1 repository
Awesome GitHub RepositoriesAlignment Error Metrics
Metrics specifically designed to quantify the accuracy of token-level sequence alignments.
Distinct from Phoneme-Based Alignment: Focuses on quantitative error rates for word/token alignment rather than the alignment model architectures themselves
Explore 1 awesome GitHub repository matching artificial intelligence & ml · Alignment Error Metrics. Refine with filters or upvote what's useful.
Awesome Alignment Error Metrics GitHub Repositories
Găsește cele mai bune repo-uri cu AI.Vom căuta cele mai potrivite repository-uri folosind AI.
- facebookresearch/fairseq
facebookresearch/fairseq
32,228Vezi pe GitHubFairseq is a PyTorch toolkit for sequence-to-sequence modeling, specializing in neural machine translation, automatic speech recognition, and large-scale language model training. It provides a framework for processing and aligning diverse data sources, including text, audio, and video, to support tasks such as speech-to-text conversion and multimodal sequence learning. The project is distinguished by its distributed training capabilities, which utilize parameter sharding, mixed-precision training, and CPU offloading to handle models that exceed single-device memory. It also includes specializ
Measures the quality of predicted word alignments using specialized metrics like Alignment Error Rate.
Python
