初始化项目,由ModelHub XC社区提供模型
Model: RedHatAI/Qwen3-14B-FP8-dynamic Source: Original Platform
This commit is contained in:
11
recipe.yaml
Normal file
11
recipe.yaml
Normal file
@@ -0,0 +1,11 @@
|
||||
quant_stage:
|
||||
quant_modifiers:
|
||||
QuantizationModifier:
|
||||
ignore: [lm_head]
|
||||
sequential_targets: [Qwen3DecoderLayer]
|
||||
config_groups:
|
||||
group0:
|
||||
targets: [Linear]
|
||||
weights: {num_bits: 8, type: float, symmetric: true, strategy: channel, observer: mse}
|
||||
input_activations: {num_bits: 8, type: float, symmetric: true, strategy: token,
|
||||
dynamic: true, observer: memoryless}
|
||||
Reference in New Issue
Block a user