OWQ: Lessons learned from activation outliers for weight quantization in large language models.
Changhun LeeJungyu JinTaesu KimHyungjun KimEunhyeok ParkPublished in: CoRR (2023)
Keyphrases
- lessons learned
- language model
- language modeling
- n gram
- document retrieval
- future directions
- case study
- retrieval model
- speech recognition
- information retrieval
- context sensitive
- query expansion
- test collection
- probabilistic model
- statistical language models
- smoothing methods
- language models for information retrieval
- language modelling
- participatory design
- query terms
- translation model
- data points
- weighting scheme
- word error rate
- ad hoc information retrieval
- vector space model
- pseudo relevance feedback
- novelty detection
- document ranking
- real world
- text retrieval
- document length
- co occurrence
- statistical language modeling