Spaces:
Sleeping
Sleeping
Download scripts/fuse.sh from zhatrix/logistics-llm: direct link, hf CLI and curl.
- Browser
- Download file 495 Bytes
-
https://huggingface.co/spaces/zhatrix/logistics-llm/resolve/main/scripts/fuse.sh
- Command line
-
hf download hf://spaces/zhatrix/logistics-llm/scripts/fuse.sh
-
curl -L -o fuse.sh https://huggingface.co/spaces/zhatrix/logistics-llm/resolve/main/scripts/fuse.sh
495 Bytes
| # 把 LoRA 合并进基座(可选,用于分发)。 | |
| # ⚠️ 实测直接合并进 4bit 量化基座会让 LoRA 失效,必须 --de-quantize 得到 fp16 模型(约 15GB)。 | |
| set -euo pipefail | |
| cd "$(dirname "$0")/.." | |
| BASE=${BASE_MODEL:-mlx-community/Qwen3-14B-4bit} | |
| uv run python -m mlx_lm fuse --model "$BASE" --adapter-path adapters/logistics-lora --save-path models/logistics-qwen-7b-fp16 --de-quantize | |
| echo "已合并 → models/logistics-qwen-7b-fp16(fp16)" | |