# Amazon SageMaker Examples

### JumpStart Foundation Models

These examples provide quick walkthroughs to get you up and running with Amazon SageMaker JumpStart Foundation Models.

- [Text Generation Llama-2](text-generation-llama-2.ipynb) and [Text Generation Llama-2-chat](text-generation-llama-2-chat.ipynb) demonstrate text generation using Meta's Llama-2 pretrained and fine-tuned chat models, respectfully.
- [Text2Text Generation Flan T5](text2text-generation-flan-t5-ul2.ipynb) demonstrates Text2Text generation using state-of-the-art pretrained Flan T5 models from [Hugging Face](https://huggingface.co/docs/transformers/model_doc/flan-t5) which take an input text containing the task and returns the output of the accomplished task. 
- [Text2Text Generation BloomZ](text2text-generation-bloomz.ipynb) demonstrates Text generation using state-of-the-art pretrained BloomZ 7B1 models from [Hugging Face](https://huggingface.co/bigscience/bloomz-7b1) which take an input text containing the task and returns the output of the accomplished task. In addition to the tasks that Flan T5 can perform, BloomZ can perform multilingual text classification, question and answering, code generation, paragraph rephrase, and More.
- [Text Generation Few Shot Learning](text-generation-few-shot-learning.ipynb) demonstrates Text generation using state-of-the-art pretrained GPT-J-6B models from [Hugging Face](https://huggingface.co/EleutherAI/gpt-j-6B) which takes a text string as input and predicts a sequence of next few words. These models can, for example, fill in incomplete text or paraphrase.
- [BloomZ 176B Few Shot and Zero Shot Learning](bloom-z-176b-few-shot-and-zero-shot-learning.ipynb) demonstrates over 20 few-shot and zero-shot learning tasks using state-of-the-art pretrained BloomZ 176B model from [Hugging Face](https://huggingface.co/bigscience/bloomz).
- [Text Generation Chatbot](text-generation-chatbot.ipynb) demonstrates text generation within a chatbot context using a variety of pretrained JumpStart models, to include a variety of models, to include [Falcon-7B-Instruct](https://huggingface.co/tiiuae/falcon-7b-instruct), [Falcon-40B-Instruct](https://huggingface.co/tiiuae/falcon-40b-instruct), [RedPajama-INCITE-Chat-3B](https://huggingface.co/togethercomputer/RedPajama-INCITE-Chat-3B-v1), [RedPajama-INCITE-7B-Chat](https://huggingface.co/togethercomputer/RedPajama-INCITE-7B-Chat), and [GPT-NeoXT-Chat-Base-20B](https://huggingface.co/togethercomputer/GPT-NeoXT-Chat-Base-20B).
- [Retrieval Augmented Generation: Question and Answering based on custom data](question_answering_retrieval_augmented_generation/question_answering_jumpstart_knn.ipynb) demonstrates how to use retrieval-augmented generation based approach to perform question and answering with JumpStart embedding and large language models.
- [Retrieval Augmented Generation: Question and Answering based on custom data with LangChain](question_answering_retrieval_augmented_generation/question_answering_langchain_jumpstart.ipynb) demonstrates how to use retrieval-augmented generation based approach to perform question and answering with LangChain library and JumpStart embedding and large language models.
- [Domain adaption fine-tuning of large language model](domain-adaption-finetuning-gpt-j-6b.ipynb) demonstrates how to use SageMaker SDK to fine-tune large language model like GPT-J-6B from [Hugging Face](https://huggingface.co/EleutherAI/gpt-j-6b) on domain-specific dataset and deploy the fine-tuned model for inference.