msModelSlim Quick Quant Skill
SkillAI & modelsProvides general quick quantization guidance for msModelSlim, including installation, minimal YAML configuration, and basic execution verification. Use when the user asks about msmodelslim installation, quick quantization, configuring yaml, or basic linear_quant or minmax parameters.
Available today. Use it from your connected AI after setup.
No other account needed.
Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
Then ask your AI: use the msModelSlim Quick Quant Skill skill
What this skill tells your AI
The instructions your AI receives, as published by kali20gakki/msagent in skills/quantizer/msmodelslim-quick-quant/SKILL.md and read by ahel’s review.
适用场景
- 需要安装
msmodelslim - 需要用 YAML 启动基础量化流程
- 仅需基础量化器配置(不涉及复杂策略)
执行规则
- 禁止使用绝对路径或仓库私有路径。
- 统一使用占位符:
${MODEL_PATH}、${SAVE_PATH}、${MODEL_TYPE}、${CONFIG_PATH}。 - YAML 仅介绍
linear_quant基础配置。 - 复杂配置(多阶段流程、MoE 混合策略、复杂 outlier 组合)不在本 Skill 覆盖范围。
- MoE 结构中路由器
gate模块一般不量化,默认应排除。 - 回答保持简洁,只解释必要参数与命令。
最小流程
1) 安装与校验
- 检查 Python 版本(>=3.8)与依赖环境(如 CANN)。
- 按安装文档执行在线/离线/源码安装。
- 安装后验证:
msmodelslim quant --helppython -c "import msmodelslim"
2) 生成最简 YAML
apiversion: "modelslim_v1"
spec:
process:
- type: "linear_quant"
qconfig:
act:
scope: "per_token"
dtype: "int8"
symmetric: false
method: "minmax"
weight:
scope: "per_channel"
dtype: "int8"
symmetric: true
method: "minmax"
include: ["*"]
exclude: ["*.gate"]
save:
- type: "ascendv1_saver"
part_file_size: 4
linear_quant 参数说明见 reference/linear_quant.md。
3) 执行量化
msmodelslim quant \
--model_path ${MODEL_PATH} \
--save_path ${SAVE_PATH} \
--device npu \
--model_type ${MODEL_TYPE} \
--config_path ${CONFIG_PATH} \
--trust_remote_code True
4) 最小验证
- 命令执行无报错退出。
save_path下生成量化结果文件。- 日志中可看到
linear_quant处理过程。
Signals
- GitHub stars
- 31
- Forks
- 8
- Last commit
- Sep 2026
Advanced
- Item type
- skill
- Key
msmodelslim-quick-quant- Source
- github.com/kali20gakki/msagent