README.md

# SliceX AI™ ELM (Efficient Language Models)
**ELM** (which stands for **E**fficient **L**anguage **M**odels) is the first version in the series of cutting-edge language models from [SliceX AI](https://slicex.ai) that is designed to achieve the best in class performance in terms of _quality_, _throughput_ & _memory_.

<div align="center">
  <img src="logo.png" width="256"/>
</div>

ELM is designed to be a modular and customizable family of neural networks that are highly efficient and performant. Today we are sharing the first version in this series: **ELM-v0.1** models. 

_Model:_ ELM introduces a new type of _(de)-composable LLM model architecture_ along with the algorithmic optimizations required to learn (training) and run (inference) these models. At a high level, we train a single ELM model in a self-supervised manner (during pre-training phase) but once trained the ELM model can be sliced in many ways to fit different user/task needs. The optimizations can be applied to the model either during the pre-training and/or fine-tuning stage. 

_Fast Inference with Customization:_ Once trained, the ELM model architecture permits flexible inference strategies at runtime depending on the deployment needs. For instance, the ELM model can  be _decomposed_ into smaller slices, i.e., smaller (or larger) models can be extracted from the original model to create multiple inference endpoints. Alternatively, the original (single) ELM model can be loaded _as is_ for inference and different slices within the model can be queried directly to power faster inference. This provides an additional level of flexibility for users to make compute/memory tradeoffs depending on their application and runtime needs.

## ELM-v0.1 Model Release
Models are located in the `models` folder. ELM models in this repository comes in three sizes (elm-1.0, elm-0.75 and elm-0.25) and supports the following use-cases.
- news_classification
- toxicity_detection
- news_content_generation
- news_summarization

## Setup ELM
### Download ELM repo
```bash
git clone [email protected]:slicexai/elm-v0.1
sudo apt-get intall git-lfs 
git lfs install
```
### Installation
```bash
cd elm-v0.1
pip install -r requirements.txt
```

(Optional) Installing git-lfs without sudo,
```bash
wget https://github.com/git-lfs/git-lfs/releases/download/v3.2.0/git-lfs-linux-amd64-v3.2.0.tar.gz
tar -xzf git-lfs-linux-amd64-v3.2.0.tar.gz
PATH=$PATH:/<absolute-path>/git-lfs-3.2.0/
git lfs install
```

## Download ELM task-specific model checkpoints

Download elm-1.0 model checkpoints
```bash
git lfs pull -I models/elm-1.0_news_classification/ckpt.pt
git lfs pull -I models/elm-1.0_toxicity_detection/ckpt.pt
git lfs pull -I models/elm-1.0_news_content_generation/ckpt.pt
git lfs pull -I models/elm-1.0_news_summarization/ckpt.pt
```

Download elm-0.75 model checkpoints
```bash
git lfs pull -I models/elm-0.75_news_classification/ckpt.pt
git lfs pull -I models/elm-0.75_toxicity_detection/ckpt.pt
git lfs pull -I models/elm-0.75_news_content_generation/ckpt.pt
git lfs pull -I models/elm-0.75_news_summarization/ckpt.pt
```

Download elm-0.25 model checkpoints
```bash
git lfs pull -I models/elm-0.25_news_classification/ckpt.pt
git lfs pull -I models/elm-0.25_toxicity_detection/ckpt.pt
git lfs pull -I models/elm-0.25_news_content_generation/ckpt.pt
```




## How to use - Run ELM on a sample task (e.g., news classification)
```bash
python run.py <elm-model-directory>
E.g. python run.py models/elm-0.75_news_classification
``` 
Prompts for the specific tasks can be found in the corresponding checkpoint directory. See an example below in the form of `models/elm-0.75_news_classification/example_prompts.json`.
```json
{
    "inputs": ["GM May Close Plant in Europe  DETROIT (Reuters) - General Motors Corp. &lt;A HREF=\"http://www.investor.reuters.com/FullQuote.aspx?ticker=GM.N target=/stocks/quickinfo/fullquote\"&gt;GM.N&lt;/A&gt; will likely  cut some jobs in Europe and may close a plant there as part of  a restructuring plan under development to try to return the  region to profitability, the U.S. automaker said on Wednesday."],
    "template": "[INST]Below is a news article. Please classify it under one of the following classes (World, Business, Sports, Sci/Tech). Please format your response as a JSON payload.\n\n### Article: {input}\n\n### JSON Response:[/INST]"
}
```

Running the above command returns the following response

```json
{
    "prompt": "[INST]Below is a news article. Please classify it under one of the following classes (World, Business, Sports, Sci/Tech). Please format your response as a JSON payload.\n\n### Article: GM May Close Plant in Europe  DETROIT (Reuters) - General Motors Corp. &lt;A HREF=\"http://www.investor.reuters.com/FullQuote.aspx?ticker=GM.N target=/stocks/quickinfo/fullquote\"&gt;GM.N&lt;/A&gt; will likely  cut some jobs in Europe and may close a plant there as part of  a restructuring plan under development to try to return the  region to profitability, the U.S. automaker said on Wednesday.\n\n### JSON Response:[/INST]",
    "response": "{'text_label': 'Business'}"
}
```