SparseFit: Few-shot Prompting with Sparse Fine-tuning for Jointly Generating Predictions and Natural Language Explanations.
Jesus SolanoMardhiyah SanniOana-Maria CamburuPasquale MinerviniPublished in: ACL (1) (2024)
Keyphrases
- fine tuning
- natural language
- fine tune
- viable alternative
- natural language processing
- natural language interface
- high dimensional
- automatically generating
- visual features
- sparse representation
- generating explanations
- semantic interpretation
- semantic representation
- sparse data
- data sets
- video shots
- fine tuned
- video sequences
- dictionary learning
- semantic analysis
- compressive sensing
- video data
- detection algorithm
- natural language generation
- information extraction
- knowledge representation
- language processing
- sparse matrix
- temporal segmentation
- generation process
- question answering
- information retrieval