diff --git a/attn_out/layer_13_width_16k_l0_big/config.json b/attn_out/layer_13_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc4cc6a1d380dcfeeb6891e90f5616ba205bd3dd --- /dev/null +++ b/attn_out/layer_13_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_16k_l0_medium/config.json b/attn_out/layer_13_width_16k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..39b47c65e806caf0c909e2bd7de6c5c30a9d1e69 --- /dev/null +++ b/attn_out/layer_13_width_16k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_16k_l0_small/config.json b/attn_out/layer_13_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a7c1d3eda57fa81f6d5ceb42e04e37bb54a95c5d --- /dev/null +++ b/attn_out/layer_13_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_262k_l0_big/config.json b/attn_out/layer_13_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..de9706a5897cf704870f598b8c6ae256465ced46 --- /dev/null +++ b/attn_out/layer_13_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_262k_l0_medium/config.json b/attn_out/layer_13_width_262k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6838eff87c8edca30016c29ed9acb48966a987f5 --- /dev/null +++ b/attn_out/layer_13_width_262k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_262k_l0_small/config.json b/attn_out/layer_13_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..137a43400124a439bc494feeb4a5786ee572b046 --- /dev/null +++ b/attn_out/layer_13_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_65k_l0_big/config.json b/attn_out/layer_13_width_65k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7e061bac86e9de49dbca6361feea24a3caafb389 --- /dev/null +++ b/attn_out/layer_13_width_65k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_65k_l0_medium/config.json b/attn_out/layer_13_width_65k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7806266a745a0862b511ca5429a3c935906b2e3b --- /dev/null +++ b/attn_out/layer_13_width_65k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_13_width_65k_l0_small/config.json b/attn_out/layer_13_width_65k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2bad52cdb955e222afe2a2435a17c7c039d506e2 --- /dev/null +++ b/attn_out/layer_13_width_65k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_16k_l0_big/config.json b/attn_out/layer_17_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ceb571107a844a1445370472f7465e58ed7ffb69 --- /dev/null +++ b/attn_out/layer_17_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_16k_l0_medium/config.json b/attn_out/layer_17_width_16k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..397af267135649c7997e0a2869b7161c41744b22 --- /dev/null +++ b/attn_out/layer_17_width_16k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_16k_l0_small/config.json b/attn_out/layer_17_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e653730f44b90fb9a531d3d65d1a7f22c6e5758d --- /dev/null +++ b/attn_out/layer_17_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_262k_l0_big/config.json b/attn_out/layer_17_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..778f76b8dd1149eaba65fbd8c440ca626a7b0d03 --- /dev/null +++ b/attn_out/layer_17_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_262k_l0_medium/config.json b/attn_out/layer_17_width_262k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c2f2b770346adadaa22afe9809fb8d28b194c205 --- /dev/null +++ b/attn_out/layer_17_width_262k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_262k_l0_small/config.json b/attn_out/layer_17_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..33f0b4c2c95bef7aa68b329278cf98614e576796 --- /dev/null +++ b/attn_out/layer_17_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_65k_l0_big/config.json b/attn_out/layer_17_width_65k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..60eed3d95d07b78e95e2029b4249772bd4b7a6df --- /dev/null +++ b/attn_out/layer_17_width_65k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_65k_l0_medium/config.json b/attn_out/layer_17_width_65k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..911d888ffffd7194dceaff7d8e87331ad2593e0c --- /dev/null +++ b/attn_out/layer_17_width_65k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_17_width_65k_l0_small/config.json b/attn_out/layer_17_width_65k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8902b14f8ebf0e1e9e99f5bf268a00bdd4866eb0 --- /dev/null +++ b/attn_out/layer_17_width_65k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.17.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_16k_l0_big/config.json b/attn_out/layer_22_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..beee6428141fc883852ba82730d1e73dbacfdd99 --- /dev/null +++ b/attn_out/layer_22_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_16k_l0_small/config.json b/attn_out/layer_22_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f75d9d2725215cbb939837e27cf57bbad2aa26b7 --- /dev/null +++ b/attn_out/layer_22_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_262k_l0_big/config.json b/attn_out/layer_22_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e6eca408eec1150d3d7905de4e2f401ea7185bba --- /dev/null +++ b/attn_out/layer_22_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_262k_l0_medium/config.json b/attn_out/layer_22_width_262k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dd350fcfa8b279fc3ad540f9b4e4977c5088eac9 --- /dev/null +++ b/attn_out/layer_22_width_262k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_262k_l0_small/config.json b/attn_out/layer_22_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..13bab4f0e0e28364d7a84c02c4e35eb0475ecce1 --- /dev/null +++ b/attn_out/layer_22_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_65k_l0_big/config.json b/attn_out/layer_22_width_65k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bbe7212f80869f622272fd468ea9ac9b2b7b282d --- /dev/null +++ b/attn_out/layer_22_width_65k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 150, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_65k_l0_medium/config.json b/attn_out/layer_22_width_65k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8fdd341c7beda3e0c457e545049ba7a1c0254a93 --- /dev/null +++ b/attn_out/layer_22_width_65k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_22_width_65k_l0_small/config.json b/attn_out/layer_22_width_65k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fc4806564332249d60507fca96f14e0cb9418702 --- /dev/null +++ b/attn_out/layer_22_width_65k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_16k_l0_big/config.json b/attn_out/layer_7_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..521ef2cd291138fe53aa7d3675b025fc89442820 --- /dev/null +++ b/attn_out/layer_7_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 132, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_16k_l0_medium/config.json b/attn_out/layer_7_width_16k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e02d6109a409f1085852e69a8dbf37b53cad9801 --- /dev/null +++ b/attn_out/layer_7_width_16k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 54, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_16k_l0_small/config.json b/attn_out/layer_7_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8f9c7832622f154258516f514da435d0169628e5 --- /dev/null +++ b/attn_out/layer_7_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_262k_l0_big/config.json b/attn_out/layer_7_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b2a220f0b270a9a321bc7b2c630860d0562b9180 --- /dev/null +++ b/attn_out/layer_7_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 132, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_262k_l0_medium/config.json b/attn_out/layer_7_width_262k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8969c09bf1a2ba42b1b06c4c74f492a9dfe086f8 --- /dev/null +++ b/attn_out/layer_7_width_262k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 54, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_262k_l0_small/config.json b/attn_out/layer_7_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2c38831e0884b198cfdb45adba8e5193284be5c8 --- /dev/null +++ b/attn_out/layer_7_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_65k_l0_big/config.json b/attn_out/layer_7_width_65k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d4bb7efee60af057acc9b0f0c8ae1f8958f59cf --- /dev/null +++ b/attn_out/layer_7_width_65k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 132, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_65k_l0_medium/config.json b/attn_out/layer_7_width_65k_l0_medium/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d14e87d26071c5def230a57b7636fe120a32b9a --- /dev/null +++ b/attn_out/layer_7_width_65k_l0_medium/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 54, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out/layer_7_width_65k_l0_small/config.json b/attn_out/layer_7_width_65k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..aa03cf26dd8e4ca80ef213c8a306892d1f8ead99 --- /dev/null +++ b/attn_out/layer_7_width_65k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.7.self_attn.o_proj.input", + "width": 65536, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_10_width_16k_l0_small/config.json b/attn_out_all/layer_10_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..54aff9fbad712a203fc44e8e06072d866b6cd32d --- /dev/null +++ b/attn_out_all/layer_10_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.10.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.10.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_12_width_16k_l0_small/config.json b/attn_out_all/layer_12_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..21d8626bcf188ee616d8e1180b174dec2a4a0b04 --- /dev/null +++ b/attn_out_all/layer_12_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.12.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_12_width_262k_l0_big/config.json b/attn_out_all/layer_12_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c7bcea5bfca8f14b7e0a30d6fecde187df1c408d --- /dev/null +++ b/attn_out_all/layer_12_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.12.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_13_width_262k_l0_big/config.json b/attn_out_all/layer_13_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4f680b48e3a93c2d06e8ecbdb58180f137f8e066 --- /dev/null +++ b/attn_out_all/layer_13_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.13.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_14_width_16k_l0_small/config.json b/attn_out_all/layer_14_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6747c156e24ff32471aef090bf3a8f18732a05d6 --- /dev/null +++ b/attn_out_all/layer_14_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.14.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.14.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_15_width_262k_l0_small/config.json b/attn_out_all/layer_15_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b19a5b21942e1321e4980fb6839d0f7f66ab18e2 --- /dev/null +++ b/attn_out_all/layer_15_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.15.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.15.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_16_width_16k_l0_big/config.json b/attn_out_all/layer_16_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..10d95f1d7e35f918abe0efc5def94841d4419bec --- /dev/null +++ b/attn_out_all/layer_16_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.16.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.16.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_18_width_262k_l0_small/config.json b/attn_out_all/layer_18_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e6dbbcd6850471d2aa583edc30747eb2059f70d4 --- /dev/null +++ b/attn_out_all/layer_18_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.18.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.18.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_1_width_262k_l0_small/config.json b/attn_out_all/layer_1_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3b7bd2144eff25e89233b0ecb5fd1ba64782affc --- /dev/null +++ b/attn_out_all/layer_1_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.1.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.1.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 11, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_21_width_16k_l0_big/config.json b/attn_out_all/layer_21_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c1d19f37c404ef8f56b895f1341357a63d69120f --- /dev/null +++ b/attn_out_all/layer_21_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.21.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.21.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_22_width_262k_l0_small/config.json b/attn_out_all/layer_22_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..13bab4f0e0e28364d7a84c02c4e35eb0475ecce1 --- /dev/null +++ b/attn_out_all/layer_22_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.22.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_24_width_16k_l0_small/config.json b/attn_out_all/layer_24_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9fd763d12cac945c112208284db9080db27a01c5 --- /dev/null +++ b/attn_out_all/layer_24_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.24.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.24.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_24_width_262k_l0_small/config.json b/attn_out_all/layer_24_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7c37f32bcce4519f401a99b1192502637085aa58 --- /dev/null +++ b/attn_out_all/layer_24_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.24.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.24.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_25_width_16k_l0_big/config.json b/attn_out_all/layer_25_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b3e6508125efed551f808462b09f741b2cc00a39 --- /dev/null +++ b/attn_out_all/layer_25_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.25.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.25.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_25_width_262k_l0_small/config.json b/attn_out_all/layer_25_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2027fa53d1cb76a6ce99878a502542485987f59f --- /dev/null +++ b/attn_out_all/layer_25_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.25.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.25.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_2_width_16k_l0_big/config.json b/attn_out_all/layer_2_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..346e563686d3264bf4ad1ac81dd4f7d90eecfcc1 --- /dev/null +++ b/attn_out_all/layer_2_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.2.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.2.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 73, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_4_width_16k_l0_small/config.json b/attn_out_all/layer_4_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ac38bb0177b6ed97cb66712ccfba5a7f96ae99b7 --- /dev/null +++ b/attn_out_all/layer_4_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.4.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 14, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_5_width_262k_l0_small/config.json b/attn_out_all/layer_5_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9e1a901502b57fca56894ae8480fb418210b81ae --- /dev/null +++ b/attn_out_all/layer_5_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.5.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.5.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 15, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_6_width_262k_l0_small/config.json b/attn_out_all/layer_6_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..86be48bcbac753ca31f5bb9e02d6adab0573458a --- /dev/null +++ b/attn_out_all/layer_6_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.6.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.6.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 16, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_8_width_262k_l0_small/config.json b/attn_out_all/layer_8_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ed42d926752779d216481174863ef07d816874c5 --- /dev/null +++ b/attn_out_all/layer_8_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.8.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.8.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 19, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_9_width_262k_l0_big/config.json b/attn_out_all/layer_9_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ff911b3806a8e64a969ccdbb6417efe8a12a9f18 --- /dev/null +++ b/attn_out_all/layer_9_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.9.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_9_width_262k_l0_small/config.json b/attn_out_all/layer_9_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ad1c74aa6b0ec05dfae03a26d5a793be9129774f --- /dev/null +++ b/attn_out_all/layer_9_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.9.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/transcoder_all/layer_0_width_16k_l0_big/config.json b/transcoder_all/layer_0_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ccf6d1ffe87cab44e0a26559d1fa21bf83055816 --- /dev/null +++ b/transcoder_all/layer_0_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_0_width_16k_l0_small/config.json b/transcoder_all/layer_0_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..71bfb57860069d1b2e29c7756c1ac1a7f714eee2 --- /dev/null +++ b/transcoder_all/layer_0_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_0_width_262k_l0_big/config.json b/transcoder_all/layer_0_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..803c81f3e67838b0e9d92048250f05e0382c3c81 --- /dev/null +++ b/transcoder_all/layer_0_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_0_width_262k_l0_big_affine/config.json b/transcoder_all/layer_0_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..06d9695320256c27190f77bd73cdedd53eb3a708 --- /dev/null +++ b/transcoder_all/layer_0_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 60, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_0_width_262k_l0_small_affine/config.json b/transcoder_all/layer_0_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f431ff5a9a733ae5b13d0e050d2171a166c770df --- /dev/null +++ b/transcoder_all/layer_0_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.0.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_10_width_16k_l0_big/config.json b/transcoder_all/layer_10_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bfa0eec1061c6ef28a8944545b8757e1b19983a6 --- /dev/null +++ b/transcoder_all/layer_10_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.10.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.10.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_10_width_16k_l0_small_affine/config.json b/transcoder_all/layer_10_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8318dc8de69667736c26cab7d6d10fda1aaa927f --- /dev/null +++ b/transcoder_all/layer_10_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.10.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.10.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_10_width_262k_l0_big/config.json b/transcoder_all/layer_10_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..906322b7258bf0520be036ea518d522bb55677ea --- /dev/null +++ b/transcoder_all/layer_10_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.10.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.10.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_11_width_262k_l0_small/config.json b/transcoder_all/layer_11_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..346c382f61408d52cb30a6f0b1a0dc7e7e7b0d61 --- /dev/null +++ b/transcoder_all/layer_11_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.11.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.11.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_12_width_16k_l0_big_affine/config.json b/transcoder_all/layer_12_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e39ba67c81f0a8298b25f7578ab6ad0ee6dc9248 --- /dev/null +++ b/transcoder_all/layer_12_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_12_width_16k_l0_small/config.json b/transcoder_all/layer_12_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..64961dba482d9a8ca4fbf7eb5d4f07b0d13327d1 --- /dev/null +++ b/transcoder_all/layer_12_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_12_width_16k_l0_small_affine/config.json b/transcoder_all/layer_12_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..789fc3d22fa395a8af022ab3131bdfeff2980998 --- /dev/null +++ b/transcoder_all/layer_12_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_12_width_262k_l0_big_affine/config.json b/transcoder_all/layer_12_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a4149eccfd68e50478fb39ab641d4741615706e6 --- /dev/null +++ b/transcoder_all/layer_12_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.12.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.12.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_13_width_16k_l0_big/config.json b/transcoder_all/layer_13_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2f536f83380e63d2d1729896c03d9860fe54cab5 --- /dev/null +++ b/transcoder_all/layer_13_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.13.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_13_width_16k_l0_big_affine/config.json b/transcoder_all/layer_13_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..320b04333d9e62d4ab500a902680c468139dad2d --- /dev/null +++ b/transcoder_all/layer_13_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.13.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_13_width_16k_l0_small/config.json b/transcoder_all/layer_13_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..609dc0b3d134bfcd8cffeda85c5062937d072f71 --- /dev/null +++ b/transcoder_all/layer_13_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.13.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.13.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_14_width_262k_l0_small/config.json b/transcoder_all/layer_14_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6fbb5cf2977eab65269cfc2786d7f63076941994 --- /dev/null +++ b/transcoder_all/layer_14_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.14.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.14.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_15_width_16k_l0_big/config.json b/transcoder_all/layer_15_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..89c3fa0bedf701737eb4df625dbc538d7a10765d --- /dev/null +++ b/transcoder_all/layer_15_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.15.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_15_width_262k_l0_big/config.json b/transcoder_all/layer_15_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9e36d3f44c0c25972335eaa46e87a94f86b8086b --- /dev/null +++ b/transcoder_all/layer_15_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.15.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_15_width_262k_l0_big_affine/config.json b/transcoder_all/layer_15_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9f99b4f5355a02f768c885780c87b1922efbf046 --- /dev/null +++ b/transcoder_all/layer_15_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.15.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.15.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_16_width_262k_l0_small_affine/config.json b/transcoder_all/layer_16_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..33ad9e30f1f2c95bda76f6d2d3192f572ad814e3 --- /dev/null +++ b/transcoder_all/layer_16_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.16.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.16.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_17_width_16k_l0_big/config.json b/transcoder_all/layer_17_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e618220e3077b0722b237dfc018ec6b81f1b4198 --- /dev/null +++ b/transcoder_all/layer_17_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.17.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_17_width_16k_l0_big_affine/config.json b/transcoder_all/layer_17_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6810e6b4312c8df9ab7fd8d98833532addacfe99 --- /dev/null +++ b/transcoder_all/layer_17_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.17.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_17_width_262k_l0_small_affine/config.json b/transcoder_all/layer_17_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7dd0d13cfc778fced37e2b9f6aeb5732bf1db8cd --- /dev/null +++ b/transcoder_all/layer_17_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.17.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.17.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_18_width_16k_l0_big/config.json b/transcoder_all/layer_18_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5667503e28ad17aef37ab8db657ec7ed704c2904 --- /dev/null +++ b/transcoder_all/layer_18_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_18_width_16k_l0_big_affine/config.json b/transcoder_all/layer_18_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..342e0de233b831c72e16db5d5b6a6d0d634232be --- /dev/null +++ b/transcoder_all/layer_18_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_18_width_16k_l0_small_affine/config.json b/transcoder_all/layer_18_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3ee074ffec6f6830026f5644f5298ad048ccf426 --- /dev/null +++ b/transcoder_all/layer_18_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_18_width_262k_l0_big_affine/config.json b/transcoder_all/layer_18_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..893f029c7ce5fb0936d4418cd4673079d690cbb8 --- /dev/null +++ b/transcoder_all/layer_18_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.18.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.18.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_19_width_16k_l0_big/config.json b/transcoder_all/layer_19_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..01cc77f5dafc846c3e3b409958b1e98c44b0bd41 --- /dev/null +++ b/transcoder_all/layer_19_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.19.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.19.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_19_width_16k_l0_big_affine/config.json b/transcoder_all/layer_19_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..671a7a980e6f3f05923dadaf3e6e920b96a2cbaa --- /dev/null +++ b/transcoder_all/layer_19_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.19.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.19.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_19_width_16k_l0_small/config.json b/transcoder_all/layer_19_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca41501ab46de3a95fa725ff28a3630578c410da --- /dev/null +++ b/transcoder_all/layer_19_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.19.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.19.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_1_width_16k_l0_big_affine/config.json b/transcoder_all/layer_1_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dc481b0e45095307a2b19a6e897821447b859234 --- /dev/null +++ b/transcoder_all/layer_1_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.1.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.1.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 66, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_1_width_262k_l0_small_affine/config.json b/transcoder_all/layer_1_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2be45f8711b5712f423b1125de023fc86b80aa56 --- /dev/null +++ b/transcoder_all/layer_1_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.1.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.1.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 11, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_20_width_16k_l0_small/config.json b/transcoder_all/layer_20_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7f7ed61ddc21f19c1989e5875b581d3d13415962 --- /dev/null +++ b/transcoder_all/layer_20_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.20.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.20.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_20_width_16k_l0_small_affine/config.json b/transcoder_all/layer_20_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3b85264acf3aede5d339f2e287725198e870ab92 --- /dev/null +++ b/transcoder_all/layer_20_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.20.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.20.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_20_width_262k_l0_big/config.json b/transcoder_all/layer_20_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2ea68a42a4ab5469aeca2afb85f0886881843641 --- /dev/null +++ b/transcoder_all/layer_20_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.20.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.20.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_21_width_16k_l0_big_affine/config.json b/transcoder_all/layer_21_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c0c2c1fc07ce8e6ebdf4a207eaaa58fceef25548 --- /dev/null +++ b/transcoder_all/layer_21_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.21.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.21.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_21_width_16k_l0_small/config.json b/transcoder_all/layer_21_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..08d648025845b6e78cbd72485c4b1f704ac10e9e --- /dev/null +++ b/transcoder_all/layer_21_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.21.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.21.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_21_width_262k_l0_small/config.json b/transcoder_all/layer_21_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4e859d559254836e98b243ad6a1b272de1dbb57f --- /dev/null +++ b/transcoder_all/layer_21_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.21.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.21.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_22_width_16k_l0_big_affine/config.json b/transcoder_all/layer_22_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e60ae0ff4c8fa410e72ca906f4f55000b6cade4e --- /dev/null +++ b/transcoder_all/layer_22_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_22_width_262k_l0_big/config.json b/transcoder_all/layer_22_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7783d07f1d917ab41865b772eb8e7bcfc7b69cfb --- /dev/null +++ b/transcoder_all/layer_22_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.22.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.22.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_23_width_262k_l0_big_affine/config.json b/transcoder_all/layer_23_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..258b1eb2cc13c88b6e80d9199ad4a4584ceeab45 --- /dev/null +++ b/transcoder_all/layer_23_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.23.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.23.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_23_width_262k_l0_small/config.json b/transcoder_all/layer_23_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..05aab9403ef2855e3fecb5b43e7728158bcd8f5b --- /dev/null +++ b/transcoder_all/layer_23_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.23.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.23.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_23_width_262k_l0_small_affine/config.json b/transcoder_all/layer_23_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..34f5768b579bcef60291bbdd6ddfa07daf443275 --- /dev/null +++ b/transcoder_all/layer_23_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.23.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.23.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_24_width_262k_l0_big/config.json b/transcoder_all/layer_24_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8ff7553a2d3ef0da6d270284f26749ed06ce6d3b --- /dev/null +++ b/transcoder_all/layer_24_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.24.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.24.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_24_width_262k_l0_small_affine/config.json b/transcoder_all/layer_24_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..734b2ccd125ed4e2fcd47b37b72764408f47e49e --- /dev/null +++ b/transcoder_all/layer_24_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.24.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.24.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_25_width_16k_l0_big_affine/config.json b/transcoder_all/layer_25_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..99002154d0a720dcd08331c4a0ade1a2b48112c3 --- /dev/null +++ b/transcoder_all/layer_25_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.25.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.25.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_2_width_16k_l0_small/config.json b/transcoder_all/layer_2_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..1452f2d14967a2266707a6c4401505a817b178c1 --- /dev/null +++ b/transcoder_all/layer_2_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 12, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_2_width_262k_l0_big/config.json b/transcoder_all/layer_2_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b556aefbbeb14e9ef5247f7794a81440a395fe09 --- /dev/null +++ b/transcoder_all/layer_2_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 73, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_2_width_262k_l0_small/config.json b/transcoder_all/layer_2_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d2c67a6a5b155fb1ea50d1ee69e1af2b109fc6a5 --- /dev/null +++ b/transcoder_all/layer_2_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.2.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.2.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 12, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_3_width_16k_l0_small/config.json b/transcoder_all/layer_3_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..31cd7e03f5ceed8bf60152872d686f2d6f8d487b --- /dev/null +++ b/transcoder_all/layer_3_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.3.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.3.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 13, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_3_width_262k_l0_big_affine/config.json b/transcoder_all/layer_3_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0fadda2fae1865561229a9abe612d5b3d90466c9 --- /dev/null +++ b/transcoder_all/layer_3_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.3.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.3.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 80, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_16k_l0_big/config.json b/transcoder_all/layer_4_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4c86a8819eafa91c6907e1346c70f0a27833af32 --- /dev/null +++ b/transcoder_all/layer_4_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 87, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_16k_l0_big_affine/config.json b/transcoder_all/layer_4_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dbad3b600d9afe71850dbabab2046284dda1d3a0 --- /dev/null +++ b/transcoder_all/layer_4_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 87, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_16k_l0_small_affine/config.json b/transcoder_all/layer_4_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..585544df75eb52380edd3522e395f5c4eb3371e8 --- /dev/null +++ b/transcoder_all/layer_4_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 14, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_262k_l0_big/config.json b/transcoder_all/layer_4_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7b396d775038a2771bc7ea8b0a547c5acf4433a4 --- /dev/null +++ b/transcoder_all/layer_4_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 87, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_262k_l0_small_affine/config.json b/transcoder_all/layer_4_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0f24bd91f30284336aaaca3caba2759594a608e8 --- /dev/null +++ b/transcoder_all/layer_4_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.4.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.4.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 14, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_5_width_16k_l0_big_affine/config.json b/transcoder_all/layer_5_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f02a2ca4671b1b0c65eeb3b2b05eec5f8868c285 --- /dev/null +++ b/transcoder_all/layer_5_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.5.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 94, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_5_width_262k_l0_big/config.json b/transcoder_all/layer_5_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3798db753f0245d9ce5e84c8fce67d871ba3e399 --- /dev/null +++ b/transcoder_all/layer_5_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.5.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.5.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 94, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_6_width_262k_l0_big_affine/config.json b/transcoder_all/layer_6_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..48799af67b80e44d9eaefe4b0ab77847f02a0124 --- /dev/null +++ b/transcoder_all/layer_6_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.6.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.6.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 101, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_7_width_16k_l0_small_affine/config.json b/transcoder_all/layer_7_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..537ffba7911d5b095604b110c4781c81878397a0 --- /dev/null +++ b/transcoder_all/layer_7_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.7.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_7_width_262k_l0_big_affine/config.json b/transcoder_all/layer_7_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6644becb30367dd8abe2bddfbbe3ae68de8cfda5 --- /dev/null +++ b/transcoder_all/layer_7_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.7.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 108, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_7_width_262k_l0_small/config.json b/transcoder_all/layer_7_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..1fb4f99bf47a3b3c91c6c6243b2a8b25840e63b5 --- /dev/null +++ b/transcoder_all/layer_7_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.7.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_7_width_262k_l0_small_affine/config.json b/transcoder_all/layer_7_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3d22e49d51a03efa3e3b7c79a871ffd62b088d69 --- /dev/null +++ b/transcoder_all/layer_7_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.7.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.7.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_9_width_16k_l0_big_affine/config.json b/transcoder_all/layer_9_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f441ce1a7f5ccaac93e27c7bd9c4381a9f66d69f --- /dev/null +++ b/transcoder_all/layer_9_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_9_width_16k_l0_small/config.json b/transcoder_all/layer_9_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e02dfa4fcfb01b9c472c7b37275d8fd90ab9622e --- /dev/null +++ b/transcoder_all/layer_9_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_9_width_16k_l0_small_affine/config.json b/transcoder_all/layer_9_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a3b77645c9c794627ade0360b74e6a606c90d073 --- /dev/null +++ b/transcoder_all/layer_9_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_9_width_262k_l0_small_affine/config.json b/transcoder_all/layer_9_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fedf3d3133856cd3bafd3c1db9399796e549231a --- /dev/null +++ b/transcoder_all/layer_9_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.9.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.9.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-1b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file