| license: apache-2.0 | |
| base_model: Qwen/Qwen2.5-Coder-1.5B-Instruct | |
| tags: | |
| - angular | |
| - lora | |
| - mlx | |
| library_name: mlx | |
| # angular-coder-1.5b-lora | |
| The LoRA adapter behind `bharathmuppa/angular-coder-1.5b`. Apply it to | |
| `Qwen/Qwen2.5-Coder-1.5B-Instruct` with `mlx-lm`: | |
| ```sh | |
| python -m mlx_lm generate --model Qwen/Qwen2.5-Coder-1.5B-Instruct \ | |
| --adapter-path <this repo, downloaded> --prompt "What is new in Angular 20?" | |
| ``` | |
| Rank 16, scale 20, dropout 0.05, last 16 layers, keys: q/k/v/o projections and | |
| gate/up/down MLP projections. See the fused model card for data and training details. | |