update instruction

README.md

@@ -1 +1,23 @@
 # Starbucks
+Starbucks: Improved Training for 2D Matryoshka Embeddings
+
+### General guidelines
+Our codebase is built on top of torch and transformers.
+
+We recommend using a conda environment to install the required dependencies.
+To install the required dependencies:
+
+```bash
+conda create -n starbucks python=3.10
+conda activate starbucks
+
+pip install torch
+pip install transformers datasets peft
+pip install deepspeed accelerate
+```
+
+For SMAE pre-training, see [smae](smae/README.md).
+
+For SRL fine-tuning on retrieval task, see [retrieval](retrieval/README.md).
+
+For SRL fine-tuning on STS task, see [sts](sts/README.md).

retrieval/README.md

@@ -5,8 +5,6 @@ Our training code for passage retrieval is based on [Tevatron](https://github.co
 
 To install Tevatron:
 ```bash
-pip install transformers datasets peft
-pip install deepspeed accelerate
 pip install faiss-cpu # or 'conda install pytorch::faiss-gpu' for faiss gpu search
 pip install wandb # for logging
 git clone https://github.com/texttron/tevatron.git

sts/README.md

@@ -10,8 +10,6 @@ This repo supports for three types of fine-tuning:
 ## Installation
 ```bash
 pip install sentence_transformers
-pip install datasets
-pip install torch
 ```