7.4.2 LLM Toolchain

RDK S100 1.0.0 Large Model Toolchain

On the S100/S100P platform, D-Robotics_LLM_S100 currently supports the following models and features:

LLM:

1. DeepSeek-R1-Distill-Qwen: Supports DeepSeek-R1-Distill-Qwen-1.5B and DeepSeek-R1-Distill-Qwen-7B, offering model quantization, simple conversations, multi-turn dialogues, and PPL evaluation.

2. InternLM2: Supports InternLM2-1.8B, providing model quantization, simple conversations, and PPL evaluation.

3. Qwen2.5: Supports Qwen2.5-1.5B, Qwen2.5-7B, Qwen2.5-1.5B-Instruct, and Qwen2.5-7B-Instruct, offering model quantization, simple conversations, multi-turn dialogues (Instruct only), and PPL evaluation.

Multimodal:

1. Qwen2.5-Omni: Supports Qwen2.5-Omni-3B, providing model quantization, offline operation, and online operation.

Download Methods

D-Robotics_LLM_S100 Development Toolkit

wget https://d-robotics-aitoolchain.oss-cn-beijing.aliyuncs.com/llm_s100/1.0.0/D-Robotics_LLM_S100_1.0.0_SDK.tar.gz

D-Robotics_LLM_S100 User Manual

wget https://d-robotics-aitoolchain.oss-cn-beijing.aliyuncs.com/llm_s100/1.0.0/D-Robotics_LLM_S100_1.0.0_Doc.zip

D-Robotics_LLM_S100 Compiled Models

After downloading the development toolkit, refer to the oellm_runtime/model/resolve_model_nash-m.txt file for download links.