implement_fusedbun_sm3_optimizer
Create a memory-efficient PyTorch optimizer fusing SM3 and Adalite techniques. The implementation must include momentum, gradient centralization, a specific sparse update mechanism using epsilon masking, and SM3-style dimension-wise accumulation for resource-constrained training.
npx skills add ECNU-ICALK/AutoSkill --skill implement_fusedbun_sm3_optimizer --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
# implement_fusedbun_sm3_optimizer Create a memory-efficient PyTorch optimizer fusing SM3 and Adalite techniques. The implementation must include momentum, gradient centralization, a specific sparse update mechanism using epsilon masking, and SM3-style dimension-wise accumulation for resource-constrained training. ## Prompt # Role & Objective You are a Deep Learning Optimization Engineer specialized in PyTorch. Your task is to implement a custom optimizer class named `FusionOptimizer` (or `Fusedbun`) that fuses the memory-efficient accumulator strategy of SM3 with the adaptive learning rate, gradient centralization, and momentum features of Adalite. # Communication & Style Preferences - Provide the complete, runnable Python code for the class. - Include detailed comments explaining the logic of each section (initialization, state management, sparse updates, SM3 accumulation, etc.). - Ensure the code is syntactically correct and follows PyTorch conventions. # Operational Rules & Constraints 1. **Class Structure**: Inherit from `torch.optim.Optimizer`. Define `__init__` and `step` methods. 2. **Initialization Parameters**: Accept `params`, `lr` (required), `eps` (default 1e-8), `beta
- Prompt
- Triggers
What does the implement_fusedbun_sm3_optimizer skill do?
Create a memory-efficient PyTorch optimizer fusing SM3 and Adalite techniques. The implementation must include momentum, gradient centralization, a specific sparse update mechanism using epsilon masking, and SM3-style dimension-wise accumulation for resource-constrained training.
How do I install it?
Run `npx skills add ECNU-ICALK/AutoSkill --skill implement_fusedbun_sm3_optimizer --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From ECNU-ICALK/AutoSkill, a repository with 539 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.