docs: update model card, context lengths, and config specs

.gitattributes

@@ -39,5 +39,6 @@ e2b/xnnpack/gemma_4_e2b_xnnpack_8da4w.pte filter=lfs diff=lfs merge=lfs -text
 e2b/xnnpack/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 e2b/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 e2b/mlx/gemma4_e2b_mlx_int4.pte filter=lfs diff=lfs merge=lfs -text
+e2b/mlx/gemma_4_e2b_mlx_int4.pte filter=lfs diff=lfs merge=lfs -text
 gemma4_e2b_omni_mlx_int4_v4.pte filter=lfs diff=lfs merge=lfs -text
 e2b/mlx/gemma4_e2b_mlx_int4_old.pte filter=lfs diff=lfs merge=lfs -text

README.md

@@ -19,24 +19,26 @@ library as ExecuTorch `.pte` programs, ready to run on device.
 
 ## Variants
 
-| Path | Backend | Precision |
-| --- | --- | --- |
-| `e2b/mlx/gemma4_e2b_mlx_int4.pte` | mlx | int4 |
-| `e2b/vulkan/gemma_4_e2b_vulkan_8da4w.pte` | vulkan | 8da4w |
-| `e2b/xnnpack/gemma_4_e2b_xnnpack_8da4w.pte` | xnnpack | 8da4w |
+| Path | Backend | Precision | Context length |
+| --- | --- | --- | --- |
+| `e2b/mlx/gemma_4_e2b_mlx_int4.pte` | mlx | int4 | 2,048 |
+| `e2b/vulkan/gemma_4_e2b_vulkan_8da4w.pte` | vulkan | 8da4w | 2,048 |
+| `e2b/xnnpack/gemma_4_e2b_xnnpack_8da4w.pte` | xnnpack | 8da4w | 4,048 |
+
+
 
 ## Repository structure
 
 ```
 config.json                                40 B
-e2b/mlx/config.json                        1.2 kB
-e2b/mlx/gemma4_e2b_mlx_int4.pte            3.0 GB
-e2b/tokenizer.json                         30.7 MB
-e2b/tokenizer_config.json                  21.8 kB
-e2b/vulkan/config.json                     1.5 kB
-e2b/vulkan/gemma_4_e2b_vulkan_8da4w.pte    4.1 GB
-e2b/xnnpack/config.json                    1.4 kB
-e2b/xnnpack/gemma_4_e2b_xnnpack_8da4w.pte  4.4 GB
+e2b/mlx/config.json                        4.7 KiB
+e2b/mlx/gemma_4_e2b_mlx_int4.pte           2.5 GiB
+e2b/tokenizer.json                         30.7 MiB
+e2b/tokenizer_config.json                  21.8 KiB
+e2b/vulkan/config.json                     4.7 KiB
+e2b/vulkan/gemma_4_e2b_vulkan_8da4w.pte    3.5 GiB
+e2b/xnnpack/config.json                    4.7 KiB
+e2b/xnnpack/gemma_4_e2b_xnnpack_8da4w.pte  3.0 GiB
 ```
 
 ## Compatibility

e2b/mlx/config.json

@@ -11,7 +11,7 @@
   "license": "gemma",
   "variants": [
     {
-      "file": "gemma4_e2b_mlx_int4.pte",
+      "file": "gemma_4_e2b_mlx_int4.pte",
       "precision": "int4",
       "quantized": true,
       "default": true,

e2b/vulkan/config.json

@@ -210,4 +210,4 @@
     "token_embedding",
     "vision_encoder"
   ]
-}
+}