Comments (4)
Will do further investigation later :)
As for the IO issue, I remember I have seen in somewhere that a thread block would instinctively load nearby memory whatever it is used or not. Have you ever tried using (N, U, T, V) layout instead of (N, T, U, V)? With the former's (and especially when gather=True), a warp (also a thread block) is able to load a chunk of consecutive memory and reuse it.
Indeed, I've been using the compact version loss function in our speech recognition tasks for a while. It should be technically correct (it's in my dev branch now, the main branch hasn't been updated for some time). I'll finish some merge from my dev to the main branch, and once it's finished, I would reopen the MR.
from warp-rnnt.
I've been following the fast_rnnt work for a while, but haven't make a successful pruned rnn-t training yet.
They also have a paper about the implementation. https://arxiv.org/pdf/2206.13236.pdf
from warp-rnnt.
Hello Huahuan Zheng, interesting theory! But I don't think it will be useful in practice. Optimising a forward pass doesn't make sense. Your can check the cuda profiler logs. The big issue is memory IO, and I really like your previous MR with compact memory version. I wish to finish reviewing it and reopen your MR in near feature.
from warp-rnnt.
Iām not familiar with memory manager for cuda threads. But you right, having TxU matrix is the main bottleneck. Fortunately, there is solution for this, fast_rnnt. It looks really promising.
from warp-rnnt.
Related Issues (20)
- WARNING: sample 0 [42, 26] has a forward/backward mismatch -52.543503 / 0.000000 HOT 3
- ImportError: libcudart.so.10.2: cannot open shared object file: No such file or directory HOT 3
- Transducer loss leads to memory leak HOT 4
- RuntimeError: rnnt_loss status 1 HOT 5
- warning that forward/backward mismatch HOT 3
- question about the gather arguments HOT 1
- Question about average_frames and reduction parmas HOT 1
- Not support for pytorch 1.7 HOT 1
- ninja: build stopped: subcommand failed. HOT 1
- Normalize the RNN-T Loss with input seq length HOT 1
- PyTorch 1.9 Support HOT 5
- Strange behavior using PyTorch DDP HOT 7
- undefined symbol: _ZNSt19basic_ostringstreamIcSt11char_traitsIcESaIcEEC1Ev HOT 2
- rnnt_loss status 1 HOT 2
- THC/THC.h: No such file or directory HOT 7
- can't install warp-rnnt HOT 2
- Exception: CPU version is not implemented HOT 2
- ImportError: libcudart.so.10.1: cannot open shared object file: No such file or directory HOT 2
- __version__ assignment breaks local build.
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
š Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. ššš
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ā¤ļø Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from warp-rnnt.