初始化项目,由ModelHub XC社区提供模型

Model: RedHatAI/Qwen2.5-0.5B-quantized.w8a8
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-04-28 14:42:56 +08:00
commit 0c98de33a8
22 changed files with 365 additions and 0 deletions

18
recipe.yaml Normal file
View File

@@ -0,0 +1,18 @@
quant_stage:
quant_modifiers:
SmoothQuantModifier:
smoothing_strength: 0.9
mappings:
- - ['re:.*q_proj', 're:.*k_proj', 're:.*v_proj']
- re:.*input_layernorm
- - ['re:.*gate_proj', 're:.*up_proj']
- re:.*post_attention_layernorm
- - ['re:.*down_proj']
- re:.*up_proj
GPTQModifier:
sequential_update: true
dampening_frac: 0.1
ignore: [lm_head]
scheme: W8A8
targets: Linear
observer: mse