Add files using upload-large-folder tool
Browse files- attn_out/layer_29_width_262k_l0_big/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_1m_l0_big/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_1m_l0_medium/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_262k_l0_big/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_262k_l0_medium/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_524k_l0_big/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_524k_l0_medium/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_65k_l0_big/config.json +9 -0
- crosscoder/layer_9_17_22_29_width_65k_l0_medium/config.json +9 -0
- resid_post_all/layer_0_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_11_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_11_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_12_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_13_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_13_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_14_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_14_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_15_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_18_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_19_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_1_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_20_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_20_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_21_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_21_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_22_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_23_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_24_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_24_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_24_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_25_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_25_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_27_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_29_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_29_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_2_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_30_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_31_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_32_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_32_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_33_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_33_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_3_width_262k_l0_small/config.json +9 -0
- resid_post_all/layer_5_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_5_width_262k_l0_big/config.json +9 -0
- resid_post_all/layer_7_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_7_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_9_width_16k_l0_big/config.json +9 -0
- resid_post_all/layer_9_width_16k_l0_small/config.json +9 -0
- resid_post_all/layer_9_width_262k_l0_small/config.json +9 -0
attn_out/layer_29_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.29.self_attn.o_proj.input",
|
| 3 |
+
"hf_hook_point_out": "model.layers.29.self_attn.o_proj.input",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_1m_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 1048576,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_1m_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 1048576,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_262k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_524k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 524288,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_524k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 524288,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_65k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 150,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
crosscoder/layer_9_17_22_29_width_65k_l0_medium/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.{9.0,17.5,22.6,29.4}.output",
|
| 4 |
+
"width": 65536,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 50,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_0_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.0.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.0.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 10,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_11_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.11.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.11.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 118,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_11_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.11.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.11.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 19,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_12_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.12.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.12.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_13_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.13.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.13.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_13_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.13.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.13.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_14_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.14.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.14.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_14_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.14.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.14.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_15_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.15.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.15.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_18_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.18.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.18.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_19_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.19.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.19.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_1_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.1.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.1.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 65,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_20_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.20.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.20.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_20_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.20.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.20.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_21_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.21.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.21.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_21_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.21.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.21.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_22_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.22.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.22.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_23_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.23.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.23.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_24_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.24.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.24.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_24_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.24.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.24.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_24_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.24.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.24.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_25_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.25.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.25.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_25_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.25.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.25.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_27_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.27.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.27.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_29_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.29.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.29.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_29_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.29.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.29.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_2_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.2.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.2.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 11,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_30_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.30.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.30.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_31_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.31.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.31.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_32_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.32.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.32.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_32_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.32.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.32.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_33_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.33.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.33.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 120,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_33_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.33.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.33.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 20,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_3_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.3.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.3.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 12,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_5_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 86,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_5_width_262k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.5.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.5.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 86,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_7_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.7.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.7.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 97,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_7_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.7.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.7.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 16,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_9_width_16k_l0_big/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 107,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_9_width_16k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.output",
|
| 4 |
+
"width": 16384,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 17,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|
resid_post_all/layer_9_width_262k_l0_small/config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hf_hook_point_in": "model.layers.9.output",
|
| 3 |
+
"hf_hook_point_out": "model.layers.9.output",
|
| 4 |
+
"width": 262144,
|
| 5 |
+
"model_name": "gemma-v3-4b-it",
|
| 6 |
+
"architecture": "jump_relu",
|
| 7 |
+
"l0": 17,
|
| 8 |
+
"affine_connection": false
|
| 9 |
+
}
|