provides task-aware dataset composition through HuggingFace Datasets' interleaving API, enabling weighted sampling of heterogeneous tasks (e.g., oversample RTE's 2.5K examples to match QQP's 364K) without manual replication logic. Preserves task identity through metadata columns for downstream loss weighting.

vs others: Enables multi-task training without custom dataset construction by providing task-aware composition utilities, vs alternatives like manual concatenation (loses task identity) or separate task-specific models (no transfer learning). Native integration with HuggingFace Transformers enables multi-task fine-tuning with minimal code changes.

via “multi-task vision-language pre-training with shared representations”

* ⭐ 02/2022: [data2vec: A General Framework for Self-supervised Learning in Speech, Vision and... (Data2vec)](https://proceedings.mlr.press/v162/baevski22a.html)

Unique: Combines multi-task learning with data bootstrapping: the same unified model is trained on both understanding tasks (retrieval) and generation tasks (captioning, VQA) using bootstrapped training data. This creates a virtuous cycle where the captioner generates training data for other tasks, and multi-task learning improves the captioner's quality.

vs others: Outperforms single-task models by leveraging shared representations and multi-task learning, achieving SOTA on multiple benchmarks simultaneously. Unlike separate task-specific models, BLIP's unified approach reduces model size and inference latency while improving generalization through positive transfer between tasks.

via “multi-task visual policy learning with task-agnostic world models”

* ⏫ 02/2023: [Grounding Large Language Models in Interactive Environments with Online RL (GLAM)](https://arxiv.org/abs/2302.02662)

Unique: DreamerV3's task-agnostic world model learns shared visual representations without explicit task conditioning, relying on the policy learning objective to extract task-relevant information from the shared latent space. This contrasts with task-conditioned approaches (e.g., MTRL baselines) that explicitly encode task identity, making DreamerV3 more flexible for discovering emergent task structure.

vs others: Achieves better sample efficiency and generalization than task-conditioned baselines by learning task-invariant visual dynamics, while avoiding the computational overhead of task-specific world models or explicit task embeddings.

via “multi-task vision model with shared representation”

* ⏫ 12/2023: [VideoPoet: A Large Language Model for Zero-Shot Video Generation (VideoPoet)](https://arxiv.org/abs/2312.14125)

Unique: Uses single encoder-decoder backbone with shared parameters across all vision tasks, trained on 5.4B diverse annotations to learn unified representation handling variable spatial hierarchies and semantic granularities. Contrasts with ensemble or task-specific approaches by consolidating capabilities into one model.

vs others: Reduces deployment complexity and memory footprint compared to maintaining separate detection (YOLO), segmentation (DeepLab), grounding (ALBEF), and captioning (BLIP) models, though individual task performance vs specialized baselines unknown.

via “multi-task adapter composition for vision-language understanding”

* ⭐ 04/2022: [Winoground: Probing Vision and Language Models for Visio-Linguistic... (Winoground)](https://arxiv.org/abs/2204.03162)

Unique: Implements task-specific adapter composition for multimodal models with explicit routing logic, enabling independent training of task adapters while maintaining shared backbone — distinct from single-task adapter approaches and multi-task learning methods that require joint training

vs others: More memory-efficient than training separate full models per task and more flexible than single-task adapters, enabling dynamic task switching without model reloading

via “multi-task and meta-learning frameworks”


Unique: Provides practical implementations of multi-task learning with systematic task weighting strategies and meta-learning approaches (MAML, Prototypical Networks) from scratch, combined with empirical analysis of when multi-task learning helps vs hurts generalization. Includes frameworks for identifying task relatedness and designing shared representations.

vs others: More practical and implementation-focused than academic meta-learning papers by providing working code and systematic frameworks for task weighting and architecture design, while more comprehensive than generic transfer learning tutorials by covering few-shot learning and rapid adaptation.

via “multi-task agent learning with shared trajectory representation”

### Other Papers <a name="2023op"></a>

Unique: Enables multi-task learning by conditioning the language model policy on task descriptions, allowing a single agent to learn from trajectories across diverse tasks and generalize to new tasks — this is distinct from task-specific agents that require separate training for each task

vs others: More sample-efficient than single-task agents because it leverages cross-task patterns, and more flexible than fixed multi-task architectures because task conditioning is learned end-to-end

via “multi-task robot policy learning from diverse demonstrations”

## Historical Papers <a name="history"></a>

Unique: Trains a single transformer model on 700+ diverse tasks without task-specific heads or explicit multi-task loss weighting, relying on language conditioning and shared token embeddings to learn task-agnostic manipulation primitives. This contrasts with prior multi-task approaches that use separate output heads or task-specific adapters.

vs others: Achieves better generalization to novel objects and scenes than task-specific policies trained on equivalent data, and scales more efficiently than ensemble or modular approaches by sharing all transformer parameters across tasks.