Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- clt/width_262k_l0_big/config.json +9 -0
- clt/width_524k_l0_medium_affine/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_1m_l0_big/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_1m_l0_medium/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_262k_l0_big/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_262k_l0_medium/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_524k_l0_big/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_524k_l0_medium/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_65k_l0_big/config.json +9 -0
- crosscoder/layer_5_9_12_15_width_65k_l0_medium/config.json +9 -0
- mlp_out/layer_12_width_16k_l0_big/config.json +9 -0
- mlp_out/layer_12_width_16k_l0_medium/config.json +9 -0
- mlp_out/layer_12_width_16k_l0_small/config.json +9 -0
- mlp_out/layer_12_width_262k_l0_big/config.json +9 -0
- mlp_out/layer_12_width_262k_l0_medium/config.json +9 -0
- mlp_out/layer_12_width_262k_l0_small/config.json +9 -0
- mlp_out/layer_12_width_65k_l0_big/config.json +9 -0
- mlp_out/layer_12_width_65k_l0_medium/config.json +9 -0
- mlp_out/layer_12_width_65k_l0_small/config.json +9 -0
- mlp_out/layer_15_width_16k_l0_big/config.json +9 -0
- mlp_out/layer_15_width_16k_l0_medium/config.json +9 -0
- mlp_out/layer_15_width_16k_l0_small/config.json +9 -0
- mlp_out/layer_15_width_262k_l0_big/config.json +9 -0
- mlp_out/layer_15_width_262k_l0_medium/config.json +9 -0
- mlp_out/layer_15_width_262k_l0_small/config.json +9 -0
- mlp_out/layer_15_width_65k_l0_big/config.json +9 -0
- mlp_out/layer_15_width_65k_l0_medium/config.json +9 -0
- mlp_out/layer_15_width_65k_l0_small/config.json +9 -0
- mlp_out/layer_5_width_16k_l0_big/config.json +9 -0
- mlp_out/layer_5_width_16k_l0_medium/config.json +9 -0
- mlp_out/layer_5_width_16k_l0_small/config.json +9 -0
- mlp_out/layer_5_width_262k_l0_big/config.json +9 -0
- mlp_out/layer_5_width_262k_l0_medium/config.json +9 -0
- mlp_out/layer_5_width_262k_l0_small/config.json +9 -0
- mlp_out/layer_5_width_65k_l0_big/config.json +9 -0
- mlp_out/layer_5_width_65k_l0_medium/config.json +9 -0
- mlp_out/layer_5_width_65k_l0_small/config.json +9 -0
- mlp_out/layer_9_width_16k_l0_big/config.json +9 -0
- mlp_out/layer_9_width_16k_l0_medium/config.json +9 -0
- mlp_out/layer_9_width_16k_l0_small/config.json +9 -0
- mlp_out/layer_9_width_262k_l0_big/config.json +9 -0
- mlp_out/layer_9_width_262k_l0_medium/config.json +9 -0
- mlp_out/layer_9_width_262k_l0_small/config.json +9 -0
- mlp_out/layer_9_width_65k_l0_big/config.json +9 -0
- mlp_out/layer_9_width_65k_l0_medium/config.json +9 -0
- mlp_out/layer_9_width_65k_l0_small/config.json +9 -0
- mlp_out_all/layer_0_width_16k_l0_big/config.json +9 -0
- mlp_out_all/layer_0_width_16k_l0_small/config.json +9 -0
- mlp_out_all/layer_0_width_262k_l0_big/config.json +9 -0
- mlp_out_all/layer_0_width_262k_l0_small/config.json +9 -0
clt/width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{all}.pre_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{all}.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262080,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
clt/width_524k_l0_medium_affine/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{all}.pre_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{all}.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 524160,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": true
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_1m_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 1048576,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_1m_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 1048576,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_524k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 524288,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_524k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 524288,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_5_9_12_15_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{5.0,9.5,12.200000000000001,15.799999999999999}.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_16k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_12_width_65k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_16k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_15_width_65k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 135,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_16k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 55,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 18,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 135,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 55,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 18,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 135,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 55,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_5_width_65k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 18,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_16k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out/layer_9_width_65k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out_all/layer_0_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.0.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out_all/layer_0_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.0.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 10,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out_all/layer_0_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.0.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 60,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
mlp_out_all/layer_0_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.0.post_feedforward_layernorm.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-270m-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 10,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|