Skip to content

Commit

Permalink
New README (#392)
Browse files Browse the repository at this point in the history 
* New README

* yolo

* yolo

* Update README.md

* Update README.md

* Trigger CI

* Trigger CI

* Trigger CI

* push

* push

* push

* push

* push

* push

* push

* push

* push

* push

* push
  • Loading branch information
@msaroufim
msaroufim authored Jun 19, 2024
1 parent 6b0ca2d commit 0304281
Showing 1 changed file with 96 additions and 110 deletions.
206 changes: 96 additions & 110 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,147 +2,133 @@

[![](https://dcbadge.vercel.app/api/server/cudamode?style=flat)](https://discord.gg/cudamode)

This repository is currently under heavy development - if you have suggestions on the API or use-cases you'd like to be covered, please open an [issue](https://github.com/pytorch/ao/issues)

## Introduction
`torchao` is a PyTorch library for quantization and sparsity.

torchao is a library to create and integrate high-performance custom data types, layouts and kernels into their PyTorch workflows with up to **2x speedups** with **65%** less VRAM for [inference](#inference) and support for [training](#training)

All with no intrusive code changes and minimal accuracy degradation.

## Benchmarks

### Inference

#### Without intrusive code changes

Quantizing your models is a 1 liner that should work on any model with `nn.Linear` including your favorite HuggingFace model. You can find a more comprehensive usage instructions [here](torchao/quantization/) and a hugginface inference example [here](scripts/hf_eval.py)

```python
from torchao.quantization.quant_api import quantize
m = quantize(m, "int4wo")
```

Benchmarks are run on a machine with a single A100 GPU using the script in `_models/llama` which generates text in a latency-optimized way (batchsize=1)

The models used were `meta-llama/Llama-2-7b-chat-hf` and `meta-llama/Meta-Llama-3-8B`.

| Model | Technique | wikitext-perplexity | Tokens/Second | Memory Bandwidth (GB/s) | Peak Memory (GB) | Model Size (GB) |
| ----------- | ------------------ | ------------------- | ------------- | ----------------------- | ---------------- | --------------- |
| Llama-2-7B | Base (bfloat16) | 12.212 | 105.02 | 1387.78 | 13.21 | 13.90 |
| | int8dq | 12.262 | 9.40 | 62.26 | 6.62 | 8.61 |
| | int8wo | 12.204 | 147.03 | 973.54 | 6.62 | 8.95 |
| | int4wo-64 | 12.843 | 199.81 | 746.45 | 3.74 | 4.75 |
| | int4wo-64-GPTQ | 12.489 | 199.81 | 746.45 | 3.74 | 4.75 |
| Llama-3-8B | Base (bfloat16) | | 94.91 | 1424.58 | 15.01 | 16.43 |
| | int8dq | | 8.41 | 63.23 | 7.52 | 9.24 |
| | int8wo | | 136.75 | 1028.38 | 7.52 | 10.42 |
| | int4wo-64 | | 179.41 | 757.45 | 4.22 | 6.88 |

note: Int8 dynamic quantization works best on compute bound as opposed to memory bound models. Some relatable examples might be [SAM](https://github.com/pytorch-labs/segment-anything-fast) which is compute bound vs Llama at batchsize=1 which is memory bound.

For int4 we make heavy use of [tinygemm](https://github.com/pytorch/ao/blob/cb3bd8c674f2123af232a0231b5e38ddafa756a8/torchao/dtypes/aqt.py#L526) of `torch.ops.aten._weight_int4pack_mm` to bitpack into a layout optimized for tensor cores

And a quick crash course on inference quantization to help parse the above table. Int4 quantization is an ambiguous term because there's the dtype in which a layer is represented and then the dtype in which the computation is done. For example, if you're using Weight-Only (wo) int4 quantization that means that the layer will be upcasted to a larger dtype like fp16 so an int4 matrix multiplication is defined as `F.linear(input, weight.to(input.dtype))`. Dynamic quantization (DQ) primarily targets activations, enabling on-the-fly quantization from higher precision formats like bf16 to lower precision formats such as int8. This process, when supported by hardware, allows for direct computation, such as performing `F.linear(input, weight)`. Naive quantization algorithms are also notoriously sensitive to outliers so we also typically set a group size that applies a scale factor per group of 64 elements in the case of `int4wo64`.


#### With intrusive code changes

In some cases we rewrote popular GenAI models to be significantly faster in native PyTorch as in no C++/CUDA to achieve at the time SOTA inference performance. These involve more intrusive code changes.

* 8x speedups for Image segmentation models with [sam-fast](https://pytorch.org/blog/accelerating-generative-ai)
* 10x speedups for Language models with [gpt-fast](https://pytorch.org/blog/accelerating-generative-ai-2)
* 3x speedup for Diffusion models with [sd-fast](https://pytorch.org/blog/accelerating-generative-ai-3)

### Training

We've added support for semi-structured 2:4 sparsity with 6% end to end speedups on ViT-L

The code change is a 1 liner with the full example available [here](torchao/sparsity/training/)


```python
swap_linear_with_semi_sparse_linear(model, {"seq.0": SemiSparseLinear})
```


## Newer dtypes

* [MX](torchao/prototype/mx_formats) implementing training and inference support with tensors using the [OCP MX spec](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf) data types, which can be described as groupwise scaled float8/float6/float4/int8, with the scales being constrained to powers of two. This work is prototype as the hardware support is not available yet.
* [nf4](torchao/dtypes/nf4tensor.py) which was used to [implement QLoRA](https://github.com/pytorch/torchtune/blob/main/docs/source/tutorials/qlora_finetune.rst) one of the most popular finetuning algorithms without writing custom Triton or CUDA code. Accessible talk [here](https://x.com/HamelHusain/status/1800315287574847701)
* [fp6](torchao/prototype/fp6_llm/) for 2x faster inference over fp16 with an easy to use wrapper api `convert_fp6_llm(model)`

## Composability

A key design principle for us is composability as in any new dtype or layout we provide needs to work with `torch.compile()` and needs to work with `FSDP`. It shouldn't matter if the kernels are written are pure PyTorch, CUDA, C++, or Triton - things should just work! And here is our current strategy
1. Write the dtype, layout or bit packing logic in pure PyTorch and code-generate efficient kernels with torch.compile. You can inspect those kernels with `TORCH_LOGS="output_code" python your_code.py` and check if a single kernel is being generated and if any unnecessary buffers are being created
2. However once you get a kernel, how do you know how good it is? The best way is to benchmark the code-generated code with the best kernel on the market. But packaging custom CPP/CUDA kernels that work on multiple devices is tedious but we've abstracted all the tedium from you with our [custom ops support](./torchao/csrc/) so if you love writing kernels but hate packaging, we'd love to accept contributions for your custom ops. One key benefit is a kernel written as a custom op will just work with no graph breaks with `torch.compile()`. Compilers are great at optimizations like fusions and overhead reduction but it's challenging for compilers to rewrite the math of an algorithm such that it's faster but also numerically stable so we are betting on both compilers and custom ops
3. Finally while historically most quantization has been done for inference there is now a thriving area of research combining lower dtypes and sharding. One popular example is [NF4](torchao/dtypes/nf4tensor.py) which is used to create the QLoRA algorithm and you can define the semantics for how custom tensors should be sharded over multiple devices. We gave an accessible talk on [how to do this](https://x.com/HamelHusain/status/1800315287574847701).

## Get Started

### Installation
`torchao` makes liberal use of several new features in pytorch, it's recommended to use it with the current nightly or latest stable version of PyTorch.
`torchao` makes liberal use of several new features in Pytorch, it's recommended to use it with the current nightly or latest stable version of PyTorch.

Stable Release
```Shell
pip install torchao
pip install torchao --extra-index-url https://download.pytorch.org/whl/test/cu121 # full options are cpu/cu118/cu121/cu124
```

Nightly Release
```Shell
pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cpu # CPU only builds
pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cu118 # CUDA 11.8
pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cu121 # CUDA 12.1
pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cu124 # CUDA 12.4

pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cu121 # full options are cpu/cu118/cu121/cu124
```

From source
## Community Contributions

* [jeromeku](https://github.com/jeromeku) has implemented
* [GaLore](torchao/prototype/galore/) a drop for the Adam Optimizer that allows you to finetune llama 7b on a single 4090 card with up to 70% speedups relative to eager PyTorch
* [DoRA](torchao/prototype/dora) a newer replacement for QLoRA with more promising convergence characteristics
* [Fused int4/fp16 Quant Matmul](torchao/prototype/hqq) which is particularly useful for compute bound kernels showing 4x speedups over tinygemm for larger batch sizes such as 512
* [gau-nernst](https://github.com/gau-nernst) fp6 kernels that are 4x faster than fp16 [torchao/prototype/fp6_llm](torchao/prototype/fp6_llm)
* [vayuda](https://github.com/vayuda) with generic bitpacking kernels that were code generated using pure PyTorch [prototype/common](torchao/prototype/common)
* [andreaskopf](https://github.com/andreaskoepf) and [melvinebenezer](https://github.com/melvinebenezer) with [1 bit LLMs](torchao/prototype/dtypes) Bitnet 1.58 bitpacked into uin2 and fully code-generated with torch.compile

## How to contribute

This repository is currently under heavy development
* If you have suggestions on the API or use cases you'd like to be covered, please open an [issue](https://github.com/pytorch/ao/issues)
* If you'd like to co-develop the library with us please join us on #torchao on [discord.gg/cudamode](https://discord.gg/cudamode) - there are a lot of dtypes out there and we could use a lot more hands to make them go brrr

Installation instructions

```Shell
git clone https://github.com/pytorch/ao
cd ao
python setup.py install
python setup.py install
```

If you plan to be developing the library run:
If you're contributing a feature ao
```Shell
pip install -r dev-requirements.txt
python setup.py develop
```

** Note:
If you are running into any issues while building `ao` cpp extensions you can instead build using
For *most* developers you probably want to skip building custom C++/CUDA extensions for faster iteration cycles

```shell
USE_CPP=0 python setup.py install
```

### Quantization

```python
import torch
import torchao

# inductor settings which improve torch.compile performance for quantized modules
torch._inductor.config.force_fuse_int_mm_with_mul = True
torch._inductor.config.use_mixed_mm = True

# Plug in your model and example input
model = torch.nn.Sequential(torch.nn.Linear(32, 64)).cuda().to(torch.bfloat16)
input = torch.randn(32,32, dtype=torch.bfloat16, device='cuda')

# perform autoquantization and compilation
q_model = torchao.autoquant(torch.compile(model, mode='max-autotune'))
q_model(input)
```

### Sparsity

```python
import torch
from torch.sparse import to_sparse_semi_structured, SparseSemiStructuredTensor
from torch.ao.pruning import WeightNormSparsifier

# bfloat16 CUDA model
model = torch.nn.Sequential(torch.nn.Linear(64, 64)).cuda().to(torch.bfloat16)

# Accuracy: Finding a sparse subnetwork
sparse_config = []
for name, mod in model.named_modules():
if isinstance(mod, torch.nn.Linear):
sparse_config.append({"tensor_fqn": f"{name}.weight"})

sparsifier = WeightNormSparsifier(sparsity_level=1.0,
sparse_block_shape=(1,4),
zeros_per_block=2)

# attach FakeSparsity
sparsifier.prepare(model, sparse_config)
sparsifier.step()
sparsifier.squash_mask()
# now we have dense model with sparse weights

# Performance: Accelerated sparse inference
for name, mod in model.named_modules():
if isinstance(mod, torch.nn.Linear):
mod.weight = torch.nn.Parameter(to_sparse_semi_structured(mod.weight))
```

To learn more try out our APIs, you can check out API examples in
* [quantization](./torchao/quantization)
* [sparsity](./torchao/sparsity)
* [dtypes](./torchao/dtypes)


## Supported Features
1. [Quantization algorithms](./torchao/quantization)
- [Int8 weight-only](https://github.com/pytorch/ao/blob/main/torchao/quantization/weight_only.py) quantization
- [Int4 weight-only](https://github.com/pytorch/pytorch/blob/main/aten/src/ATen/native/cuda/int4mm.cu) quantization
- [GPTQ](https://github.com/pytorch/ao/blob/main/torchao/quantization/GPTQ.py) and [Smoothquant](https://github.com/pytorch/ao/blob/main/torchao/quantization/smoothquant.py) for low latency inference
- High level [torchao.autoquant API](https://github.com/pytorch/ao/blob/main/torchao/quantization/autoquant.py) and [kernel autotuner](https://github.com/pytorch/ao/blob/main/torchao/kernel/autotuner.py) targeting SOTA performance across varying model shapes on consumer and enterprise GPUs
2. [Sparsity algorithms](./torchao/sparsity) such as Wanda that help improve accuracy of sparse networks
3. Support for lower precision [dtypes](./torchao/dtypes) such as
- [nf4](https://github.com/pytorch/ao/blob/main/torchao/dtypes/nf4tensor.py) which was used to [implement QLoRA](https://github.com/pytorch/torchtune/blob/main/docs/source/tutorials/qlora_finetune.rst) without writing custom Triton or CUDA code
- [uint4](https://github.com/pytorch/ao/blob/main/torchao/dtypes/uint4.py)
- [MX](https://github.com/pytorch/ao/blob/main/torchao/prototype/mx_formats) implementing training and inference support with tensors using the [OCP MX spec](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf) data types, which can be described as groupwise scaled float8/float6/float4/int8, with the scales being constrained to powers of two. This work is prototype as the hardware support is not available yet.
4. [Bleeding Edge Kernels](./torchao/prototype/) for experimental kernels without backwards compatibility guarantees
- [GaLore](https://github.com/pytorch/ao/tree/main/torchao/prototype/galore) for memory efficient finetuning
- [fused HQQ Gemm Kernel](https://github.com/pytorch/ao/tree/main/torchao/prototype/hqq) for compute bound workloads
- [FP6-LLM](torchao/prototype/fp6_llm) mixed matmul FP16 x FP6 kernel for io bound workloads

## Our Goals

* Composability with `torch.compile`: We rely heavily on `torch.compile` to write pure PyTorch code and codegen efficient kernels. There are however limits to what a compiler can do so we don't shy away from writing our custom CUDA/Triton kernels
* Composability with `FSDP`: The new support for FSDP per parameter sharding means engineers and researchers alike can experiment with different quantization and distributed strategies concurrently.
* Performance: We measure our performance on every commit using an A10G. We also regularly run performance benchmarks on the [torchbench](https://github.com/pytorch/benchmark) suite
* Heterogeneous Hardware: Efficient kernels that can run on CPU/GPU based server (w/ torch.compile) and mobile backends (w/ ExecuTorch).
* Packaging kernels should be easy: We support custom [CUDA and Triton extensions](./torchao/csrc/) so you can focus on writing your kernels and we'll ensure that they work on most operating systems and devices

## Integrations

torchao has been integrated with other libraries including

* [torchtune](https://github.com/pytorch/torchtune/blob/main/recipes/quantization.md) leverages our 8 and 4 bit weight-only quantization techniques with optional support for GPTQ
* [Executorch](https://github.com/pytorch/executorch/tree/main/examples/models/llama2#quantization) leverages our GPTQ implementation for both 8da4w (int8 dynamic activation with int4 weight) and int4 weight-only quantization.
* [HQQ](https://github.com/mobiusml/hqq/blob/master/hqq/backends/torchao.py) leverages our int4mm kernel for low latency inference

## Success stories
Our kernels have been used to achieve SOTA inference performance on

* Image segmentation models with [sam-fast](https://pytorch.org/blog/accelerating-generative-ai)
* Language models with [gpt-fast](https://pytorch.org/blog/accelerating-generative-ai-2)
* Diffusion models with [sd-fast](https://pytorch.org/blog/accelerating-generative-ai-3)

## License

`torchao` is released under the [BSD 3](https://github.com/pytorch-labs/ao/blob/main/LICENSE) license.

0 comments on commit 0304281

Please sign in to comment.