Add new LLM models to MNN by downloading, analyzing, mapping, and exporting.
npx skills add https://github.com/alibaba/MNN --skill support-new-llm
Please help me install this Agent Skill. Skill: support-new-llm Source: https://github.com/alibaba/MNN/tree/main/skills/support-new-llm Command: npx skills add https://github.com/alibaba/MNN --skill support-new-llm
This Skill automates the complex process of adding support for new Large Language Models (LLMs) to the MNN framework, enabling broader model compatibility and faster integration.
Use the support-new-llm skill to add support for the 'google/gemma-2-9b-it' model.
High-intent search queries and answers about installing and using this skill.
Adding a new LLM to MNN involves downloading the model from HuggingFace or ModelScope, mapping its architecture to MNN's internal representation, running alignment tests, and exporting the validated MNN model.
Yes, you can integrate HuggingFace models into MNN for on-device inference by downloading the model, analyzing its architecture, mapping fields to MNN's representation, and exporting the validated MNN model.
Yes, the MNN model integration process supports multimodal architectures. It analyzes custom model architectures and maps them to MNN's internal representation during the export workflow.
The best way to ensure accuracy when exporting LLMs to MNN is to follow a Test-Driven Development methodology. This approach validates each step through alignment tests before proceeding to the final model export.
Yes, alignment tests are required for every step of MNN model integration. The workflow follows a strict Test-Driven Development approach to validate each stage before proceeding to the final export.
PaddlePaddleAutomate adding inference deployment support for open-source LLMs to FastDeploy.
maziyarpanahiLoads OpenMed NER models for reuse in offline clinical text processing workflows.
alibabaGuide adding new operators to the MNN framework with TDD steps.
erinclemmerRegister HuggingFace transformer architectures for Language Pipes inference.
