Search Results for "encoder"

JEPA (Joint-Embedding Predictive Architecture) captures the idea of predicting missing high-level representations rather than reconstructing pixels, aiming for robust, scalable self-supervised learning. A context encoder ingests visible regions and predicts target embeddings for masked regions produced by a separate target encoder, avoiding low-level reconstruction losses that can overfit to texture. This makes learning focus on semantics and structure, yielding features that transfer well with simple linear probes and minimal fine-tuning. The repository provides training recipes, data pipelines, and evaluation utilities for image JEPA variants and often includes ablations that illuminate which masking and architectural choices matter. ...

...It introduces Momentum Contrast (MoCo), a scalable approach to self-supervised learning that enables visual representation learning without labeled data. The core idea of MoCo is to maintain a dynamic dictionary with a momentum-updated encoder, allowing efficient contrastive learning across large batches. The repository includes implementations for both MoCo v1 and MoCo v2, the latter improving training stability and performance through architectural and augmentation enhancements. Training is optimized for distributed multi-GPU environments, using DistributedDataParallel for speed and simplicity.

VJEPA2 is a next-generation self-supervised learning framework for video that extends the “predict in representation space” idea from i-JEPA to the temporal domain. Instead of reconstructing pixels, it predicts the missing high-level embeddings of masked space-time regions using a context encoder and a slowly updated target encoder. This objective encourages the model to learn semantics, motion, and long-range structure without the shortcuts that pixel-level losses can invite. The architecture is designed to scale: spatiotemporal ViT backbones, flexible masking schedules, and efficient sampling let it train on long clips while remaining stable. ...

...Historically, this repository provided open-source benchmarks and codes for flash flood and river flow forecasting. Full transformer (SimpleTransformer in model_dict): The full original transformer with all 8 encoder and decoder blocks. Requires passing the target in at inference.

...Notebooks for extensive model benchmarking. Modular (but not too much) code base enabling easy testing for new ideas. Text2Spec models (Tacotron, Tacotron2, Glow-TTS, SpeedySpeech). Speaker Encoder to compute speaker embeddings efficiently. Vocoder models (MelGAN, Multiband-MelGAN, GAN-TTS, ParallelWaveGAN, WaveGrad, WaveRNN). If you are only interested in synthesizing speech with the released TTS models, installing from PyPI is the easiest option.