初始化项目，由ModelHub XC社区提供模型

Model: RedHatAI/Mistral-Small-24B-Instruct-2501-quantized.w8a8 Source: Original Platform
2026-06-03 19:44:14 +08:00
commit 25f653a739
17 changed files with 497 additions and 0 deletions
--- a/recipe.yaml
+++ b/recipe.yaml
@@ -0,0 +1,20 @@
+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:
+      ignore: [lm_head]
+      dampening_frac: 0.09999999999999999
+      config_groups:
+        group_0:
+          targets: [Linear]
+          weights: {num_bits: 8, type: int, symmetric: true, strategy: channel, observer: mse}
+          input_activations: {num_bits: 8, type: int, symmetric: true, strategy: token, dynamic: true,
+            observer: memoryless}