[Hybrid Parallel] Add model parallel support in dygraph #32248
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for your contribution! |
sandyhouse
requested changes
Apr 16, 2021
| message HybridConfig { | ||
| optional int32 num_data_parallel = 1 [ default = -1 ]; | ||
| optional int32 num_model_parallel = 2 [ default = 1 ]; | ||
| optional int32 num_pipeline_parallel = 3 [ default = 1 ]; |
There was a problem hiding this comment.
Use dp_degree, mp_degree and pp_degree to be consistent with static graph.
| @@ -21,6 +21,7 @@ | |||
| from .data_generator import MultiSlotDataGenerator, MultiSlotStringDataGenerator | |||
| from . import metrics | |||
| from .base.topology import CommunicateTopology, HybridCommunicateGroup | |||
| from .mpu import random, layers | |||
| "num_data_parallel": 1, | ||
| "num_model_parallel": 2, | ||
| "num_pipeline_parallel": 1} | ||
| """ |
There was a problem hiding this comment.
use dp_degree, mp_degree and pp_degree
| check_configs_key(self.strategy.model_parallel_configs, configs, | ||
| "model_parallel_configs") | ||
| assign_configs_value(self.strategy.model_parallel_configs, configs) | ||
|
|
There was a problem hiding this comment.
delete unnecessary model_parallel_configs in this pr.
|
|
||
| self._topology = tp.CommunicateTopology( | ||
| hybrid_group_names=["data", "model", "pipe"], | ||
| dims=[self.dp_num, self.mp_num, self.pp_num]) |
There was a problem hiding this comment.
use ['data', 'pipe', 'model'] instead.
| per_part_size += 1 # make the last row as the padding index | ||
| self.per_part_size = per_part_size | ||
|
|
||
| with get_rng_state_tracker().rng_state(): |
| self._weight_attr = weight_attr | ||
| self._dtype = self._helper.get_default_dtype() | ||
|
|
||
| with get_rng_state_tracker().rng_state(): |
|
|
||
| self.input_size_per_partition = in_features // self.world_size | ||
|
|
||
| with get_rng_state_tracker().rng_state(): |
XieYunshen
previously approved these changes
Apr 16, 2021
XieYunshen
approved these changes
Apr 16, 2021
fuyinno4
approved these changes
Apr 16, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR types
New features
PR changes
APIs
Describe
[Hybrid Parallel] Add model parallel support in dygraph,