Why is autograd so complicated

PyTorch Developer Podcast

Player FM - Internet Radio Done Right

32 subscribers

اضافه شده در four سال پیش

محتوای ارائه شده توسط PyTorch, Edward Yang, and Team PyTorch. تمام محتوای پادکست شامل قسمت‌ها، گرافیک‌ها و توضیحات پادکست مستقیماً توسط PyTorch, Edward Yang, and Team PyTorch یا شریک پلتفرم پادکست آن‌ها آپلود و ارائه می‌شوند. اگر فکر می‌کنید شخصی بدون اجازه شما از اثر دارای حق نسخه‌برداری شما استفاده می‌کند، می‌توانید روندی که در اینجا شرح داده شده است را دنبال کنید.https://fa.player.fm/legal

Silver Linings with The Old Gays

1
Our Chosen Family: "The gay community is much bolder today." 33:19

18 روز پیش33:19

پخش در آینده

لیست ها

پسندیدن

دوست داشته شد

33:19

What’s the secret to lasting friendships? How does queer community show up through the ebbs and flows of life? And what’s the REAL story behind the “YMCA” song? In the first episode of Silver Linings, The Old Gays dive into an essential part of queer life: chosen family. They discuss the vital love, support, and sense of belonging that community provides, especially during life's toughest moments. They open up about what “queer” means to them, how chosen family has impacted their lives, and how to maintain close bonds over time–including their love for each other! “We’ve come a long way, baby.” Family isn’t just what you’re born with; it’s the people who show up, shape you, and stick around. See omnystudio.com/listener for privacy information.…

۴ سال پیش 15:45

MP3•خانه قسمت

Why is autograd so complicated? What are the constraints and features that go into making it complicated? What's up with it being written in C++? What's with derivatives.yaml and code generation? What's going on with views and mutation? What's up with hooks and anomaly mode? What's reentrant execution? Why is it relevant to checkpointing? What's the distributed autograd engine?

Further reading.

Autograd notes in the docs https://pytorch.org/docs/stable/notes/autograd.html
derivatives.yaml https://github.com/pytorch/pytorch/blob/master/tools/autograd/derivatives.yaml
Paper on autograd engine in PyTorch https://openreview.net/pdf/25b8eee6c373d48b84e5e9c6e10e7cbbbce4ac73.pdf

83 قسمت

#Tech #PyTorch #Edward Yang #Team PyTorch #Deep Learning #Machine Learning

Why is autograd so complicated

PyTorch Developer Podcast

32 subscribers

published ۴ سال پیش

اشتراک گذاری

MP3•خانه قسمت

Further reading.

Autograd notes in the docs https://pytorch.org/docs/stable/notes/autograd.html
derivatives.yaml https://github.com/pytorch/pytorch/blob/master/tools/autograd/derivatives.yaml
Paper on autograd engine in PyTorch https://openreview.net/pdf/25b8eee6c373d48b84e5e9c6e10e7cbbbce4ac73.pdf

83 قسمت

#Tech #PyTorch #Edward Yang #Team PyTorch #Deep Learning #Machine Learning

همه قسمت ها

1
Compiler collectives 16:33

43 weeks پیش16:33

16:33

Compiler collectives are a PT2 feature where by compiler instances across multiple ranks use NCCL collectives to communicate information to other instances. This is used to ensure we consistently decide if inputs or static or dynamic across all ranks. See also PR at https://github.com/pytorch/pytorch/pull/130935…

1
TORCH_TRACE and tlparse 15:28

1 year پیش15:28

15:28

TORCH_TRACE and tlparse are a structured log and log parser for PyTorch 2. It gives useful information about what code was compiled and what the intermediate build products look like.

1
Higher order operators 17:10

1 year پیش17:10

17:10

Higher order operators are a special form of operators in torch.ops which have relaxed input argument requirements: in particular, they can accept any form of argument, including Python callables. Their name is based off of their most common use case, which is to represent higher order functions like control flow operators. However, they are also used to implement other variants of basic operators and can also be used to smuggle in Python data that is quite unusual. They are implemented using a Python dispatcher.…

1
Inductor - Post-grad FX passes 24:07

1 year پیش24:07

24:07

The post-grad FX passes in Inductor run after AOTAutograd has functionalized and normalized the input program into separate forward/backward graphs. As such, they generally can assume that the graph in question is functionalized, except for some mutations to inputs at the end of the graph. At the end of post-grad passes, there are special passes that reintroduce mutation into the graph before going into the rest of Inductor lowering which is generally aware of passes. The post-grad FX passes are varied but are typically domain specific passes making local changes to specific parts of the graph.…

1
CUDA graph trees 20:50

1 year پیش20:50

20:50

CUDA graph trees are the internal implementation of CUDA graphs used in PT2 when you say mode="reduce-overhead". Their primary innovation is that they allow the reuse of memory across multiple CUDA graphs, as long as they form a tree structure of potential paths you can go down with the CUDA graph. This greatly reduced the memory usage of CUDA graphs in PT2. There are some operational implications to using CUDA graphs which are described in the podcast.…

1
Min-cut partitioner 15:56

1 year پیش15:56

15:56

The min-cut partitioner makes decisions about what to save for backwards when splitting the forward and backwards graph from the joint graph traced by AOTAutograd. Crucially, it doesn't actually do a "split"; instead, it is deciding how much of the joint graph should be used for backwards. I also talk about the backward retracing problem.…

1
AOTInductor 17:30

1 year پیش17:30

17:30

AOTInductor is a feature in PyTorch that lets you export an inference model into a self-contained dynamic library, which can subsequently be loaded and used to run optimized inference. It is aimed primarily at CUDA and CPU inference applications, for situations when your model export once to be exported once while your runtime may still get continuous updates. One of the big underlying organizing principles is a limited ABI which does not include libtorch, which allows these libraries to stay stable over updates to the runtime. There are many export-like use cases you might be interested in using AOTInductor for, and some of the pieces should be useful, but AOTInductor does not necessarily solve them.…

1
Tensor subclasses and PT2 13:25

1 year پیش13:25

13:25

Tensor subclasses allow you to add extend PyTorch with new types of tensors without having to write any C++. They have been used to implement DTensor, FP8, Nested Jagged Tensor and Complex Tensor. Recent work by Brian Hirsh means that we can compile tensor subclasses in PT2, eliminating their overhead. The basic mechanism by which this compilation works is a desugaring process in AOTAutograd. There are some complications involving views, dynamic shapes and tangent metadata mismatch.…

1
Compiled autograd 18:07

1 year پیش18:07

18:07

Compiled autograd is an extension to PT2 that permits compiling the entirety of a backward() call in PyTorch. This allows us to fuse accumulate grad nodes as well as trace through arbitrarily complicated Python backward hooks. Compiled autograd is an important part of our plans for compiled DDP/FSDP as well as for whole-graph compilation.…

1
PT2 extension points 15:54

1 year پیش15:54

15:54

We discuss some extension points for customizing PT2 behavior across Dynamo, AOTAutograd and Inductor.

1
Inductor - Define-by-run IR 12:06

1 year پیش12:06

12:06

Define-by-run IR is how Inductor defines the internal compute of a pointwise/reduction operation. It is characterized by a function that calls a number of functions in the 'ops' namespace, where these ops can be overridden by different handlers depending on what kind of semantic analysis you need to do. The ops Inductor supports include regular arithmetic operators, but also memory load/store, indirect indexing, masking and collective operations like reductions.…

1
Unsigned integers 13:07

1 year پیش13:07

13:07

Traditionally, unsigned integer support in PyTorch was not great; we only support uint8. Recently, we added support for uint16, uint32 and uint64. Bare bones functionality works, but I'm entreating the community to help us build out the rest. In particular, for most operations, we plan to use PT2 to build anything else. But if you have an eager kernel you really need, send us a PR and we'll put it in. While most of the implementation was straightforward, there are some weirdnesses related to type promotion inconsistencies with numpy and dealing with the upper range of uint64. There is also upcoming support for sub-byte dtypes uint1-7, and these will exclusively be implemented via PT2.…

1
Inductor - IR 18:00

1 year پیش18:00

18:00

Inductor IR is an intermediate representation that lives between ATen FX graphs and the final Triton code generated by Inductor. It was designed to faithfully represent PyTorch semantics and accordingly models views, mutation and striding. When you write a lowering from ATen operators to Inductor IR, you get a TensorBox for each Tensor argument which contains a reference to the underlying IR (via StorageBox, and then a Buffer/ComputedBuffer) that says how the Tensor was computed. The inner computation is represented via define-by-run, which allows for compact definition of IR representation, while still allowing you to extract an FX graph out if you desire. Scheduling then takes buffers of inductor IR and decides what can be fused. Inductor IR may have too many nodes, this would be a good thing to refactor in the future.…

1
Dynamo - VariableTracker 15:55

1 year پیش15:55

15:55

I talk about VariableTracker in Dynamo. VariableTracker is Dynamo's representation of the Python. I talk about some recent changes, namely eager guards and mutable VT. I also tell you how to find the functionality you care about in VariableTracker ( https://docs.google.com/document/d/1XDPNK3iNNShg07jRXDOrMk2V_i66u1hEbPltcsxE-3E/edit#heading=h.i6v7gqw5byv6 ).…

1
Unbacked SymInts 21:31

۲ years پیش21:31

21:31

This podcast goes over the basics of unbacked SymInts. You might want to listen to this one before listening to https://pytorch-dev-podcast.simplecast.com/episodes/zero-one-specialization Some questions we answer (h/t from Gregory Chanan): - Are unbacked symints only for export? Because otherwise I could just break / wait for the actual size. But maybe I can save some retracing / graph breaks perf if I have them too? So the correct statement is "primarily" for export? - Why am I looking into the broadcasting code at all? Naively, I would expect the export graph to be just a list of ATen ops strung together. Why do I recurse that far down? Why can't I annotate DONT_TRACE_ME_BRO? - How does 0/1 specialization fit into this? I understand we may want to 0/1 specialize in a dynamic shape regime in "eager" mode (is there a better term?), but that doesn't seem to matter for export? - So far we've mainly been talking about how to handle our own library code. There is a worry about pushing complicated constraints downstream, similar to torchscript. What constraints does this actually push?…

PyTorch Developer Podcast

1
Zero-one specialization 21:07

۲ years پیش21:07

21:07

Mikey Dagistes joins me to ask some questions about the recent recent composability sync https://www.youtube.com/watch?v=NJV7YFbtoR4 where we discussed 0/1 specialization and its implications on export in PT2. What's the fuss all about? What do I need to understand about PT2 to understand why 0/1 specialization is a thing?…

PyTorch Developer Podcast

1
torchdynamo 25:35

۲ years پیش25:35

25:35

What is torchdynamo? From a bird's eye view, what exactly does it do? What are some important things to know about it? How does it differ from other graph capture mechanisms? For more reading, check out https://docs.google.com/document/d/13K03JN4gkbr40UMiW4nbZYtsw8NngQwrTRnL3knetGM/edit#

PyTorch Developer Podcast

1
PyTorch 2.0 17:51

۲ years پیش17:51

17:51

Soumith's keynote on PT2.0: https://youtu.be/vbtGZL7IrAw?t=1037 PT2 Manifesto: https://docs.google.com/document/d/1tlgPcR2YmC3PcQuYDPUORFmEaBPQEmo8dsh4eUjnlyI/edit# PT2 Architecture: https://docs.google.com/document/d/1wpv8D2iwGkKjWyKof9gFdTf8ISszKbq1tsMVm-3hSuU/edit#

PyTorch Developer Podcast

1
History of functorch 19:10

۳ years پیش19:10

19:10

Join me with Richard Zou to talk about the history of functorch. What was the thought process behind the creation of functorch? How did it get started? JAX’s API and model is fairly different from PyTorch’s, how did we validate that it would work in PyTorch? Where did functorch go after the early user studies? Where is it going next?…

PyTorch Developer Podcast

1
Learning rate schedulers 19:35

۳ years پیش19:35

19:35

What’s a learning rate? Why might you want to schedule it? How does the LR scheduler API in PyTorch work? What the heck is up with the formula implementation? Why is everything terrible?

PyTorch Developer Podcast

1
Weak references 16:46

۳ years پیش16:46

16:46

What are they good for? (Caches. Private fields.) C++ side support, how it’s implemented / release resources. Python side support, how it’s implemented. Weak ref tensor hazard due to resurrection. Downsides of weak references in C++. Scott Wolchok’s release resources optimization. Other episodes to listen to first: https://pytorch-dev-podcast.simplecast.com/episodes/reference-counting https://pytorch-dev-podcast.simplecast.com/episodes/pyobject-preservation…

PyTorch Developer Podcast

1
Strides 20:31

۳ years پیش20:31

20:31

Mike Ruberry has an RFC about stride-agnostic operator semantics ( https://github.com/pytorch/pytorch/issues/78050 ), so let's talk about strides. What are they? How are they used to implement views and memory format? How do you handle them properly when writing kernels? In what sense are strides overspecified, and therefore, not worth slavishly reimplementing in a system like PrimTorch? What does Edward think we should do about them? My blog post that covers strides along with other topics can be found at http://blog.ezyang.com/2019/05/pytorch-internals/…

PyTorch Developer Podcast

1
AOTAutograd 19:12

۳ years پیش19:12

19:12

AOTAutograd is a cool new feature in functorch for capturing both forward and backward traces of PyTorch operators, letting you run them through a compiler and then drop the compiled kernels back into a normal PyTorch eager program. Today, Horace joins me to tell me how it works, what it is good to use for, and what our future plans for it are.…

PyTorch Developer Podcast

1
Dispatcher questions with Sherlock 18:36

۳ years پیش18:36

18:36

Sherlock recently joined the PyTorch team, having previously worked on ONNX Runtime at Microsoft, and Sherlock’s going to ask me some questions about the dispatcher, and I’m going to answer them. We talked about the history of the dispatcher, how to override dispatching order, multiple dispatch, how to organize various dispatch keys and torch function mode. The companion video is at https://youtu.be/_qB2Ho1O3u4…

PyTorch Developer Podcast

1
New CI 16:12

۳ years پیش16:12

16:12

PyTorch recently moved all of its CI from CircleCI to GitHub Actions. There were a lot of improvements in the process, making my old podcast about CI obsolete! Today, Eli Uriegas joins me to talk about why we moved to GitHub Actions, how the new CI system is put together, and what some cool features about our new CI.…

PyTorch Developer Podcast

1
Python exceptions 14:47

۳ years پیش14:47

14:47

C++ has exceptions, Python has exceptions. But they’re not the same thing! How do exceptions work in CPython, how do we translate exceptions from C++ to Python (hint: it’s different for direct bindings versus pybind11), and what do warnings (which we also translate from C++ to Python) have in common with this infrastructure?…

PyTorch Developer Podcast

1
Torch vs ATen APIs 15:03

۳ years پیش15:03

15:03

PyTorch’s torch API is the Python API everyone knows and loves, but there’s also another API, the ATen API, which most of PyTorch’s internal subsystems are built on. How to tell them apart? What implications do these have on our graph mode IR design? Also, a plug for PrimTorch, a new set of operators, not designed for eager mode, that is supposed to be even lower level than ATen.…

PyTorch Developer Podcast

1
All about NVIDIA GPUs 19:29

۴ years پیش19:29

19:29

PyTorch is in the business of shipping numerical software that can run fast on your CUDA-enabled NVIDIA GPU, but it turns out there is a lot of heterogeneity in NVIDIA’s physical GPU offering and when it comes to what is fast and what is slow, what specific GPU you have on hand matters quite a bit. Yet there are literally hundreds of distinct NVIDIA GPU models on the market, how do you make sense of the madness? Today, Natalia Gimelshein joins me to talk about everything that’s going on in the NVIDIA GPU market, and what, as a framework developer, you have to care about to make sense of it all. Further reading. NVIDIA microarchitectures on Wikipedia https://en.wikipedia.org/wiki/Category:Nvidia_microarchitectures A slightly old post about matching SM to architecture https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/…

PyTorch Developer Podcast

1
Tensor subclasses and Liskov substitution principle 19:13

۴ years پیش19:13

19:13

A lot of recent work going in PyTorch is all about adding new and interesting Tensor subclasses, and this all leads up to the question of, what exactly is OK to make a tensor subclass? One answer to this question comes from an old principle from Barbara Liskov called the Liskov substitution principle, which informally can be stated as S is a subtype of T if anywhere you have T, it can be replaced with S without altering "desirable" properties of this program. In this podcast I'll talk about LSP and how it relates to the design of Tensor subclasses and a hypothetical "abstract Tensor specification" which really doesn't exist but which sort of implicitly exists in the corpus of existing PyTorch programs. Further reading: This is a cool interview with Barbara Liskov that I quote in the podcast https://www.youtube.com/watch?v=-Z-17h3jG0A Max Balandat talking about linear operators in PyTorch https://github.com/pytorch/pytorch/issues/28341 At the end I talk a little bit about multiple dispatch; an earlier discussion about this topic is in this podcast https://pytorch-dev-podcast.simplecast.com/episodes/multiple-dispatch-in-torch-function…

PyTorch Developer Podcast

1
Half precision 18:00

۴ years پیش18:00

18:00

In this episode I talk about reduced precision floating point formats float16 (aka half precision) and bfloat16. I'll discuss what floating point numbers are, how these two formats vary, and some of the practical considerations that arise when you are working with numeric code in PyTorch that also needs to work in reduced precision. Did you know that we do all CUDA computations in float32, even if the source tensors are stored as float16? Now you know! Further reading. The Wikipedia article on IEEE floating point is pretty great https://en.wikipedia.org/wiki/IEEE_754 How bfloat16 works out when doing training https://arxiv.org/abs/1905.12322 Definition of acc_type in PyTorch https://github.com/pytorch/pytorch/blob/master/aten/src/ATen/AccumulateType.h…

به Player FM خوش آمدید!

Player FM در سراسر وب را برای یافتن پادکست های با کیفیت اسکن می کند تا همین الان لذت ببرید. این بهترین برنامه ی پادکست است که در اندروید، آیفون و وب کار می کند. ثبت نام کنید تا اشتراک های شما در بین دستگاه های مختلف همگام سازی شود.