feature(tj): addd monitoring for the gradient conflict metric of MoE in ScaleZero #416
Conversation
This file contains hidden or 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
puyuan1996
added
research
puyuan1996
changed the base branch from
dev-uz-mt-balance
to
dev-multitask-balance-clean
puyuan1996
reviewed
Sep 18, 2025
| from tensorboardX import SummaryWriter | ||
|
|
||
| from lzero.entry.utils import log_buffer_memory_usage, TemperatureScheduler | ||
| # 添加性能监控相关导入 |
There was a problem hiding this comment.
改一下注释为英文，统一用类似DI-engine/ding/model/common/encoder.py at main · opendilab/DI-engine这样的规范格式
There was a problem hiding this comment.
你只改相对dev-multitask-balance-clean分支的moe grad改动这块哈，dev-multitask-balance-clean分支内部的注释规范我去修改哈
puyuan1996
changed the title
puyuan1996
requested changes
Sep 18, 2025
| @@ -0,0 +1,2207 @@ | |||
| diff --git a/lzero/entry/train_unizero_multitask_segment_ddp.py b/lzero/entry/train_unizero_multitask_segment_ddp.py | |||
|
|
||
| import sys | ||
| import os | ||
| PROJECT_ROOT = os.path.abspath("/fs-computility/niuyazhe/tangjia/github/LightZero") # 或者直接写死路径 |
There was a problem hiding this comment.
这个是必要的吗 如果是写成相对路径，不要带有个人信息
| # ------------------------------------------------------------ | ||
| # MOE专家选择统计相关函数 | ||
| # ------------------------------------------------------------ | ||
| def merge_expert_stats_across_ranks(all_expert_stats): |
| _GLOBAL_HEATMAP_FIG.set_size_inches(figsize) | ||
| return _GLOBAL_HEATMAP_FIG, _GLOBAL_HEATMAP_AX | ||
|
|
||
| def create_heatmap_with_values_fast(matrix, task_ids, title="Task-Expert Selection Frequencies"): |
| 'wasserstein_stats': wasserstein_stats | ||
| } | ||
|
|
||
| def create_similarity_heatmap_no_diagonal(similarity_matrix, task_ids, metric_name, title_suffix=""): |
There was a problem hiding this comment.
工具函数已经放在utils里面了，就把lzero/entry/train_unizero_multitask_segment_ddp.py里面没用的删除了哈
| betas=(0.9, 0.95), | ||
| ) | ||
|
|
||
| # self.a=1 |
| # "max_moe_layer_grad_conflict", | ||
| ] | ||
|
|
||
| # # # If the model uses MoE, add expert gradient conflict variables |
| import os | ||
| PROJECT_ROOT = os.path.abspath("/fs-computility/niuyazhe/tangjia/github/LightZero") # 或者直接写死路径 | ||
| sys.path.insert(0, PROJECT_ROOT) | ||
| # /fs-computility/niuyazhe/tangjia/github/LightZero/zoo/atari/config/atari_unizero_multitask_segment_ddp_config_debug.py |
|
|
||
| # 手动销毁进程组 | ||
| # 手动销毁进程组 /fs-computility/niuyazhe/tangjia/github/LightZero/zoo/atari/config/atari_unizero_multitask_segment_ddp_config_debug.py | ||
| if dist.is_initialized(): |
There was a problem hiding this comment.
config中只保留我们scalezero论文里面用到分析的吧
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.
No description provided.