https://github.com/miccunifi/kdpl
공식 코드Improving Zero-shot Generalization of Learned Prompts via Unsupervised Knowledge Distillation
Papers with Code 아카이브
Knowledge Distillation
태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.
- 하위 태스크
- 2
- 평가표
- 11
- 보존된 평가 기록
- 90
Knowledge distillation is the process of transferring knowledge from a large model to a smaller one. While large models (such as very deep neural networks or ensembles of many models) have higher knowledge capacity than small models, this capacity might not be fully utilized.
코드
관련 논문과 평가 결과에 연결된 구현입니다. 출처에서 확인된 저자 구현에는 공식 코드 표시가 있습니다.
https://github.com/micdoh/XLRON
공식 코드Reinforcement Learning for Dynamic Resource Allocation in Optical Networks: Hype or Hope?
https://github.com/michaelsaxon/T2IScoreScore
공식 코드https://github.com/michaelsaxon/cococrola
공식 코드https://github.com/michalspiegel/imgtb
공식 코드IMGTB: A Framework for Machine-Generated Text Detection Benchmarking
https://github.com/microsoft/AirSim-NeurIPS2019-Drone-Racing
공식 코드https://github.com/microsoft/BIPIA
공식 코드Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models
https://github.com/microsoft/ContextualSP
공식 코드MultiSpider: Towards Benchmarking Multilingual Text-to-SQL Semantic Parsing
https://github.com/microsoft/DeepSpeed
공식 코드Extreme Compression for Pre-trained Transformers Made Simple and Efficient
https://github.com/microsoft/Distilled-Sentence-Embedding
공식 코드Scalable Attentive Sentence-Pair Modeling via Distilled Sentence Embedding
https://github.com/microsoft/JARVIS
공식 코드TaskBench: Benchmarking Large Language Models for Task Automation
https://github.com/microsoft/KC
공식 코드Multi-Level Knowledge Distillation for Out-of-Distribution Detection in Text
https://github.com/microsoft/Lightweight-Low-Resource-NMT
공식 코드https://github.com/microsoft/VISOR
공식 코드Benchmarking Spatial Relationships in Text-to-Image Generation
https://github.com/microsoft/cream
공식 코드TinyViT: Fast Pretraining Distillation for Small Vision Transformers
https://github.com/microsoft/deepspeed-mii
공식 코드DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference
https://github.com/microsoft/encoder-decoder-slm
공식 코드Return of the Encoder: Maximizing Parameter Efficiency for SLMs
https://github.com/microsoft/fastformers
공식 코드FastFormers: Highly Efficient Transformer Models for Natural Language Understanding
https://github.com/microsoft/fnl_paper
공식 코드Initialization and Regularization of Factorized Neural Layers
https://github.com/microsoft/lamar-benchmark
공식 코드LaMAR: Benchmarking Localization and Mapping for Augmented Reality
https://github.com/microsoft/lmops
공식 코드https://github.com/microsoft/private-benchmarking
공식 코드TRUCE: Private Benchmarking to Prevent Contamination and Improve Comparative Evaluation of LLMs
https://github.com/microsoft/probts
공식 코드ProbTS: Benchmarking Point and Distributional Forecasting across Diverse Prediction Horizons
https://github.com/microsoft/semiparametric-distillation
공식 코드https://github.com/microsoft/simxns
공식 코드https://github.com/microsoft/subseasonal_data
공식 코드SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking
https://github.com/microsoft/subseasonal_toolkit
공식 코드SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking
https://github.com/microsoft/syntheseus
공식 코드https://github.com/microsoft/unilm
공식 코드EdgeFormer: A Parameter-Efficient Transformer for On-Device Seq2seq Generation
https://github.com/microsoft/unilm/tree/master/beit2
공식 코드BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers
https://github.com/microsoft/vert-papers
공식 코드CoLaDa: A Collaborative Label Denoising Framework for Cross-lingual Named Entity Recognition
https://github.com/microsoft/vert-papers/tree/master/papers/UniTrans
공식 코드https://github.com/microsoft/xtreme-distil-transformers
공식 코드XtremeDistilTransformers: Task Transfer for Task-agnostic Distillation
https://github.com/mihaioltean/evolve-algorithms
공식 코드https://github.com/mikelasz/benchmarking-fairness-imageupsampling
공식 코드https://github.com/mikelma/craftium
공식 코드Craftium: An Extensible Framework for Creating Reinforcement Learning Environments
출처: Papers with Code · CC BY-SA 4.0. 아카이브 자료와 새로 수집한 논문은 별도로 관리합니다.