Accelerating the Training of Large Language Models using Efficient Activation Rematerialization and Optimal Hybrid Parallelism.
Tailing YuanYuliang LiuXucheng YeShenglong ZhangJianchao TanBin ChenChengru SongDi ZhangPublished in: USENIX ATC (2024)
Keyphrases
- language model
- language modeling
- probabilistic model
- document retrieval
- statistical language models
- information retrieval
- speech recognition
- n gram
- retrieval model
- language modelling
- test collection
- training set
- ad hoc information retrieval
- machine learning
- query expansion
- vector space model
- relevance model
- language models for information retrieval
- query terms
- context sensitive
- smoothing methods
- hidden markov models