Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- transcoder_all/layer_0_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_10_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_12_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_14_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_15_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_18_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_1_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_21_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_22_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_22_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_23_width_262k_l0_big/config.json +3 -2
- transcoder_all/layer_24_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_25_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_26_width_16k_l0_big/config.json +3 -2
- transcoder_all/layer_27_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_28_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_28_width_262k_l0_big/config.json +3 -2
- transcoder_all/layer_29_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_2_width_16k_l0_big/config.json +3 -2
- transcoder_all/layer_2_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_31_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_31_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_31_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_32_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_32_width_262k_l0_big/config.json +3 -2
- transcoder_all/layer_33_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_34_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_36_width_16k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_36_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_38_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_38_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_3_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_42_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_42_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_43_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_44_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_45_width_16k_l0_big/config.json +3 -2
- transcoder_all/layer_46_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_47_width_16k_l0_big/config.json +3 -2
- transcoder_all/layer_47_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_48_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_48_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_48_width_262k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_49_width_262k_l0_big/config.json +3 -2
- transcoder_all/layer_49_width_262k_l0_small/config.json +3 -2
- transcoder_all/layer_4_width_262k_l0_big/config.json +3 -2
- transcoder_all/layer_50_width_16k_l0_small/config.json +3 -2
- transcoder_all/layer_51_width_16k_l0_small_affine/config.json +3 -2
- transcoder_all/layer_58_width_262k_l0_big_affine/config.json +3 -2
- transcoder_all/layer_59_width_16k_l0_small_affine/config.json +3 -2
transcoder_all/layer_0_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_10_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.10.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.10.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 89,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.10.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.10.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 89,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_12_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 94,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 94,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_14_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.14.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.14.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 16,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.14.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.14.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 16,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_15_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.15.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 17,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.15.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 17,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_18_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 18,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 18,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_1_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.1.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.1.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.1.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.1.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_21_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.21.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.21.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.21.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.21.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_22_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_22_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_23_width_262k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.23.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.23.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.23.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.23.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_24_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.24.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.24.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.24.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.24.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_25_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.25.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.25.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.25.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.25.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_26_width_16k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.26.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.26.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.26.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.26.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_27_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_28_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_28_width_262k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_29_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_2_width_16k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 65,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 65,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_2_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 10,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_31_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_31_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_31_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_32_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.32.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.32.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.32.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.32.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_32_width_262k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.32.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.32.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.32.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.32.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_33_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_34_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.34.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.34.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.34.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.34.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_36_width_16k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.36.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.36.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.36.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.36.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_36_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.36.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.36.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.36.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.36.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_38_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.38.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.38.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.38.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.38.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_38_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.38.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.38.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.38.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.38.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_3_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.3.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.3.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 11,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.3.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.3.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 11,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_42_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.42.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.42.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.42.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.42.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_42_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.42.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.42.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.42.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.42.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_43_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.43.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.43.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.43.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.43.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_44_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.44.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.44.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.44.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.44.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_45_width_16k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.45.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.45.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.45.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.45.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_46_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.46.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.46.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.46.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.46.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_47_width_16k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.47.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.47.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.47.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.47.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_47_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.47.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.47.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.47.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.47.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_48_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_48_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_48_width_262k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.48.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.48.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_49_width_262k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.49.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.49.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.49.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.49.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_49_width_262k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.49.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.49.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.49.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.49.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_4_width_262k_l0_big/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 71,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 71,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_50_width_16k_l0_small/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.50.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.50.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": false
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.50.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.50.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": false,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_51_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.51.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.51.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.51.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.51.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_58_width_262k_l0_big_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.58.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.58.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.58.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.58.post_feedforward_layernorm.output",
|
| 4 |
"width": 262144,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 120,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|
transcoder_all/layer_59_width_16k_l0_small_affine/config.json
CHANGED
|
@@ -2,8 +2,9 @@
|
|
| 2 |
"hf_hook_point_in": "model.layers.59.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.59.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
-
"model_name": "gemma-
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
-
"affine_connection": true
|
|
|
|
| 9 |
}
|
|
|
|
| 2 |
"hf_hook_point_in": "model.layers.59.pre_feedforward_layernorm.output",
|
| 3 |
"hf_hook_point_out": "model.layers.59.post_feedforward_layernorm.output",
|
| 4 |
"width": 16384,
|
| 5 |
+
"model_name": "google/gemma-3-27b-it",
|
| 6 |
"architecture": "jump_relu",
|
| 7 |
"l0": 20,
|
| 8 |
+
"affine_connection": true,
|
| 9 |
+
"type": "transcoder"
|
| 10 |
}
|