# Gemma

In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on Gemma models. For illustration purposes, we utilize the [google/gemma-7b-it](https://huggingface.co/google/gemma-7b-it) and [google/gemma-2b-it](https://huggingface.co/google/gemma-2b-it) as a reference Gemma model.

## 0. Requirements
To run these examples with IPEX-LLM, we have some recommended requirements for your machine, please refer to [here](../README.md#recommended-requirements) for more information.

## Example: Predict Tokens using `generate()` API
In the example [generate.py](./generate.py), we show a basic use case for a Gemma model to predict the next N tokens using `generate()` API, with IPEX-LLM INT4 optimizations.
### 1. Install
We suggest using conda to manage the Python environment:

After installing conda, create a Python environment for IPEX-LLM:

On Linux:

```bash
conda create -n llm python=3.11 # recommend to use Python 3.11
conda activate llm

# install the latest ipex-llm nightly build with 'all' option
pip install --pre --upgrade ipex-llm[all] --extra-index-url https://download.pytorch.org/whl/cpu

# According to Gemma's requirement, please make sure you are using a stable version of Transformers, 4.38.1 or newer.
pip install "transformers>=4.38.1"
```

On Windows:

```cmd
conda create -n llm python=3.11
conda activate llm

pip install --pre --upgrade ipex-llm[all]

pip install "transformers>=4.38.1"
```

### 2. Run
```
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
```

Arguments info:
- `--repo-id-or-model-path REPO_ID_OR_MODEL_PATH`: argument defining the huggingface repo id for the Gemma model (e.g. `google/gemma-7b-it`) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be `'google/gemma-7b-it'`.
- `--prompt PROMPT`: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be `'What is AI?'`.
- `--n-predict N_PREDICT`: argument defining the max number of tokens to predict. It is default to be `32`.

> **Note**: When loading the model in 4-bit, IPEX-LLM converts linear layers in the model into INT4 format. In theory, a *X*B model saved in 16-bit will requires approximately 2*X* GB of memory for loading, and ~0.5*X* GB memory for further inference.
>
> Please select the appropriate size of the Gemma model based on the capabilities of your machine.

#### 2.1 Client
On client Windows machine, it is recommended to run directly with full utilization of all cores:
```cmd
python ./generate.py 
```

#### 2.2 Server
For optimal performance on server, it is recommended to set several environment variables (refer to [here](../README.md#best-known-configuration-on-linux) for more information), and run the example with all the physical cores of a single socket.

E.g. on Linux,
```bash
# set IPEX-LLM env variables
source ipex-llm-init

# e.g. for a server with 48 cores per socket
export OMP_NUM_THREADS=48
numactl -C 0-47 -m 0 python ./generate.py
```

#### 2.4 Sample Output
#### [google/gemma-7b-it](https://huggingface.co/google/gemma-7b-it)
```log
Inference time: xxxx s
-------------------- Output --------------------
user
What is AI?
model
**Artificial Intelligence (AI)** is a field of computer science that involves the creation of intelligent machines capable of performing tasks typically requiring human intelligence, such as learning,
```

#### [google/gemma-2b-it](https://huggingface.co/google/gemma-2b-it)
```log
Inference time: xxxx s
-------------------- Output --------------------
user
What is AI?
model
**Artificial intelligence (AI)** is the simulation of human cognitive functions, such as learning, reasoning, and problem-solving, by machines. AI systems are designed
```