Upload folder using huggingface_hub

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:02bfb07b081a97e16ba1222a980a4cb7bf5071981da538b0ed8247566773d870
-size 4998338056
+oid sha256:7a6d7ef044775c8fb2ab485f33a87de2dde4a3c44968ebfe03e53617691ffd84
+size 4998337960

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:feb5b364c9438d060ab89931bd62c7065e32495d8cd2c5aff6bea5baf5d1a724
-size 543078152
+oid sha256:bd49c9b55dd05b309de0173327cf383bae563b8e9733476bf75257703eb583ab
+size 543078144

model.safetensors.index.json

@@ -72,7 +72,7 @@
     "model.mamba_layers.0.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.0.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.0.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.0.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.0.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.0.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.0.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -81,7 +81,7 @@
     "model.mamba_layers.1.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.1.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.1.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.1.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.1.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.1.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.1.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -90,7 +90,7 @@
     "model.mamba_layers.10.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.10.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.10.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.10.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.10.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.10.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.10.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -99,7 +99,7 @@
     "model.mamba_layers.11.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.11.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.11.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.11.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.11.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.11.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.11.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -108,7 +108,7 @@
     "model.mamba_layers.12.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.12.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.12.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.12.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.12.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.12.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.12.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -117,7 +117,7 @@
     "model.mamba_layers.13.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.13.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.13.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.13.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.13.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.13.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.13.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -126,7 +126,7 @@
     "model.mamba_layers.14.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.14.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.14.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.14.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.14.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.14.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.14.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -135,7 +135,7 @@
     "model.mamba_layers.15.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.15.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.15.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.15.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.15.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.15.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.15.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -144,7 +144,7 @@
     "model.mamba_layers.16.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.16.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.16.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.16.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.16.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.16.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.16.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -153,7 +153,7 @@
     "model.mamba_layers.17.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.17.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.17.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.17.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.17.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.17.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.17.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -162,7 +162,7 @@
     "model.mamba_layers.18.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.18.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.18.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.18.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.18.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.18.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.18.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -171,7 +171,7 @@
     "model.mamba_layers.19.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.19.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.19.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.19.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.19.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.19.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.19.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -180,7 +180,7 @@
     "model.mamba_layers.2.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.2.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.2.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.2.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.2.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.2.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.2.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -189,7 +189,7 @@
     "model.mamba_layers.20.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.20.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.20.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.20.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.20.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.20.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.20.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -198,7 +198,7 @@
     "model.mamba_layers.21.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.21.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.21.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.21.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.21.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.21.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.21.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -207,7 +207,7 @@
     "model.mamba_layers.22.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.22.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.22.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.22.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.22.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.22.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.22.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -216,7 +216,7 @@
     "model.mamba_layers.23.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.23.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.23.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.23.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.23.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.23.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.23.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -225,7 +225,7 @@
     "model.mamba_layers.24.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.24.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.24.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.24.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.24.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.24.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.24.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -234,7 +234,7 @@
     "model.mamba_layers.25.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.25.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.25.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.25.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.25.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.25.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.25.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -243,7 +243,7 @@
     "model.mamba_layers.26.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.26.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.26.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.26.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.26.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.26.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.26.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -252,7 +252,7 @@
     "model.mamba_layers.27.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.27.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.27.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.27.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.27.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.27.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.27.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.28.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -261,7 +261,7 @@
     "model.mamba_layers.28.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.28.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.28.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.28.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.28.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.28.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.28.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.29.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -270,7 +270,7 @@
     "model.mamba_layers.29.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.29.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.29.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.29.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.29.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.29.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.29.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -279,7 +279,7 @@
     "model.mamba_layers.3.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.3.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.3.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.3.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.3.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.3.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.3.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.30.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -288,7 +288,7 @@
     "model.mamba_layers.30.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.30.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.30.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.30.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.30.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.30.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.30.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.31.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -297,7 +297,7 @@
     "model.mamba_layers.31.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.31.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.31.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.31.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.31.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.31.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.31.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.32.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -306,7 +306,7 @@
     "model.mamba_layers.32.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.32.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.32.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.32.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.32.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.32.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.32.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.33.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -315,7 +315,7 @@
     "model.mamba_layers.33.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.33.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.33.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.33.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.33.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.33.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.33.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.34.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -324,7 +324,7 @@
     "model.mamba_layers.34.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.34.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.34.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.34.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.34.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.34.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.34.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.35.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -333,7 +333,7 @@
     "model.mamba_layers.35.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.35.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.35.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.35.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.35.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.35.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.35.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.36.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -342,7 +342,7 @@
     "model.mamba_layers.36.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.36.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.36.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.36.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.36.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.36.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.36.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.37.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -351,7 +351,7 @@
     "model.mamba_layers.37.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.37.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.37.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.37.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.37.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.37.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.37.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.38.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -360,7 +360,7 @@
     "model.mamba_layers.38.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.38.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.38.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.38.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.38.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.38.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.38.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.39.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -369,7 +369,7 @@
     "model.mamba_layers.39.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.39.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.39.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.39.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.39.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.39.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.39.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -378,7 +378,7 @@
     "model.mamba_layers.4.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.4.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.4.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.4.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.4.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.4.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.4.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.40.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -387,7 +387,7 @@
     "model.mamba_layers.40.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.40.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.40.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.40.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.40.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.40.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.40.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.41.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -396,7 +396,7 @@
     "model.mamba_layers.41.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.41.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.41.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.41.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.41.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.41.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.41.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.42.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -405,7 +405,7 @@
     "model.mamba_layers.42.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.42.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.42.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.42.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.42.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.42.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.42.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.43.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -414,7 +414,7 @@
     "model.mamba_layers.43.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.43.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.43.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.43.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.43.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.43.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.43.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.44.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -423,7 +423,7 @@
     "model.mamba_layers.44.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.44.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.44.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.44.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.44.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.44.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.44.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.45.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -432,7 +432,7 @@
     "model.mamba_layers.45.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.45.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.45.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.45.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.45.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.45.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.45.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.46.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -441,7 +441,7 @@
     "model.mamba_layers.46.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.46.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.46.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.46.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.46.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.46.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.46.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.47.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -450,7 +450,7 @@
     "model.mamba_layers.47.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.47.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.47.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.47.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.47.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.47.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.47.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.48.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -459,7 +459,7 @@
     "model.mamba_layers.48.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.48.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.48.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.48.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.48.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.48.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.48.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.49.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -468,7 +468,7 @@
     "model.mamba_layers.49.mamba.conv1d.bias": "model-00002-of-00002.safetensors",
     "model.mamba_layers.49.mamba.conv1d.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.49.mamba.dt_bias": "model-00002-of-00002.safetensors",
-    "model.mamba_layers.49.mamba.in_proj.0.weight": "model-00002-of-00002.safetensors",
+    "model.mamba_layers.49.mamba.in_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.49.mamba.norm.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.49.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -477,7 +477,7 @@
     "model.mamba_layers.5.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.5.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.5.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.5.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.5.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.5.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.5.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.50.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -486,7 +486,7 @@
     "model.mamba_layers.50.mamba.conv1d.bias": "model-00002-of-00002.safetensors",
     "model.mamba_layers.50.mamba.conv1d.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.50.mamba.dt_bias": "model-00002-of-00002.safetensors",
-    "model.mamba_layers.50.mamba.in_proj.0.weight": "model-00002-of-00002.safetensors",
+    "model.mamba_layers.50.mamba.in_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.50.mamba.norm.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.50.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.51.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -495,7 +495,7 @@
     "model.mamba_layers.51.mamba.conv1d.bias": "model-00002-of-00002.safetensors",
     "model.mamba_layers.51.mamba.conv1d.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.51.mamba.dt_bias": "model-00002-of-00002.safetensors",
-    "model.mamba_layers.51.mamba.in_proj.0.weight": "model-00002-of-00002.safetensors",
+    "model.mamba_layers.51.mamba.in_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.51.mamba.norm.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.51.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.52.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -504,7 +504,7 @@
     "model.mamba_layers.52.mamba.conv1d.bias": "model-00002-of-00002.safetensors",
     "model.mamba_layers.52.mamba.conv1d.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.52.mamba.dt_bias": "model-00002-of-00002.safetensors",
-    "model.mamba_layers.52.mamba.in_proj.0.weight": "model-00002-of-00002.safetensors",
+    "model.mamba_layers.52.mamba.in_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.52.mamba.norm.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.52.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.53.input_layernorm.weight": "model-00002-of-00002.safetensors",
@@ -513,7 +513,7 @@
     "model.mamba_layers.53.mamba.conv1d.bias": "model-00002-of-00002.safetensors",
     "model.mamba_layers.53.mamba.conv1d.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.53.mamba.dt_bias": "model-00002-of-00002.safetensors",
-    "model.mamba_layers.53.mamba.in_proj.0.weight": "model-00002-of-00002.safetensors",
+    "model.mamba_layers.53.mamba.in_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.53.mamba.norm.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.53.mamba.out_proj.weight": "model-00002-of-00002.safetensors",
     "model.mamba_layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -522,7 +522,7 @@
     "model.mamba_layers.6.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.6.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.6.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.6.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.6.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.6.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.6.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -531,7 +531,7 @@
     "model.mamba_layers.7.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.7.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.7.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.7.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.7.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.7.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.7.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -540,7 +540,7 @@
     "model.mamba_layers.8.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.8.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.8.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.8.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.8.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.8.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.8.mamba.out_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
@@ -549,8 +549,8 @@
     "model.mamba_layers.9.mamba.conv1d.bias": "model-00001-of-00002.safetensors",
     "model.mamba_layers.9.mamba.conv1d.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.9.mamba.dt_bias": "model-00001-of-00002.safetensors",
-    "model.mamba_layers.9.mamba.in_proj.0.weight": "model-00001-of-00002.safetensors",
+    "model.mamba_layers.9.mamba.in_proj.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.9.mamba.norm.weight": "model-00001-of-00002.safetensors",
     "model.mamba_layers.9.mamba.out_proj.weight": "model-00001-of-00002.safetensors"
   }
-}
+}