Advances in Deep Learning

Title		Materials	References
Getting Started
	Welcome to Advances in Deep Learning	Slides

Introduction
	Structures of Deep Networks	Slides
	Training Deep Networks	Slides
	Modern GPU Architectures	Slides Materials

Advanced Training
	Training Large Models	Slides
	Mixed Precision Training	Slides	[1]
	Distributed Training	Slides	[2] [3]
	Zero Redundancy Training	Slides	[4]
	Low-Rank Adapters	Slides	[5]
	Quantization	Slides	[6] [7] [8]
	Quantized Low-Rank Adapters	Slides	[9]
	Low-Rank Projections	Slides	[10]
	Checkpointing	Slides	[11]
	FlashAttention	Slides	[12] [13] [14]
	Open-Source Infrastructure for Model Training	Slides Materials	[15] [16]

Generative Models
	Generative Models	Slides
	Variational Auto Encoders	Slides	[17]
	Generative Adversarial Networks	Slides	[18] [19] [20]
	Flow-Based Models	Slides	[21] [22] [23]
	Auto-Regressive Generation	Slides	[24] [25] [26] [27]
	Vector Quantization	Slides	[28] [29] [30]
	Dall-E	Slides	[28] [31] [32] [33] [34] [35] [36]
	Diffusion Models	Slides	[37] [38] [39]
	Latent Diffusion and State-of-the-Art Models	Slides	[40] [31] [41] [42] [43] [44] [45] [46]
	Which Generative Model Should I Use?	Slides

Large Language Models
	Large Language Models	Slides
	Architectures	Slides	[47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60]
	Generation	Slides Materials	[51] [61] [62]
	Instruction Tuning	Slides
	RLHF	Slides	[63] [64] [65]
	Dpo	Slides	[63] [66]
	Tasks and Datasets	Slides	[67] [68] [69] [70] [71] [72] [73] [74] [75]
	Efficient LLM Training and Inference	Slides
	Sequence Parallelism	Slides	[76] [77] [78] [79]
	Page Attention	Slides	[80] [81] [82]
	Speculative Decoding	Slides	[83] [84]
	Open-Source Infrastructure for LLMs	Slides	[85] [86] [87] [88] [89] [90]
	Tool Use	Slides	[91] [92] [93]
	Structured Outputs	Slides
	Constrained Decoding	Slides	[94] [95] [96]
	Long Context	Slides	[97] [98] [99]
	Retrieval Augmented Generation	Slides	[100] [101] [102] [103] [104]
	Structured Dialogues	Slides	[105] [106] [56] [107] [108] [109] [110] [111] [112]
	Limitations of LLMs	Slides	[113] [114] [115] [116]

References