kali20gakki/msAgent / skills/quantizer/quantization-expert-experience-tuning-rules
quantization-expert-experience-tuning-rules
Verified量化专家经验调优库:按「L1 通用专家调优意见 → L2 结构化专家调优意见 → L3 模型专属专家调优意见」三级递进, 输出量化精度调优的完整手段(离群值抑制 / 量化方法·粒度·对称性选型 / 校准集调整 / 敏感层回退 / 模型专属策略), 每条结论附专家原因分析与专家意见可信度等级。 本 Skill 是专家经验库,负责说明可选离群值抑制算法并询问用户选择哪些算法,以及回答「怎么调、哪些层需要回退及为什么」;不运行离群值抑制 processor,不修改 YAML、不执行量化、不做 EP 检查 / 服务化 / 任务评测。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 10.8 KB |
| cross-model-pitfalls.md | markdown | 8.3 KB |
| models/README.md | markdown | 1.3 KB |
| models/deepseek/deepseek-v3.md | markdown | 2.9 KB |
| models/deepseek/deepseek-v4.md | markdown | 2.5 KB |
| models/glm/glm-4-moe.md | markdown | 1.3 KB |
| models/glm/glm-5.md | markdown | 1.4 KB |
| models/minimax/minimax-m.md | markdown | 1.6 KB |
| scripts/README.md | markdown | 3.1 KB |
| scripts/apply_rollback.py | script | 10.1 KB |
| scripts/inspect_model_structure.py | script | 7.1 KB |
| scripts/yaml_utils.py | script | 4.2 KB |
| structure-family-pitfalls.md | markdown | 14.2 KB |
models/minimax/minimax-m.md
View on GitHub# MiniMax-M2.7 / M3 专属经验
> 层级:L3。结构模型类背景同 L2「MoE」「自定义 modeling」。本节记该模型独有坑。
## 1. `block_sparse_moe.experts` 命名(M2.7 独有)
- M2.7 的专家命名用 `*block_sparse_moe.experts*`(块稀疏 MoE),非标准 `*mlp.experts*`;若照搬 L2 常规前缀会匹配不到层。
- 原因:M2.7 采用了自定义 modeling 实现,模块命名约定与标准 transformers 不同,必须取真实 `named_modules` 确认。专家意见可信度:高。
## 2. W8A8 mxfp8 路径(M2.7)
- `quarot(export_extra_info: True)` + `linear_quant`(mxfp8 per_block + minmax),include 仅 `*block_sparse_moe.experts*` 与 `*self_attn*`,其余不量化。
- 原因:MXFP8 per_block 粒度在块内分布更集中,适合该模型稳定低比特;仅量化 experts 与 attention 主体,其他结构保持浮点以保精度。专家意见可信度:中。
## 3. W8A8 int 路径(M3)
- `iter_smooth`(alpha 0.5、scale_min 1e-5、symmetric True、enable `norm-linear`)+ 全局 `w8a8_dynamic`。
- 大范围排除:`*norm*`/`*embed_tokens*`/`*lm_head*`/`*vision_tower*`/`*multi_modal_projector*`/`*patch_merge_mlp*`/`*mlp.down_proj*`/`*shared_experts.down_proj*`/`*mlp.gate`/`*indexer*`/`*index_*`。
- 说明 M3 是 VLM 且含 indexer/跨模态结构,回退面较大,属该模型类「先保 LLM 主体、大量外围排除」的实践。原因:跨模态投影与索引结构连接不同模态空间,量化误差易破坏模态对齐,因此初版先全部排除,仅量化 LLM 主体。专家意见可信度:中。Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |