fix: apply proven configs from mlx-community/unsloth references

Fresh weights from full multimodal Google base model.
Configs matched against repos with millions of downloads.

Co-Authored-By: Cladius Maximus <cladius@lethean.io>

README.md

@@ -1,7 +1,9 @@
 ---
-language: en
-pipeline_tag: text-generation
 library_name: mlx
+license: apache-2.0
+license_link: https://ai.google.dev/gemma/docs/gemma_4_license
+pipeline_tag: text-generation
+base_model: google/gemma-4-26b-a4b-it
 tags:
 - mlx
 ---

config.json

@@ -10,11 +10,7 @@
     "eoa_token_id": 258883,
     "eoa_token_index": 258883,
     "eoi_token_id": 258882,
-    "eos_token_id": [
-        1,
-        106,
-        50
-    ],
+    "eos_token_id": 1,
     "image_token_id": 258880,
     "initializer_range": 0.02,
     "model_type": "gemma4",
@@ -22,250 +18,970 @@
         "group_size": 64,
         "bits": 4,
         "mode": "affine",
+        "language_model.model.layers.0.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.0.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.0.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
         "language_model.model.layers.0.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.1.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.2.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.3.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.4.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.5.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.6.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.7.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.8.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.9.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.10.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.11.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.12.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.13.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.14.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.15.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.16.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.17.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.18.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.19.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.20.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.21.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.22.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.23.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.24.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.25.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.26.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.27.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.28.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.29.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
         }
     },
     "quantization_config": {
         "group_size": 64,
         "bits": 4,
         "mode": "affine",
+        "language_model.model.layers.0.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.0.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.0.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
         "language_model.model.layers.0.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.1.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.1.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.2.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.2.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.3.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.3.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.4.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.4.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.5.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.5.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.6.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.6.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.7.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.7.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.8.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.8.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.9.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.9.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.10.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.10.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.11.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.11.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.12.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.12.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.13.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.13.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.14.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.14.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.15.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.15.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.16.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.16.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.17.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.17.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.18.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.18.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.19.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.19.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.20.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.20.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.21.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.21.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.22.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.22.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.23.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.23.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.24.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.24.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.25.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.25.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.26.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.26.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.27.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.27.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.28.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.28.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.gate_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.down_proj": {
+            "group_size": 64,
+            "bits": 4
+        },
+        "language_model.model.layers.29.mlp.up_proj": {
+            "group_size": 64,
+            "bits": 4
         },
         "language_model.model.layers.29.router.proj": {
             "group_size": 64,
-            "bits": 8
+            "bits": 4
         }
     },
     "text_config": {
@@ -350,5 +1066,48 @@
     "tie_word_embeddings": true,
     "transformers_version": "5.5.0.dev0",
     "video_token_id": 258884,
+    "vision_config": {
+        "_name_or_path": "",
+        "architectures": null,
+        "attention_bias": false,
+        "attention_dropout": 0.0,
+        "chunk_size_feed_forward": 0,
+        "default_output_length": 280,
+        "dtype": "bfloat16",
+        "global_head_dim": 72,
+        "head_dim": 72,
+        "hidden_activation": "gelu_pytorch_tanh",
+        "hidden_size": 1152,
+        "id2label": {
+            "0": "LABEL_0",
+            "1": "LABEL_1"
+        },
+        "initializer_range": 0.02,
+        "intermediate_size": 4304,
+        "is_encoder_decoder": false,
+        "label2id": {
+            "LABEL_0": 0,
+            "LABEL_1": 1
+        },
+        "max_position_embeddings": 131072,
+        "model_type": "gemma4_vision",
+        "num_attention_heads": 16,
+        "num_hidden_layers": 27,
+        "num_key_value_heads": 16,
+        "output_attentions": false,
+        "output_hidden_states": false,
+        "patch_size": 16,
+        "pooling_kernel_size": 3,
+        "position_embedding_size": 10240,
+        "problem_type": null,
+        "return_dict": true,
+        "rms_norm_eps": 1e-06,
+        "rope_parameters": {
+            "rope_theta": 100.0,
+            "rope_type": "default"
+        },
+        "standardize": true,
+        "use_clipped_linears": false
+    },
     "vision_soft_tokens_per_image": 280
-}
+}

generation_config.json

@@ -1,11 +1,7 @@
 {
   "bos_token_id": 2,
   "do_sample": true,
-  "eos_token_id": [
-    1,
-    106,
-    50
-  ],
+  "eos_token_id": 1,
   "pad_token_id": 0,
   "temperature": 1.0,
   "top_k": 64,

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79044a795347440cadc6356659b328ac910063e5453d75289916db136bcf454a
+oid sha256:683f420ee09550b8027bf0335c4202e196b37eb520b60a4bcbe8690b9e388c07
 size 5320218487

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8b2fbff6ea86ca622457c6cff481404e83fba26a91ae81610e00b4ff9f137798
+oid sha256:feab2873c2976fb7ed666f8462549564a29409364388e0c8f6b015729d165e15
 size 5363328422

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ec1c68e5bc23c05e87b261b7a200c09d2c36b5b01ee6dcc5d8461078363aecf2
+oid sha256:1de55874e782f8ebb1f16759a676809c2915ee14a42c63c7cfea80e068a41b01
 size 3516685531

processor_config.json

@@ -1,27 +1,5 @@
 {
-  "audio_ms_per_token": 40,
   "audio_seq_length": 750,
-  "feature_extractor": {
-    "dither": 0.0,
-    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
-    "feature_size": 128,
-    "fft_length": 512,
-    "fft_overdrive": false,
-    "frame_length": 320,
-    "hop_length": 160,
-    "input_scale_factor": 1.0,
-    "max_frequency": 8000.0,
-    "mel_floor": 0.001,
-    "min_frequency": 0.0,
-    "padding_side": "right",
-    "padding_value": 0.0,
-    "per_bin_mean": null,
-    "per_bin_stddev": null,
-    "preemphasis": 0.0,
-    "preemphasis_htk_flavor": true,
-    "return_attention_mask": true,
-    "sampling_rate": 16000
-  },
   "image_processor": {
     "do_convert_rgb": true,
     "do_normalize": false,
@@ -43,33 +21,12 @@
     "patch_size": 16,
     "pooling_kernel_size": 3,
     "resample": 3,
-    "rescale_factor": 0.00392156862745098
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "height": 224,
+      "width": 224
+    }
   },
   "image_seq_length": 280,
-  "processor_class": "Gemma4Processor",
-  "video_processor": {
-    "do_convert_rgb": true,
-    "do_normalize": true,
-    "do_rescale": true,
-    "do_resize": true,
-    "do_sample_frames": true,
-    "image_mean": [
-      0.0,
-      0.0,
-      0.0
-    ],
-    "image_std": [
-      1.0,
-      1.0,
-      1.0
-    ],
-    "max_soft_tokens": 70,
-    "num_frames": 32,
-    "patch_size": 16,
-    "pooling_kernel_size": 3,
-    "resample": 3,
-    "rescale_factor": 0.00392156862745098,
-    "return_metadata": false,
-    "video_processor_type": "Gemma4VideoProcessor"
-  }
+  "processor_class": "Gemma4Processor"
 }

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
-size 32169626
+oid sha256:12bac982b793c44b03d52a250a9f0d0b666813da566b910c24a6da0695fd11e6
+size 32170070

tokenizer_config.json

@@ -17,52 +17,32 @@
     "<|video|>"
   ],
   "image_token": "<|image|>",
+  "is_local": true,
   "mask_token": "<mask>",
   "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
   "pad_token": "<pad>",
   "padding_side": "left",
   "processor_class": "Gemma4Processor",
-  "response_schema": {
-    "type": "object",
-    "properties": {
-      "role": {
-        "const": "assistant"
-      },
-      "thinking": {
-        "type": "string"
-      },
-      "content": {
-        "type": "string"
-      },
-      "tool_calls": {
-        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>",
-        "type": "array",
-        "items": {
-          "type": "object",
-          "properties": {
-            "type": {
-              "const": "function"
-            },
-            "function": {
-              "type": "object",
-              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})",
-              "properties": {
-                "name": {
-                  "type": "string"
-                },
-                "arguments": {
-                  "type": "object",
-                  "x-parser": "gemma4-tool-call",
-                  "additionalProperties": {}
-                }
-              }
-            }
-          }
-        }
-      }
-    },
-    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<content>(?:(?!\\<\\|tool_call\\>)(?!\\<turn\\|\\>).)+)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?:\\<turn\\|\\>)?"
-  },
   "soc_token": "<|channel>",
   "sot_token": "<|turn>",
   "stc_token": "<|tool_call>",