From 940e11a4cebe76fcc846ca89e5062e69ee51bea7 Mon Sep 17 00:00:00 2001
From: Sarthak Malhotra <142551729+sarthakwer@users.noreply.github.com>
Date: Thu, 10 Apr 2025 11:08:12 -0700
Subject: [PATCH] Update README.md with distillation + finetuning colab
 notebook

Added example "Distillation + Finetuning" under subheading Finetuning. The example distills the capabilities of a large model into a small model using Together's finetuning API.
Details: I demonstrate how to use Curator to distill capabilities from a large language model to a much smaller 8B parameter model. I use Yelp restaurant reviews dataset to train a sentiment analysis model. I then generate a synthetic dataset using Bespokelabs's curator and finetune a model using Together's finetuning API. The finetuned model shows a 12% improvement in accuracy while being 14x cheaper than LLM
---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index abff837..ad5dc0d 100644
--- a/README.md
+++ b/README.md
@@ -38,6 +38,7 @@ While the code examples are primarily written in Python/JS, the concepts can be
 | [Long Context Finetuning For Repetition](https://github.com/togethercomputer/together-cookbook/blob/main/LongContext_Finetuning_RepetitionTask.ipynb) | Fine-tuning LLMs to repeat back words in long sequences. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/togethercomputer/together-cookbook/blob/main/LongContext_Finetuning_RepetitionTask.ipynb) |
 | [Summarization Long Context Finetuning](https://github.com/togethercomputer/together-cookbook/blob/main/Summarization_LongContext_Finetuning.ipynb) | Long context fine-tuning to improve summarization capabilities. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/togethercomputer/together-cookbook/blob/main/Summarization_LongContext_Finetuning) |
 | [Conversation Finetuning](https://github.com/togethercomputer/together-cookbook/blob/main/Multiturn_Conversation_Finetuning.ipynb) | Fine-tuning LLMs on multi-step conversations. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/togethercomputer/together-cookbook/blob/main/Multiturn_Conversation_Finetuning.ipynb) |
+| [Distillation and Finetuning](https://colab.research.google.com/drive/1Zfl3g7POsqqYQqkzXdyhYRSAymLhZugn?usp=sharing) | Distillation using Curator and finetuning using Together's API | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1Zfl3g7POsqqYQqkzXdyhYRSAymLhZugn?usp=sharing) |
 | Retrieval-augmented generation  |  | |
 | [RAG_with_Reasoning_Models](https://github.com/togethercomputer/together-cookbook/blob/main/RAG_with_Reasoning_Models.ipynb) | RAG + source citations with DeepSeek R1. | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/togethercomputer/together-cookbook/blob/main/RAG_with_Reasoning_Models.ipynb) |
 | [MultiModal_RAG_with_Nvidia_Deck](https://github.com/togethercomputer/together-cookbook/blob/main/MultiModal_RAG_with_Nvidia_Investor_Slide_Deck.ipynb) | Multimodal RAG using Nvidia investor slides | [![Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/togethercomputer/together-cookbook/blob/main/MultiModal_RAG_with_Nvidia_Investor_Slide_Deck.ipynb) [![](https://uohmivykqgnnbiouffke.supabase.co/storage/v1/object/public/landingpage/youtubebadge.svg)](https://youtu.be/IluARWPYAUc?si=gG90hqpboQgNOAYG)|