Upload folder using huggingface_hub

config.json

@@ -28,7 +28,7 @@
           "Linear"
         ],
         "weights": {
-          "actorder": null,
+          "actorder": "group",
           "block_structure": null,
           "dynamic": false,
           "group_size": 128,
@@ -42,7 +42,7 @@
       }
     },
     "format": "pack-quantized",
-    "global_compression_ratio": 1.883165566487463,
+    "global_compression_ratio": 1.8833636632816793,
     "ignore": [
       "lm_head"
     ],
@@ -51,7 +51,7 @@
     "quantization_status": "compressed",
     "sparsity_config": {
       "format": "dense",
-      "global_sparsity": 0.14375328644374685,
+      "global_sparsity": 0.1407252015594666,
       "ignore": [],
       "registry_requires_subclass": false,
       "sparsity_structure": "unstructured",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:892873b79abcf66508b526a74179177e303de9b4b1464a6177096348faa7fd76
-size 761968800
+oid sha256:a526374315ca3ab080a4c930dac9144a8a2a738a69ba9446ef6027e9fd2ab1e9
+size 763563248

recipe.yaml

@@ -1,6 +1,9 @@
-DEFAULT_stage:
-  DEFAULT_modifiers:
+quant_stage:
+  quant_modifiers:
     GPTQModifier:
-      targets: Linear
       ignore: [lm_head]
-      scheme: W4A16
+      config_groups:
+        group_0:
+          weights: {num_bits: 4, type: int, symmetric: true, strategy: group, group_size: 128,
+            actorder: group}
+          targets: [Linear]