REPOGEO 报告 · LITE
Tencent/AngelSlim
默认分支 main · commit f76d316c · 扫描时间 2026/5/7 23:02:19
星标 1,031 · Fork 106
行动计划告诉你下一步要做什么——按影响力排序、可直接复制粘贴的修改。品类可见性是真正的 GEO 测试:当用户向 AI 提一个不带品牌、本应让 Tencent/AngelSlim 浮出水面的问题时,AI 是真的推荐了你,还是推荐了你的竞品?客观检查验证 AI 引擎最先权衡的那些元数据信号。自指检查判断 AI 是否还认识你的名字。
行动计划 — 可复制粘贴的修复
3 条由 gemini-2.5-flash 生成、按优先级排序的修改。修完后请把对应条目标记为完成。
- highreadme#1Add a direct opening sentence to the README
原因:
当前English | [简体中文](README_cn.md) <p align="center"> <picture> <source media="(prefers-color-scheme: dark)" srcset="./docs/source/assets/logos/angelslim_logo_light.png"> </picture> </p> <h3 align="center"> A more accessible, comprehensive, and efficient toolkit for large model compression. </h3>复制粘贴的修复English | [简体中文](README_cn.md) Tencent/AngelSlim is a comprehensive toolkit for large model compression, focusing on quantization and efficiency for LLMs and VLMs. <p align="center"> <picture> <source media="(prefers-color-scheme: dark)" srcset="./docs/source/assets/logos/angelslim_logo_light.png"> </picture> </p> <h3 align="center"> A more accessible, comprehensive, and efficient toolkit for large model compression. </h3> - hightopics#2Expand topics with broader model optimization terms
原因:
当前audio, deepseek, dflash, diffusion, eagle, fp4, hunyuan, llm, llm-compression, quantization, qwen, speculative-decoding, vlm
复制粘贴的修复audio, deepseek, dflash, diffusion, eagle, fp4, hunyuan, llm, llm-compression, quantization, qwen, speculative-decoding, vlm, model-optimization, deep-learning-optimization, ai-acceleration, inference-optimization, model-quantization
- mediumreadme#3Clarify the project's license in the README
原因:
复制粘贴的修复## License This project is licensed under the terms specified in the [LICENSE](LICENSE) file.
本次扫描解析到的品类 GEO 通道:google/gemini-2.0-flash-001, deepseek/deepseek-chat
品类可见性 — 真正的 GEO 测试
向 google/gemini-2.0-flash-001 提出的不带品牌问题。AI 推荐了你,还是推荐了别人?
各模型使用同一组问题 — 切换标签对比回答与排名。
- TensorRT · 被推荐 2 次
- microsoft/onnxruntime · 被推荐 1 次
- huggingface/optimum · 被推荐 1 次
- GPTQ · 被推荐 1 次
- neuralmagic/sparseml · 被推荐 1 次
- 品类问题How can I reduce the memory footprint and inference latency of large language models?你:未被推荐AI 推荐顺序:
- TensorRT
- ONNX Runtime (microsoft/onnxruntime)
- Optimum (huggingface/optimum)
- GPTQ
- SparseML (neuralmagic/sparseml)
- Neural Magic DeepSparse Engine (neuralmagic/deepsparse)
- TensorFlow (tensorflow/tensorflow)
- PyTorch (pytorch/pytorch)
- DeepSpeed (microsoft/deepspeed)
- Megatron-LM (NVIDIA/Megatron-LM)
- PyTorch DistributedDataParallel (DDP)
- PEFT library from Hugging Face (huggingface/peft)
- NVIDIA GPUs
- Google TPUs
- AWS Inferentia
- AWS Trainium
- Habana Gaudi
- FlashAttention (Dao-AILab/flash-attention)
- Linear Attention
- NVIDIA Triton Inference Server (triton-inference-server/server)
AI 推荐了 20 个替代方案,却始终没点名 Tencent/AngelSlim。这就是要补上的差距。
查看 AI 完整回答
- 品类问题Seeking a comprehensive toolkit for quantizing large models to improve deployment efficiency.你:未被推荐AI 推荐顺序:
- ONNX Runtime
- TensorRT
- PyTorch (torch.quantization)
- TensorFlow Model Optimization Toolkit
- Intel Neural Compressor
- Optimum (Hugging Face)
- TVM
AI 推荐了 7 个替代方案,却始终没点名 Tencent/AngelSlim。这就是要补上的差距。
查看 AI 完整回答
客观检查
针对 AI 引擎最看重的元数据信号的规则审计。
- Metadata completenesspass
- README presencepass
自指检查
当被直接问到你时,AI 是否还知道你的仓库存在?
- Compared to common alternatives in this category, what is the core differentiator of Tencent/AngelSlim?passAI 明确点名了 Tencent/AngelSlim
AI 的回答可能信誓旦旦却是错的。请按事实核对:技术栈、目标人群、差异化点是不是和你实际的对得上?
- If a team adopts Tencent/AngelSlim in production, what risks or prerequisites should they evaluate first?passAI 明确点名了 Tencent/AngelSlim
AI 的回答可能信誓旦旦却是错的。请按事实核对:技术栈、目标人群、差异化点是不是和你实际的对得上?
- In one sentence, what problem does the repo Tencent/AngelSlim solve, and who is the primary audience?passAI 明确点名了 Tencent/AngelSlim
AI 的回答可能信誓旦旦却是错的。请按事实核对:技术栈、目标人群、差异化点是不是和你实际的对得上?
嵌入你的 GEO 徽章
把这个徽章贴进 Tencent/AngelSlim 的 README。每次重新扫描都会自动更新,并跳到最新报告——是「我在乎 AI 可发现性」最简单的公开证明。
[](https://repogeo.com/zh/r/Tencent/AngelSlim)<a href="https://repogeo.com/zh/r/Tencent/AngelSlim"><img src="https://repogeo.com/badge/Tencent/AngelSlim.svg" alt="RepoGEO" /></a>订阅 Pro,解锁深度诊断
Tencent/AngelSlim — 轻量扫描仍免费;本卡列出 Pro 相对轻量的深度额度。
- 深度报告每月 10 次
- 无品牌品类查询5,轻量 2
- 优先行动项8,轻量 3