tract
fairseq
Our great sponsors
tract | fairseq | |
---|---|---|
20 | 89 | |
2,050 | 29,205 | |
2.9% | 1.6% | |
9.8 | 6.6 | |
4 days ago | 3 days ago | |
Rust | Python | |
Apache 2.0/MIT | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
tract
-
Are there any ML crates that would compile to WASM?
Tract is the most well known ML crate in Rust, which I believe can compile to WASM - https://github.com/sonos/tract/. Burn may also be useful - https://github.com/burn-rs/burn.
-
[Discussion] What crates would you like to see?
tract!!
-
tract VS burn - a user suggested alternative
2 projects | 25 Mar 2023
-
Machine Learning Inference Server in Rust?
we use tract for inference, integrated into our runtime and services.
- onnxruntime
- Rust Native ML Frameworks?
-
Neural networks - what crates to use?
Not for training, but for inference this looks nice: https://github.com/sonos/tract
-
Brain.js: GPU Accelerated Neural Networks in JavaScript
There's also tract, from sonos[0]. 100% rust.
I'm currently trying to use it to do speech recognition with a variant of the Conformer architecture (exported to ONNX).
The final goal is to do it in WASM client-side.
[0] https://github.com/sonos/tract
-
Serving ML at the Speed of Rust
As the article notes, there isn't any official Rust-native support for any common frameworks.
tract (https://github.com/sonos/tract) seems like the most mature for ONNX (for which TF/PT export is good nowadays), and recently it successfully implemented BERT.
-
Run deep neural network models from scratch
There are some DL libraries written in Rust: https://github.com/sonos/tract , https://docs.rs/neuronika/latest/neuronika/index.html . The second one could be used for training, I think.
fairseq
- Sequence-to-Sequence Toolkit Written in Python
- Unsupervised (Semi-Supervised) ASR/STT training recipes
-
Nvidia's 900 tons of GPU muscle bulks up server market, slims down wallets
> Is there really no way to partition the workload to run with 16gb memory per card?
It really depends and this can get really complicated really fast. I'll give a tldr and then a longer explanation.
TLDR:
Yes, you can easily split networks up. If your main bottleneck is batch size (i.e. training) then there aren't huge differences in spreading across multiple GPUs assuming you have good interconnects (GPU direct is supported). If you're running inference and the model fits on the card you're probably fine too unless you need to do things like fancy inference batching (i.e. you have LOTS of users)
Longer version:
You can always split things up. If we think about networks we recognize some nice properties about how they operate as mathematical groups. Non-residual networks are compositional, meaning each layer can be treated as a sub network (every residual block can be treated this way too). Additionally, we may have associative and distributive properties depending on the architecture (some even have commutative!). So we can use these same rules to break apart networks in many different ways. There are often performance hits for doing this though, as it practically requires you touching the disk more often but in some more rare cases (at least to me, let me know if you know more) they can help.
I mentioned the batching above and this can get kinda complicated. There are actually performance differences when you batch in groups of data (i.e. across GPUs) compared to batching on a single accelerator. This difference isn't talked about a lot. But it is going to come down to how often your algorithm depends on batching and what operations are used, such as batch norm. The batch norm is calculated across the GPU's batch, not the distributed batch (unless you introduce blocking). This is because your gradients AND inference are going to be computed differently. In DDP your whole network is cloned across cards so you basically run inference on multiple networks and then do an all reduce on the loss then calculate the gradient and then recopy the weights to all cards. There is even a bigger difference when you use lazy regularization (don't compute gradients for n-minibatches). GANs are notorious for using this and personally I've seen large benefits to distributed training for these. GANs usually have small batch sizes and aren't getting anywhere near the memory of the card anyways (GANs are typically unstable so large batch sizes can harm them), but also pay attention to this when evaluating papers (of course as well as how much hyper-parameter tuning has been done. This is always tricky when comparing works, especially between academia and big labs. You can easily be fooled by which is a better model. Evaluating models is way tougher than people give credit to and especially in the modern era of LLMs. I could rant a lot about just this alone). Basically in short, we can think of this as an ensembling method, except our models are actually identical (you could parallel reduce lazily too and that will create some periodic divergence between your models but that's not important for conceptually understanding, just worth noting).
There is are also techniques to split a single model up called model sharding and checkpointing. Model sharding is where you split a single model across multiple GPUs. You're taking advantage of the compositional property of networks, meaning that as long as there isn't a residual layer between your split location you can actually treat one network as a series of smaller networks. This has obvious drawbacks as you need to feed one into another and so the operations have to be synchronous, but sometimes this isn't too bad. Checkpointing is very similar but you're just doing the same thing on the same GPU. Your hit here is in I/O, but may or may not be too bad with GPU Direct and highly depends on your model size (were you splitting because batch size or because model size?).
This is all still pretty high level but if you want to dig into it more META developed a toolkit called fairseq that will do a lot of this for you and they optimized it
https://engineering.fb.com/2021/07/15/open-source/fsdp/
https://github.com/facebookresearch/fairseq
TLDR: really depends on your use case, but it is a good question.
-
Talk back and forth with AI like you would with a person
How do they do the text to voice conversion so fast? https://github.com/facebookresearch/fairseq/tree/main (open source takes sub-minute to do text to voice.
-
Voice generation AI (TTS)
It might be worth checking out Meta's TTS tho, I haven't gotten the chance to fiddle around with it but it looks somewhat promising https://github.com/facebookresearch/fairseq/tree/main/examples/mms
-
Translation app with TTS (text-to-speech) for Persian?
They have instructions on how to use it in command line and a notebook on how to use it as a python library.
- Why no work on open source TTS (Text to speech) models
-
Meta's Massively Multilingual Speech project supports 1k languages using self supervised learning
Github - https://github.com/facebookresearch/fairseq/tree/main/examples/mms Paper - https://research.facebook.com/publications/scaling-speech-technology-to-1000-languages/
-
AI — weekly megathread!
Meta released a new open-source model, Massively Multilingual Speech (MMS) that can do both speech-to-text and text-to-speech in 1,107 languages and can also recognize 4,000+ spoken languages. Existing speech recognition models only cover approximately 100 languages out of the 7,000+ known spoken languages. [Details | Research Paper | GitHub].
- Meta's MMS: Scaling Speech Technology to 1000+ languages (How to Run colab)
What are some alternatives?
onnxruntime-rs - Rust wrapper for Microsoft's ONNX Runtime (version 1.8)
gpt-neox - An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
MTuner - MTuner is a C/C++ memory profiler and memory leak finder for Windows, PlayStation 4 and 3, Android and other platforms
transformers - 🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
wonnx - A WebGPU-accelerated ONNX inference run-time written 100% in Rust, ready for native and the web
DeepSpeed - DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
ncurses-rs - A low-level ncurses wrapper for Rust
text-to-text-transfer-transformer - Code for the paper "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer"
linfa - A Rust machine learning framework.
espnet - End-to-End Speech Processing Toolkit
tangram - Tangram makes it easy for programmers to train, deploy, and monitor machine learning models.
Pytorch - Tensors and Dynamic neural networks in Python with strong GPU acceleration