Files
Qwen2.5-VL-3B-Instruct-FP8-…/recipe.yaml
ModelHub XC c3e36c1725 初始化项目,由ModelHub XC社区提供模型
Model: neuralmagic/Qwen2.5-VL-3B-Instruct-FP8-Dynamic
Source: Original Platform
2026-10-23 02:01:14 +08:00

7 lines
158 B
YAML

DEFAULT_stage:
DEFAULT_modifiers:
QuantizationModifier:
ignore: ['re:.*lm_head', 're:visual.*']
targets: [Linear]
scheme: FP8_DYNAMIC