Image captioning with weight pruning in PyTorch
COMIC: This is the code repo of our TMM2019 work titled "COMIC: Towards a Compact Image Captioning Model with Attention".
Simple but Comprehensive PyTorch Implementation of Image Captioning Models.
SacreEOS is a signature generator and implementation helper for the Self Critical Sequence Training
SacreEOS experiments
PyTorch image captioning: ResNet50 encoder + Bahdanau attention + LSTM decoder, trained in two stages — cross-entropy warmup then CIDEr-optimized Self-Critical Sequence Training (SCST). Outperforms Show, Attend and Tell (Xu et al., 2015) on Flickr30k. Gradio demo included.
LoRA/QLoRA/AdaLoRA/DoRA/VeRA fine-tuning for RAG faithfulness Arabic (XQuAD) & Malay (Belebele) QA with mT5, self-critical sequence training, and independent NLI verification.
Add a description, image, and links to the self-critical-sequence-training topic page so that developers can more easily learn about it.
To associate your repository with the self-critical-sequence-training topic, visit your repo's landing page and select "manage topics."