At [Finegrain](https://finegrain.ai), we're on a mission to automate product photography. Given our "no human in the loop approach", nailing the quality of the outputs we generate is paramount to our success.
That's why we're building Refiners.
It's a framework to easily bridge the last mile quality gap of foundational models like Stable Diffusion or Segment Anything Model (SAM), by adapting them to specific tasks with lightweight trainable and composable patches.
We decided to build Refiners in the open.
It's because model adaptation is a new paradigm that goes beyond our specific use cases. Our hope is to help people looking at creating their own adapters save time, whatever the foundation model they're using.
## Design
We are huge fans of PyTorch (we actually were core committers to [Torch](http://torch.ch/) in another life), but we felt it's too low level for the specific model adaptation task: PyTorch models are generally hard to understand, and their adaptation requires intricate ad hoc code.
Instead, we needed:
- A model structure that's human readable so that you know what models do and how they work right here, right now
- A mechanism to easily inject parameters in some target layers, or between them
- A way to easily pass data (like a conditioning input) between layers even when deeply nested
- Native support for iconic adapter types like LoRAs and their community trained incarnations (hosted on [Civitai](http://civitai.com/) and the likes)
Refiners is designed to tackle all these challenges while remaining just one abstraction away from our beloved PyTorch.
## Downsides
As they say, there is no free lunch. Given Refiners comes with a new model structure, it can only work with models implemented that way. For now, we support Stable Diffusion 1.5, but more is in the making (SDXL, SAM, ...) - stay tuned.
## Overview
The Refiners library is made of:
1. An abstraction layer (called Fluxion) on top of [PyTorch](https://pytorch.org/) to easily build models
2. A zoo of compatible foundational models
3. Adapter APIs to easily patch supported foundational models
4. Training utils to train concrete adapters
5. Conversion scripts to easily use existing community adapters
## Key Concepts
### The Chain class
The `Chain` class is at the core of Refiners. It basically lets you express models as a composition of basic layers (linear, convolution, attention, etc) in a **declarative way**.
E.g.: this is how a Vision Transformer (ViT) looks like with Refiners:
```python
import torch
import refiners.fluxion.layers as fl
class ViT(fl.Chain):
# The Vision Transformer model structure is entirely defined in the constructor. It is
# ready-to-use right after i.e. no need to implement any forward function or add extra logic
The `Adapter` API lets you **easily patch models** by injecting parameters in targeted layers. It comes with built-in support for canonical adapter types like LoRA, but you can also implement your custom adapters with it.
E.g. to inject LoRA layers in all attention's linear layers:
> Note: this will download the original weights from https://huggingface.co/runwayml/stable-diffusion-v1-5 which takes some time. If you already have this repo cloned locally, use the `--from /path/to/stable-diffusion-v1-5` option instead.
Step 2: download and convert a community Pokemon LoRA, e.g. [this one](https://huggingface.co/pcuenq/pokemon-lora)
