feat: Add deepseek flops tracker#1250
Conversation
Signed-off-by: Guyue Huang <guyueh@nvidia.com>
📝 WalkthroughWalkthroughAdds NCCL debug environment variables during MegatronPolicyWorker initialization. Extends FLOPS tracking to support DeepSeekV3Config by importing the config and formula symbol, and adding a new branch in convert_config_to_flops_config to construct a DeepSeekV3-specific FLOPSConfig. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor Caller
participant FT as flops_tracker.convert_config_to_flops_config
participant HF as transformers.Config (Qwen2/Qwen3/Llama/DeepSeekV3)
participant FF as flops_formulas
Caller->>FT: convert_config_to_flops_config(config)
FT->>HF: inspect type(config)
alt DeepSeekV3Config (new)
FT->>FT: build FLOPSConfig (DeepSeekV3 fields)
FT->>FF: select formula: deepseekv3
else Qwen2/Qwen3/Qwen3Moe/Llama
FT->>FT: build corresponding FLOPSConfig
FT->>FF: select respective formula
else Unsupported
FT-->>Caller: raise ValueError
end
FT-->>Caller: (FLOPSConfig, formula)
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Actionable comments posted: 1
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
nemo_rl/models/policy/megatron_policy_worker.py(1 hunks)nemo_rl/utils/flops_tracker.py(2 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Follow the Google Python Style Guide for all Python code
Target Python 3.12+ for all Python code in NeMo-RL
Indent Python code with 4 spaces; do not use tabs
Python filenames should be snake_case (e.g., some_file.py)
Class names should be PascalCase
Function and method names should be snake_case
Local variable names should be snake_case; if starting with a number, prefix with k (e.g., k_99th_percentile)
Global variables should be UPPER_SNAKE_CASE and prefixed with G_ (e.g., G_MY_GLOBAL)
Constants should be UPPER_SNAKE_CASE
Avoid shadowing variables declared in an outer scope
Initialize all externally visible members of a class in the constructor
For public interfaces used outside a file, prefer docstrings over comments
Use comments mainly for code within a function or interfaces local to a file
Commented-out code must include a nearby comment explaining usage and why it is commented out; otherwise remove before merging
Use Google-style docstrings for classes and functions (Sphinx-parseable)
Avoid using reflection when functionality can be easily achieved without it
Limit except clauses to the smallest specific set of exceptions possible
For duck-typing via try/except, keep the try body minimal and use else for main logic
Add the NVIDIA copyright header (with current year) at the top of all Python files, excluding tests/ and test-only scripts
Files:
nemo_rl/models/policy/megatron_policy_worker.pynemo_rl/utils/flops_tracker.py
nemo_rl/**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
nemo_rl/**/*.py: Do not set non-None configuration defaults in code; YAML is the single source of truth for defaults
Access required config attributes directly (e.g., policy_cfg["precision"]) and assume presence; do not introduce hidden defaults
Express configuration optionality via TypedDict using typing.NotRequired
When adding a new config key to a TypedDict subclass, document the key’s purpose, valid values/types, and recommended default in code
For any class or function decorated with @ray.remote, add '# pragma: no cover' on the class/def line (and on remote functions)
Files:
nemo_rl/models/policy/megatron_policy_worker.pynemo_rl/utils/flops_tracker.py
🧬 Code graph analysis (1)
nemo_rl/utils/flops_tracker.py (2)
nemo_rl/models/policy/utils.py (1)
sliding_window_overwrite(169-195)nemo_rl/utils/flops_formulas.py (2)
FLOPSConfig(21-59)deepseekv3(386-458)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: Lint check
- GitHub Check: Post submodule check comment / Comment on PR
- GitHub Check: Post automodel integration comment / Comment on PR
terrykong
left a comment
There was a problem hiding this comment.
awesome. thanks @guyueh1 !
could you add a unit test here https://github.com/NVIDIA-NeMo/RL/blob/main/tests/unit/utils/test_flops_counter.py#L23-L30 ?
Signed-off-by: Guyue Huang <140554423+guyueh1@users.noreply.github.com>
Signed-off-by: Guyue Huang <guyueh@nvidia.com>
|
@terrykong added test for this |
|
closing in favor of #1305 |
Pull request was closed
…NeMo#1305) Signed-off-by: Guyue Huang <guyueh@nvidia.com> Signed-off-by: Guyue Huang <140554423+guyueh1@users.noreply.github.com>
…NeMo#1305) Signed-off-by: Guyue Huang <guyueh@nvidia.com> Signed-off-by: Guyue Huang <140554423+guyueh1@users.noreply.github.com> Signed-off-by: yuanhangs <yuanhangs@nvidia.com>
What does this PR do ?
Add deepseek v3 flops tracker
Issues
List issues that this PR closes (syntax):
Usage
# Add a code snippet demonstrating how to use thisBefore your PR is "Ready for review"
Pre checks:
Additional Information
Summary by CodeRabbit
New Features
Chores