Olive Recipes For AI Model Optimization Toolkit

This repository compliments Olive, the AI model optimization toolkit, and includes recipes demonstrating its extensive features and use cases. Users of Olive can use these recipes as a reference to either optimize publicly available AI models or to optimize their own proprietary models.

Supported models, architectures, devices and execution providers

Below are list of available recipes grouped by different criteria. Click the link to expand.

Models grouped by model architecture

bert	clip	deepseek	llama	llama3	mistral	phi3	phi4	qwen2	resnet	vit
google-bert-bert-base-multilingual-cased	laion-CLIP-ViT-B-32-laion2B-s34B-b79K	deepseek-ai-DeepSeek-R1-Distill-Llama-8B	meta-llama-Llama-3.2-1B-Instruct	meta-llama-Llama-3.2-1B-Instruct	mistralai-Mistral-7B-Instruct-v0.3	microsoft-Phi-3-mini-128k-instruct	microsoft-Phi-4-mini-instruct	Qwen-Qwen2.5-0.5B-Instruct	microsoft-resnet-50	google-vit-base-patch16-224
intel-bert-base-uncased-mrpc	openai-clip-vit-base-patch16	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B				microsoft-Phi-3-mini-4k-instruct	microsoft-Phi-4-mini-reasoning	Qwen-Qwen2.5-0.5B
	openai-clip-vit-base-patch32	deepseek-ai-DeepSeek-R1-Distill-Qwen-14B				microsoft-Phi-3.5-mini-instruct	microsoft-Phi-4-reasoning-plus	Qwen-Qwen2.5-1.5B-Instruct
		deepseek-ai-DeepSeek-R1-Distill-Qwen-7B				microsoft-Phi-3.5-mini-instruct	microsoft-Phi-4-reasoning	Qwen-Qwen2.5-1.5B-Instruct
								Qwen-Qwen2.5-14B-Instruct
								Qwen-Qwen2.5-3B-Instruct
								Qwen-Qwen2.5-7B-Instruct
								Qwen-Qwen2.5-Coder-0.5B-Instruct
								Qwen-Qwen2.5-Coder-1.5B-Instruct
								Qwen-Qwen2.5-Coder-14B-Instruct
								Qwen-Qwen2.5-Coder-3B-Instruct
								Qwen-Qwen2.5-Coder-7B-Instruct
								deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B

Models grouped by device

cpu	gpu	npu
Qwen-Qwen2.5-1.5B-Instruct	DeepSeek_R1_1.5B_FP16_Model_Builder	Qwen-Qwen2.5-1.5B-Instruct
deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	Llama3.2_1B_Instruct_NVMO_INT4_AWQ	Qwen-Qwen2.5-1.5B-Instruct
google-bert-bert-base-multilingual-cased	Phi3.5_Mini_Instruct_NVMO_INT4_AWQ	Qwen-Qwen2.5-1.5B-Instruct
google-vit-base-patch16-224	Qwen-Qwen2.5-0.5B-Instruct	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B
intel-bert-base-uncased-mrpc (ov)	Qwen-Qwen2.5-0.5B	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B
laion-CLIP-ViT-B-32-laion2B-s34B-b79K	Qwen-Qwen2.5-1.5B-Instruct	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B
meta-llama-Llama-3.2-1B-Instruct	Qwen-Qwen2.5-1.5B-Instruct	google-bert-bert-base-multilingual-cased
microsoft-Phi-3.5-mini-instruct	Qwen-Qwen2.5-14B-Instruct	google-bert-bert-base-multilingual-cased
microsoft-Phi-4-mini-reasoning	Qwen-Qwen2.5-3B-Instruct	google-bert-bert-base-multilingual-cased
microsoft-resnet-50	Qwen-Qwen2.5-7B-Instruct	google-vit-base-patch16-224
openai-clip-vit-base-patch16	Qwen-Qwen2.5-Coder-0.5B-Instruct	google-vit-base-patch16-224
openai-clip-vit-base-patch32	Qwen-Qwen2.5-Coder-1.5B-Instruct	google-vit-base-patch16-224
	Qwen-Qwen2.5-Coder-14B-Instruct	intel-bert-base-uncased-mrpc (AMD)
	Qwen-Qwen2.5-Coder-3B-Instruct	intel-bert-base-uncased-mrpc (ov)
	Qwen-Qwen2.5-Coder-7B-Instruct	intel-bert-base-uncased-mrpc
	Qwen2.5_1.5B_Instruct_NVMO_INT4_AWQ	laion-CLIP-ViT-B-32-laion2B-s34B-b79K (Text)
	deepseek-ai-DeepSeek-R1-Distill-Llama-8B	laion-CLIP-ViT-B-32-laion2B-s34B-b79K (Vision)
	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	laion-CLIP-ViT-B-32-laion2B-s34B-b79K
	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	laion-CLIP-ViT-B-32-laion2B-s34B-b79K
	deepseek-ai-DeepSeek-R1-Distill-Qwen-14B	meta-llama-Llama-3.2-1B-Instruct
	deepseek-ai-DeepSeek-R1-Distill-Qwen-7B	meta-llama-Llama-3.2-1B-Instruct
	google-bert-bert-base-multilingual-cased	meta-llama-Llama-3.2-1B-Instruct
	google-bert-bert-base-multilingual-cased	microsoft-Phi-3.5-mini-instruct
	google-bert-bert-base-multilingual-cased	microsoft-Phi-3.5-mini-instruct
	google-vit-base-patch16-224	microsoft-Phi-3.5-mini-instruct
	google-vit-base-patch16-224	microsoft-Phi-4-mini-reasoning
	google-vit-base-patch16-224	microsoft-resnet-50
	intel-bert-base-uncased-mrpc (ov)	microsoft-resnet-50
	intel-bert-base-uncased-mrpc	microsoft-resnet-50
	intel-bert-base-uncased-mrpc	openai-clip-vit-base-patch16 (Text)
	laion-CLIP-ViT-B-32-laion2B-s34B-b79K	openai-clip-vit-base-patch16 (Vision)
	laion-CLIP-ViT-B-32-laion2B-s34B-b79K	openai-clip-vit-base-patch16
	laion-CLIP-ViT-B-32-laion2B-s34B-b79K	openai-clip-vit-base-patch16
	meta-llama-Llama-3.2-1B-Instruct	openai-clip-vit-base-patch32 (Text)
	meta-llama-Llama-3.2-1B-Instruct	openai-clip-vit-base-patch32 (Vision)
	microsoft-Phi-3-mini-128k-instruct	openai-clip-vit-base-patch32
	microsoft-Phi-3-mini-4k-instruct	openai-clip-vit-base-patch32
	microsoft-Phi-3.5-mini-instruct
	microsoft-Phi-3.5-mini-instruct
	microsoft-Phi-4-mini-instruct
	microsoft-Phi-4-mini-reasoning
	microsoft-Phi-4-reasoning-plus
	microsoft-Phi-4-reasoning
	microsoft-resnet-50
	microsoft-resnet-50
	microsoft-resnet-50
	mistralai-Mistral-7B-Instruct-v0.3
	openai-clip-vit-base-patch16
	openai-clip-vit-base-patch16
	openai-clip-vit-base-patch16
	openai-clip-vit-base-patch32
	openai-clip-vit-base-patch32
	openai-clip-vit-base-patch32

Models grouped by EP

Dml	NvTensorRTRTX	OpenVINO	QNN	VitisAI
Qwen-Qwen2.5-1.5B-Instruct	DeepSeek_R1_1.5B_FP16_Model_Builder	Qwen-Qwen2.5-0.5B-Instruct	Qwen-Qwen2.5-1.5B-Instruct	Qwen-Qwen2.5-1.5B-Instruct
deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	Llama3.2_1B_Instruct_NVMO_INT4_AWQ	Qwen-Qwen2.5-0.5B	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B
google-bert-bert-base-multilingual-cased	Phi3.5_Mini_Instruct_NVMO_INT4_AWQ	Qwen-Qwen2.5-1.5B-Instruct	google-bert-bert-base-multilingual-cased	google-bert-bert-base-multilingual-cased
google-vit-base-patch16-224	Qwen2.5_1.5B_Instruct_NVMO_INT4_AWQ	Qwen-Qwen2.5-14B-Instruct	google-vit-base-patch16-224	google-vit-base-patch16-224
intel-bert-base-uncased-mrpc	google-bert-bert-base-multilingual-cased	Qwen-Qwen2.5-3B-Instruct	intel-bert-base-uncased-mrpc	intel-bert-base-uncased-mrpc (AMD)
laion-CLIP-ViT-B-32-laion2B-s34B-b79K	google-vit-base-patch16-224	Qwen-Qwen2.5-7B-Instruct	laion-CLIP-ViT-B-32-laion2B-s34B-b79K (Text)	laion-CLIP-ViT-B-32-laion2B-s34B-b79K
meta-llama-Llama-3.2-1B-Instruct	intel-bert-base-uncased-mrpc	Qwen-Qwen2.5-Coder-0.5B-Instruct	laion-CLIP-ViT-B-32-laion2B-s34B-b79K (Vision)	meta-llama-Llama-3.2-1B-Instruct
microsoft-Phi-3.5-mini-instruct	laion-CLIP-ViT-B-32-laion2B-s34B-b79K	Qwen-Qwen2.5-Coder-1.5B-Instruct	meta-llama-Llama-3.2-1B-Instruct	microsoft-Phi-3.5-mini-instruct
microsoft-resnet-50	microsoft-resnet-50	Qwen-Qwen2.5-Coder-14B-Instruct	microsoft-Phi-3.5-mini-instruct	microsoft-resnet-50
openai-clip-vit-base-patch16	openai-clip-vit-base-patch16	Qwen-Qwen2.5-Coder-3B-Instruct	microsoft-resnet-50	openai-clip-vit-base-patch16
openai-clip-vit-base-patch32	openai-clip-vit-base-patch32	Qwen-Qwen2.5-Coder-7B-Instruct	openai-clip-vit-base-patch16 (Text)	openai-clip-vit-base-patch32
		deepseek-ai-DeepSeek-R1-Distill-Llama-8B	openai-clip-vit-base-patch16 (Vision)
		deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B	openai-clip-vit-base-patch32 (Text)
		deepseek-ai-DeepSeek-R1-Distill-Qwen-14B	openai-clip-vit-base-patch32 (Vision)
		deepseek-ai-DeepSeek-R1-Distill-Qwen-7B
		google-bert-bert-base-multilingual-cased
		google-vit-base-patch16-224
		intel-bert-base-uncased-mrpc (ov)
		laion-CLIP-ViT-B-32-laion2B-s34B-b79K
		meta-llama-Llama-3.2-1B-Instruct
		microsoft-Phi-3-mini-128k-instruct
		microsoft-Phi-3-mini-4k-instruct
		microsoft-Phi-3.5-mini-instruct
		microsoft-Phi-4-mini-instruct
		microsoft-Phi-4-mini-reasoning
		microsoft-Phi-4-reasoning-plus
		microsoft-Phi-4-reasoning
		microsoft-resnet-50
		mistralai-Mistral-7B-Instruct-v0.3
		openai-clip-vit-base-patch16
		openai-clip-vit-base-patch32

Learn more

• Olive Github Repository
• Getting Started
• Olive Documentation

🤝 Contributions and Feedback

• We welcome contributions! Please read the contribution guidelines for more details on how to contribute to the Olive project.
• For feature requests or bug reports, file a GitHub Issue.
• For general discussion or questions, use GitHub Discussions.

⚖️ License

Copyright (c) Microsoft Corporation. All rights reserved.

Licensed under the MIT License.