TinyDevAssist

A small developer-support model adapted with six LoRA training steps. Its answers may be wrong; check commands before running them. Observed errors include fabricated API calls and incorrect Git commands. The six-prompt comparison does not establish better quality than the base model.

Developer question

Temperature

0 1

Maximum new tokens

16 256

Response

Examples

Base: HuggingFaceTB/SmolLM2-360M-Instruct at a10cc1512eabd3dde888204e902eca88bddb4951. The adapter is served from the configured Hugging Face model repository.