parent
1c5616bcbf
commit
285e7ac531
@ -1,50 +0,0 @@
|
||||
# Design Doc: Lookup Remote Table while Distributed training
|
||||
|
||||
## Abstract
|
||||
|
||||
We propose an approach to pre-fetch the parameters from a Parameter Server while distributed training so that Fluid can train a model with the very large parameter that cannot be stored in one trainer's memory.
|
||||
|
||||
## Background
|
||||
|
||||
For an embedding layer, the trainable parameter may be very large, and it is likely that it may not be able to be stored in one trainer's memory. In Fluid distributed training,
|
||||
the [Distributed Transpiler](./parameter_server.md#distributed-transpiler) would split every parameter into some small parameters that stored on the Parameter Server. Hence, we can pre-fetch the parameter from the specified Parameter Server using the input `Ids`.
|
||||
|
||||
## Design
|
||||
|
||||
Prior to reading this design, it would be useful for the reader to make themselves familiar with Fluid [Distributed Training Architecture](./distributed_architecture.md) and
|
||||
[Parameter Server](./parameter_server.md).
|
||||
|
||||
The execution of `lookup local table` is as follows:
|
||||
|
||||
<img src="src/lookup_local_table.png" width="700" />
|
||||
|
||||
For some cases, the parameter(`weight`) may be very large, such as 10 billion features, the entire
|
||||
data could not be stored in one trainer's memory, so we need to partition this parameter and
|
||||
pre-fetch it at the beginning of each mini-batch, and we call it `lookup remote table`:
|
||||
|
||||
<img src="src/lookup_remote_table.png" width="700" />
|
||||
|
||||
The processing flow of `lookup remote table` is as follows:
|
||||
|
||||
1. partitioned parameter
|
||||
|
||||
<img src="src/split_parameter.png" width="400" />
|
||||
|
||||
- **Distributed Transpiler** would split the large parameters
|
||||
(`weight`) into some partitioned parameters (`weight_0`, `weight_1`, `weight_2`) as shown in the figure above.
|
||||
- We can use some algorithms to distribute the partitioned parameters, such as `round-robin` or `hash-name`.
|
||||
|
||||
1. pre-fetching parameter at the beginning of each mini-batch
|
||||
|
||||
- `prefetch_rpc` operator would prefetch the parameter from different Parameter
|
||||
Servers using the input `Ids`. We use [SelectedRows](../../../design/selected_rows.md)
|
||||
as the received variable type.
|
||||
- `merge_selected_rows` operator would merge the received parameters into one
|
||||
`SelectedRows` variable.
|
||||
|
||||
## TODO
|
||||
|
||||
- `prefetch_rpc` operator to send rows index and receive SelectedRows variables.
|
||||
- `lookup_table` need to support `SelectedRows` variable type as input `Weight`.
|
||||
- Async Update, To avoid slow-node, Async update is important for distributed training,
|
||||
we need a design doc and implement it in future.
|
Binary file not shown.
After Width: | Height: | Size: 317 KiB |
Binary file not shown.
Before Width: | Height: | Size: 297 KiB |
Binary file not shown.
Before Width: | Height: | Size: 284 KiB |
Binary file not shown.
Before Width: | Height: | Size: 77 KiB |
Loading…
Reference in new issue