diff --git a/attn_out_all/layer_0_width_262k_l0_small/config.json b/attn_out_all/layer_0_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..aab54b6430bc1d7de5d248035d9b7a1c04807ef8 --- /dev/null +++ b/attn_out_all/layer_0_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.0.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.0.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_11_width_16k_l0_small/config.json b/attn_out_all/layer_11_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..109f7e0ae7cde666499bb4e1157d442ed75a380c --- /dev/null +++ b/attn_out_all/layer_11_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.11.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.11.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 19, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_11_width_262k_l0_big/config.json b/attn_out_all/layer_11_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3009fb1af8502a91aa9b9fe3e28f01b13f9ac25c --- /dev/null +++ b/attn_out_all/layer_11_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.11.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.11.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 118, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_13_width_262k_l0_small/config.json b/attn_out_all/layer_13_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..1f4225a14ac6acec6e0896d0d9fec95f909ec4c6 --- /dev/null +++ b/attn_out_all/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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_14_width_16k_l0_big/config.json b/attn_out_all/layer_14_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fe98bb377814700a5b27869cd9d2a5dfcb769d9c --- /dev/null +++ b/attn_out_all/layer_14_width_16k_l0_big/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-3-4b-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..e869fdf845bba42aeb0c8a3a2794ca84bd27cb1b --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_14_width_262k_l0_big/config.json b/attn_out_all/layer_14_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b82e4ce3fa985339772e1906591773a2221b3795 --- /dev/null +++ b/attn_out_all/layer_14_width_262k_l0_big/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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_15_width_16k_l0_big/config.json b/attn_out_all/layer_15_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5f3fc5198c4448f95f1e387cb8f156b621719670 --- /dev/null +++ b/attn_out_all/layer_15_width_16k_l0_big/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_16_width_262k_l0_big/config.json b/attn_out_all/layer_16_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fdf958a70ff84dbe2f9ea59315da2a2073ca124d --- /dev/null +++ b/attn_out_all/layer_16_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_17_width_262k_l0_small/config.json b/attn_out_all/layer_17_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ba574b860568df80c85a42d78472eef8447bebc5 --- /dev/null +++ b/attn_out_all/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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_18_width_16k_l0_big/config.json b/attn_out_all/layer_18_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7f6c3f53cc4aa36bcb4ae454a5ebbbe50681cde --- /dev/null +++ b/attn_out_all/layer_18_width_16k_l0_big/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_19_width_262k_l0_big/config.json b/attn_out_all/layer_19_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0f6da3b5755a88bdb42af60170b05cbb64f9da6b --- /dev/null +++ b/attn_out_all/layer_19_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.19.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.19.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "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..1a15fff60bf6c07fa659cd0bf96830cacca97e5d --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_20_width_262k_l0_big/config.json b/attn_out_all/layer_20_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a20fe48f66939f2820ea117e335b3cf3c821508b --- /dev/null +++ b/attn_out_all/layer_20_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.20.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.20.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-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_16k_l0_big/config.json b/attn_out_all/layer_22_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..89358c23f21da7b7c4e1d0bb724febe3c5c8c9c6 --- /dev/null +++ b/attn_out_all/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-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_23_width_16k_l0_big/config.json b/attn_out_all/layer_23_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a736d7541b83c74c12bc97229cdc5b522455f5ce --- /dev/null +++ b/attn_out_all/layer_23_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.23.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.23.self_attn.o_proj.input", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "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..8155e0cf37d3edb48844e8dc59d9de2f7af7f7ff --- /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-3-4b-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_small/config.json b/attn_out_all/layer_25_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..26e7d8249ff90fe14c0e623fde00c5e66ef344b6 --- /dev/null +++ b/attn_out_all/layer_25_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_29_width_262k_l0_small/config.json b/attn_out_all/layer_29_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bbe3772ab3472239634f6b5e8e67e7f4955fd619 --- /dev/null +++ b/attn_out_all/layer_29_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.29.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_30_width_262k_l0_big/config.json b/attn_out_all/layer_30_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..73fbfeee30a8332b4f228dcef8af2af34fd47ad3 --- /dev/null +++ b/attn_out_all/layer_30_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.30.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.30.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_30_width_262k_l0_small/config.json b/attn_out_all/layer_30_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3e3a9390e5c75badaaff2fee84b2a30da061e802 --- /dev/null +++ b/attn_out_all/layer_30_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.30.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.30.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_31_width_262k_l0_big/config.json b/attn_out_all/layer_31_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..154ae3389d21fd248be736ca6b3b8c4aa994ceaf --- /dev/null +++ b/attn_out_all/layer_31_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.31.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.31.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_31_width_262k_l0_small/config.json b/attn_out_all/layer_31_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d81b0c4a99fb1d75acd4567afeb649dc4bbf6791 --- /dev/null +++ b/attn_out_all/layer_31_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.31.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.31.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_32_width_262k_l0_big/config.json b/attn_out_all/layer_32_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5c443127c220e03707912bb5dddeb7658828ad5b --- /dev/null +++ b/attn_out_all/layer_32_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.32.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.32.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_33_width_262k_l0_big/config.json b/attn_out_all/layer_33_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..917bc98e7356176894244c681873cef4dc35484a --- /dev/null +++ b/attn_out_all/layer_33_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.33.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.33.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_3_width_262k_l0_big/config.json b/attn_out_all/layer_3_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..76dd9130ea66db06909a46c6b28ab08937e0277e --- /dev/null +++ b/attn_out_all/layer_3_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.3.self_attn.o_proj.input", + "hf_hook_point_out": "model.layers.3.self_attn.o_proj.input", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 75, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_4_width_262k_l0_big/config.json b/attn_out_all/layer_4_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5f08c3805d509e5705178f3b851226d7e7b2226a --- /dev/null +++ b/attn_out_all/layer_4_width_262k_l0_big/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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 81, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_5_width_16k_l0_big/config.json b/attn_out_all/layer_5_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9a58210a76475bec8b43b1e3be2220ed039b7535 --- /dev/null +++ b/attn_out_all/layer_5_width_16k_l0_big/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 86, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_8_width_16k_l0_small/config.json b/attn_out_all/layer_8_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d8b2ec4f06895518b8cff7d2efc70e57b1796db --- /dev/null +++ b/attn_out_all/layer_8_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 17, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_9_width_16k_l0_big/config.json b/attn_out_all/layer_9_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd9d743fdccd2da279655db7b4b16db5adab26f --- /dev/null +++ b/attn_out_all/layer_9_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 107, + "affine_connection": false, + "type": "sae" +} \ No newline at end of file diff --git a/attn_out_all/layer_9_width_16k_l0_small/config.json b/attn_out_all/layer_9_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a98f4e13347f090839f2c5da4678951dea08c12c --- /dev/null +++ b/attn_out_all/layer_9_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 17, + "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..d66240679283ae28a8055fa8bfdb2f726646232c --- /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-3-4b-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_big_affine/config.json b/transcoder_all/layer_0_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b0cfa9b35e92acac3d8d170a28eab3045620cb3 --- /dev/null +++ b/transcoder_all/layer_0_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-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_16k_l0_small/config.json b/transcoder_all/layer_0_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fcbd5c33b5660c22d30334829b8d93110eee45ab --- /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-3-4b-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..2524bb8b8d0a74d0dac82f25379560153a630667 --- /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-3-4b-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..e5a1dd1632301c68d7c72d5b2676f0dc2a56ae1a --- /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-3-4b-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..29f4b5d6356420ccc59526759547ccfc55690374 --- /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-3-4b-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..29a4e6ae47d2a372d731e8435fcb7fd2716c80ae --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 112, + "affine_connection": false, + "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..7231a7e9b87e6a5376823743119bbeddeb504f51 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 112, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_10_width_262k_l0_small/config.json b/transcoder_all/layer_10_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..36469ab83c5552c41cd191246379264d95cffa63 --- /dev/null +++ b/transcoder_all/layer_10_width_262k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_10_width_262k_l0_small_affine/config.json b/transcoder_all/layer_10_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2e8aeec6b28dabc8f3ab84303bfd4d486b40548d --- /dev/null +++ b/transcoder_all/layer_10_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 18, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_11_width_262k_l0_small_affine/config.json b/transcoder_all/layer_11_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7222afca10f813619f460c1325aa82d1056c3321 --- /dev/null +++ b/transcoder_all/layer_11_width_262k_l0_small_affine/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-3-4b-it", + "architecture": "jump_relu", + "l0": 19, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_12_width_16k_l0_big/config.json b/transcoder_all/layer_12_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..81202b492e50899c119a5da84e50b9cf8f339fc4 --- /dev/null +++ b/transcoder_all/layer_12_width_16k_l0_big/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-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "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..02659e9f6f0b825001652fe60cd9454c085afc8f --- /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-3-4b-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..e197dbdec909ce558562cce4d84564ee095f0b6c --- /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-3-4b-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_262k_l0_big_affine/config.json b/transcoder_all/layer_12_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5199299ea4c203bda4377a6db5611ec8163d3fbb --- /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-3-4b-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_262k_l0_small/config.json b/transcoder_all/layer_12_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b6b7080d12124574a2780e5b718ad1eba0661b3e --- /dev/null +++ b/transcoder_all/layer_12_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-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_262k_l0_small_affine/config.json b/transcoder_all/layer_12_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0827314fabc0a2332ff5ac2fc3e6898521001698 --- /dev/null +++ b/transcoder_all/layer_12_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "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..ecf0eeb0b686db613225a3576a427b7bbd8203a1 --- /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-3-4b-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_small_affine/config.json b/transcoder_all/layer_13_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b271b49625cfcc3a863cde25a79fbea6d462ee25 --- /dev/null +++ b/transcoder_all/layer_13_width_16k_l0_small_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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "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..4bcf6abe1aeefbcf5cb2efa7e8c9340c792380a7 --- /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-3-4b-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_small/config.json b/transcoder_all/layer_15_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f93157adb5f924be6d23109a6dc10d39a9a2fdcb --- /dev/null +++ b/transcoder_all/layer_15_width_16k_l0_small/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-3-4b-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_262k_l0_big/config.json b/transcoder_all/layer_15_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..277b9ffb2ae1c6e3c3d5179f4ed43a22125d5a9b --- /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-3-4b-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_small_affine/config.json b/transcoder_all/layer_15_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8aa65475d2819b1f2cc33d785855b72a82c00186 --- /dev/null +++ b/transcoder_all/layer_15_width_262k_l0_small_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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_16_width_16k_l0_big/config.json b/transcoder_all/layer_16_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..da278566f5705c870bd6e8d7558f876e4663f259 --- /dev/null +++ b/transcoder_all/layer_16_width_16k_l0_big/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_16_width_16k_l0_big_affine/config.json b/transcoder_all/layer_16_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..55f1b88009415de696775045bcb5fd4d7b29166c --- /dev/null +++ b/transcoder_all/layer_16_width_16k_l0_big_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": 16384, + "model_name": "google/gemma-3-4b-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_16k_l0_small/config.json b/transcoder_all/layer_16_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..60eec6592fe3f0eeb5ca085d1cef17744f783f5f --- /dev/null +++ b/transcoder_all/layer_16_width_16k_l0_small/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_16_width_262k_l0_big_affine/config.json b/transcoder_all/layer_16_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b550f8c65ef566c3b210e994dec007c6b32e0fdd --- /dev/null +++ b/transcoder_all/layer_16_width_262k_l0_big_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-3-4b-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/config.json b/transcoder_all/layer_16_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5213a3d2f8abca86b05987c6fcc10d577b5b81c9 --- /dev/null +++ b/transcoder_all/layer_16_width_262k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "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..2eb6cc0135da2ca3a5ca795bd885a0b8b18be8af --- /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-3-4b-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..cdd8927e7538550c526a7bc0eb415bdffa123b6d --- /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-3-4b-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_small_affine/config.json b/transcoder_all/layer_17_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4bb62f1e8eb86ed64f947f586b1dabcbd63ddeec --- /dev/null +++ b/transcoder_all/layer_17_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-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_262k_l0_big/config.json b/transcoder_all/layer_17_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0f65d384f12c0f30785806c7dcacd39478200031 --- /dev/null +++ b/transcoder_all/layer_17_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-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..2cb9c9a59d21b8d5f0c5b5d22ce487a6e80a1a98 --- /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-3-4b-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/config.json b/transcoder_all/layer_18_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..77353b36e583ca74e20c6c11f76e936b77c28902 --- /dev/null +++ b/transcoder_all/layer_18_width_16k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_18_width_262k_l0_small/config.json b/transcoder_all/layer_18_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5d2d0459631d9d9e02c3e89d819de4d9636cfda6 --- /dev/null +++ b/transcoder_all/layer_18_width_262k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "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..50023de8d2cd61eb1076e6b6bca6f2b4f781cf60 --- /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-3-4b-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_262k_l0_small_affine/config.json b/transcoder_all/layer_19_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2cecf6ffe8e72c17370db04e32e313673337c813 --- /dev/null +++ b/transcoder_all/layer_19_width_262k_l0_small_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "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..fafd21866f39b423a5b51f8bf2dca4c1ffcc31f3 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 65, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_1_width_16k_l0_small/config.json b/transcoder_all/layer_1_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..de23339d2e735ca058d042836ffbca61bbebccd4 --- /dev/null +++ b/transcoder_all/layer_1_width_16k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_1_width_16k_l0_small_affine/config.json b/transcoder_all/layer_1_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b09f0230fe95c58c8eff53a5acabc3e8855a037a --- /dev/null +++ b/transcoder_all/layer_1_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 10, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_1_width_262k_l0_big/config.json b/transcoder_all/layer_1_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..90ac4fffea294b6f92b8a38c7b2b49fc8cf814fc --- /dev/null +++ b/transcoder_all/layer_1_width_262k_l0_big/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-3-4b-it", + "architecture": "jump_relu", + "l0": 65, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_20_width_262k_l0_big_affine/config.json b/transcoder_all/layer_20_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f5c445142079e1616ad3a0e9533ac9d24754d99e --- /dev/null +++ b/transcoder_all/layer_20_width_262k_l0_big_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_20_width_262k_l0_small/config.json b/transcoder_all/layer_20_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a76b2e2c5dab9bad8f1b1ba67dc393437e5c4c2d --- /dev/null +++ b/transcoder_all/layer_20_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-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_16k_l0_small_affine/config.json b/transcoder_all/layer_21_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..aa57f68e0096b965338ac1bc01f3cc9666aa487e --- /dev/null +++ b/transcoder_all/layer_21_width_16k_l0_small_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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_21_width_262k_l0_big/config.json b/transcoder_all/layer_21_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..82a740854b4f3a041c354ffaaeee7942207abcf3 --- /dev/null +++ b/transcoder_all/layer_21_width_262k_l0_big/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-3-4b-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_262k_l0_small/config.json b/transcoder_all/layer_21_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..81f4cf99dffa1f385938956e4eda94c0cf3ebcb7 --- /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-3-4b-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_affine/config.json b/transcoder_all/layer_21_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3c24ed1269a01d14f74c317046e3ddcc51aa867b --- /dev/null +++ b/transcoder_all/layer_21_width_262k_l0_small_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "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..9ab19977fbf457cf354f9f6f30e85fc9a0938003 --- /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-3-4b-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_16k_l0_small/config.json b/transcoder_all/layer_22_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5de2bf888e2a459fb0fb50b4602b9e24cccfb3ce --- /dev/null +++ b/transcoder_all/layer_22_width_16k_l0_small/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-3-4b-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_small_affine/config.json b/transcoder_all/layer_22_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d2defdfe64f2e21221b8487c4dd721a92e26ec5b --- /dev/null +++ b/transcoder_all/layer_22_width_16k_l0_small_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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "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..47436a23066c6fd4f4a7d20fb770496c5ae503c0 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_22_width_262k_l0_small_affine/config.json b/transcoder_all/layer_22_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..54d2d7190f14a39afb4f3179e5c2693e655a7bd8 --- /dev/null +++ b/transcoder_all/layer_22_width_262k_l0_small_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_23_width_16k_l0_big_affine/config.json b/transcoder_all/layer_23_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c4bec3c3e4f66fa3bb05aba0728ef04b2c20442 --- /dev/null +++ b/transcoder_all/layer_23_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-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_16k_l0_small_affine/config.json b/transcoder_all/layer_23_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c61c9dad1c9d5d8459882fd04f5471e256aefe --- /dev/null +++ b/transcoder_all/layer_23_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_23_width_262k_l0_big/config.json b/transcoder_all/layer_23_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dabd5965917b25cd5e120d16b3cc703de2651f17 --- /dev/null +++ b/transcoder_all/layer_23_width_262k_l0_big/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-3-4b-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..bc655d9c2a7dbeb5134fa90d1b30068cdc3862ce --- /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-3-4b-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..3f6386201b2d55053e82defaa598d52ec891d375 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_24_width_16k_l0_big_affine/config.json b/transcoder_all/layer_24_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..af283bb3994b6c395d903d5aea44e2f13c1e9158 --- /dev/null +++ b/transcoder_all/layer_24_width_16k_l0_big_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_24_width_262k_l0_big_affine/config.json b/transcoder_all/layer_24_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..984e40956b656b3789543fba37090fec47542114 --- /dev/null +++ b/transcoder_all/layer_24_width_262k_l0_big_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-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "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..cd93c9970381ee53620a06f207aef9d36c5214e9 --- /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-3-4b-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_small_affine/config.json b/transcoder_all/layer_25_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3041f4ad932c6145751ce9b7d249a620644c82a5 --- /dev/null +++ b/transcoder_all/layer_25_width_16k_l0_small_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-3-4b-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_262k_l0_big/config.json b/transcoder_all/layer_25_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..624a1ad98be19b1464560bec4fbd6b0571d86f1b --- /dev/null +++ b/transcoder_all/layer_25_width_262k_l0_big/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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_25_width_262k_l0_big_affine/config.json b/transcoder_all/layer_25_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..40352dcc64bf98c6b78551ea9a35967ce8f1edca --- /dev/null +++ b/transcoder_all/layer_25_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_26_width_262k_l0_small/config.json b/transcoder_all/layer_26_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..78185339f5fb46cfab06658226c34d3a0b354d83 --- /dev/null +++ b/transcoder_all/layer_26_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.26.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.26.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_27_width_16k_l0_big/config.json b/transcoder_all/layer_27_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0a07a6897c00fd3edb5f0db6084298a82461c1c9 --- /dev/null +++ b/transcoder_all/layer_27_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_27_width_16k_l0_big_affine/config.json b/transcoder_all/layer_27_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..075f58513648b9ef7e1f00529f533d022f7af4fe --- /dev/null +++ b/transcoder_all/layer_27_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_27_width_16k_l0_small/config.json b/transcoder_all/layer_27_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bc1a7d3d251d8e72ccccf727416bc986afcc81ad --- /dev/null +++ b/transcoder_all/layer_27_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_27_width_16k_l0_small_affine/config.json b/transcoder_all/layer_27_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..815abcb1d2f8b004d4f2c5da283f0ef79989cac6 --- /dev/null +++ b/transcoder_all/layer_27_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_27_width_262k_l0_big/config.json b/transcoder_all/layer_27_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ae14ba2ac4be5928a277eb746334f41d61a278dd --- /dev/null +++ b/transcoder_all/layer_27_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.27.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.27.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_16k_l0_big/config.json b/transcoder_all/layer_28_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7037d5d18f2b73aa94bbe7136b99ae9ebd41f63a --- /dev/null +++ b/transcoder_all/layer_28_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_16k_l0_big_affine/config.json b/transcoder_all/layer_28_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..41033a7dbb93cd8310f07f27f30a0789e294420e --- /dev/null +++ b/transcoder_all/layer_28_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_262k_l0_big/config.json b/transcoder_all/layer_28_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0bf9a93bc31c80fbd1aac14eb19a624b2a68ffbd --- /dev/null +++ b/transcoder_all/layer_28_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_262k_l0_big_affine/config.json b/transcoder_all/layer_28_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc464f847edd8e32a429f6d74f06d0fe58dd314f --- /dev/null +++ b/transcoder_all/layer_28_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_262k_l0_small/config.json b/transcoder_all/layer_28_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2438ea43f6f64b7e8d24cad40e4fd453a3c5eab1 --- /dev/null +++ b/transcoder_all/layer_28_width_262k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_28_width_262k_l0_small_affine/config.json b/transcoder_all/layer_28_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a1d752486087d869566d0843f13577b88d4a9223 --- /dev/null +++ b/transcoder_all/layer_28_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.28.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.28.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_29_width_16k_l0_big_affine/config.json b/transcoder_all/layer_29_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..666a2f5d1b92739662bf96f91b17d7dc870dd8ab --- /dev/null +++ b/transcoder_all/layer_29_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_29_width_16k_l0_small_affine/config.json b/transcoder_all/layer_29_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8efe2b865cdb7de37ba7ad94fb889e997e68fe17 --- /dev/null +++ b/transcoder_all/layer_29_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_29_width_262k_l0_big/config.json b/transcoder_all/layer_29_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c8a2cb9a7f2fbd4123a1d1da1f0846e740982178 --- /dev/null +++ b/transcoder_all/layer_29_width_262k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_29_width_262k_l0_big_affine/config.json b/transcoder_all/layer_29_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..51eecc59c9c48a8b7476e46b5ac218060ca2a6bb --- /dev/null +++ b/transcoder_all/layer_29_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_29_width_262k_l0_small_affine/config.json b/transcoder_all/layer_29_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8302daae315f9e0e9cb69de8ae1410dcbaecf0f8 --- /dev/null +++ b/transcoder_all/layer_29_width_262k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.29.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.29.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_2_width_16k_l0_big/config.json b/transcoder_all/layer_2_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..15b3f48b7cd086d5b39db561e3b63fd1b2deb41a --- /dev/null +++ b/transcoder_all/layer_2_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 70, + "affine_connection": false, + "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..37baa3ad9213153515805589f701245dfbbd360c --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 11, + "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..b02e8eb19776cbf4f9fb7978072257781d5f1d3a --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 70, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_30_width_16k_l0_small/config.json b/transcoder_all/layer_30_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..76331e9587091e4aed80ceaf007c705d76fc290c --- /dev/null +++ b/transcoder_all/layer_30_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.30.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.30.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_30_width_262k_l0_big_affine/config.json b/transcoder_all/layer_30_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5877ca3bc7def2f4871373da032f88c43063b8dd --- /dev/null +++ b/transcoder_all/layer_30_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.30.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.30.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_31_width_16k_l0_big/config.json b/transcoder_all/layer_31_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4a67f61e0dcbe0723a3e7d6f3bb7f0bf6918566f --- /dev/null +++ b/transcoder_all/layer_31_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_31_width_16k_l0_small/config.json b/transcoder_all/layer_31_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..09d98ba1acc0088d8bbd575a35e34164355a4c7d --- /dev/null +++ b/transcoder_all/layer_31_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_31_width_262k_l0_big_affine/config.json b/transcoder_all/layer_31_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..66ced36277f045eaf485f01775d111255bcb727b --- /dev/null +++ b/transcoder_all/layer_31_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.31.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.31.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_33_width_16k_l0_big/config.json b/transcoder_all/layer_33_width_16k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7ebe90d57c5ffd11b50013c1a48cd1457963c307 --- /dev/null +++ b/transcoder_all/layer_33_width_16k_l0_big/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_33_width_16k_l0_big_affine/config.json b/transcoder_all/layer_33_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47105152bcc590d5530a61c391f6d7d02ba0ffdd --- /dev/null +++ b/transcoder_all/layer_33_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_33_width_16k_l0_small/config.json b/transcoder_all/layer_33_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7aa81ce2ac5b78dc1359f6895dce4db45f4612e6 --- /dev/null +++ b/transcoder_all/layer_33_width_16k_l0_small/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 20, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_33_width_262k_l0_big_affine/config.json b/transcoder_all/layer_33_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cd7f3e3dd7496169ce6e2908d156e995053a5194 --- /dev/null +++ b/transcoder_all/layer_33_width_262k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.33.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.33.post_feedforward_layernorm.output", + "width": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 120, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_3_width_16k_l0_big_affine/config.json b/transcoder_all/layer_3_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cca23eb3b2fac24917ab45835c55fcdedd29bc2c --- /dev/null +++ b/transcoder_all/layer_3_width_16k_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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 75, + "affine_connection": true, + "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..d21b41fb18a595e072cf8126d7883c246f3dcd26 --- /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-3-4b-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_262k_l0_big_affine/config.json b/transcoder_all/layer_3_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..cff356e3e287e3f62c3abb4aa3da699f4663b9e1 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 75, + "affine_connection": true, + "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..c5e3ec8c16f6925213a56a7aca8ad6b1e8a2da2c --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 81, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_16k_l0_small/config.json b/transcoder_all/layer_4_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d79165d9615efa1289c98596878620deebd75f08 --- /dev/null +++ b/transcoder_all/layer_4_width_16k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 13, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_4_width_262k_l0_big_affine/config.json b/transcoder_all/layer_4_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f8981f9d6920a31dcba22b0b9c83eca356145273 --- /dev/null +++ b/transcoder_all/layer_4_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 81, + "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..4bbf962cd3098b235ddb1389cd5d8846c859c01d --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 86, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_5_width_16k_l0_small/config.json b/transcoder_all/layer_5_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..bfe9b7275b5b8e7dfab703a30f6432e5893c2024 --- /dev/null +++ b/transcoder_all/layer_5_width_16k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 14, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_5_width_16k_l0_small_affine/config.json b/transcoder_all/layer_5_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..65c49fb5c0be5ab8388c533488184a36176ce003 --- /dev/null +++ b/transcoder_all/layer_5_width_16k_l0_small_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-3-4b-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_262k_l0_big_affine/config.json b/transcoder_all/layer_5_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..93214a70bae966aae63f4ca3f3fb324054f8e849 --- /dev/null +++ b/transcoder_all/layer_5_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 86, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_5_width_262k_l0_small/config.json b/transcoder_all/layer_5_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..f90e6c9f80b95ad8b6fd3434a9fd4ad27e7e2718 --- /dev/null +++ b/transcoder_all/layer_5_width_262k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 14, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_6_width_16k_l0_small/config.json b/transcoder_all/layer_6_width_16k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c2c055c168462111e999b2a0505fcbe9f68e3e97 --- /dev/null +++ b/transcoder_all/layer_6_width_16k_l0_small/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": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 15, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_6_width_262k_l0_big/config.json b/transcoder_all/layer_6_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9b55bc838352688ba01dc33d0f382448690a3dcc --- /dev/null +++ b/transcoder_all/layer_6_width_262k_l0_big/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-3-4b-it", + "architecture": "jump_relu", + "l0": 91, + "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..76a154f975d3b5a5646361d1728e34d83da39d91 --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 91, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_6_width_262k_l0_small/config.json b/transcoder_all/layer_6_width_262k_l0_small/config.json new file mode 100644 index 0000000000000000000000000000000000000000..54fab4de9c9e3c6bfff824710bd64659a024b618 --- /dev/null +++ b/transcoder_all/layer_6_width_262k_l0_small/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-3-4b-it", + "architecture": "jump_relu", + "l0": 15, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_6_width_262k_l0_small_affine/config.json b/transcoder_all/layer_6_width_262k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..88f8b7857b9a7293c9a2edbd3770019e8c47863a --- /dev/null +++ b/transcoder_all/layer_6_width_262k_l0_small_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-3-4b-it", + "architecture": "jump_relu", + "l0": 15, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_7_width_262k_l0_big/config.json b/transcoder_all/layer_7_width_262k_l0_big/config.json new file mode 100644 index 0000000000000000000000000000000000000000..0a780c1d5a7e50162782cad9256aa1cc81bfa8a9 --- /dev/null +++ b/transcoder_all/layer_7_width_262k_l0_big/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-3-4b-it", + "architecture": "jump_relu", + "l0": 97, + "affine_connection": false, + "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..9dbbcf9bb4bab90e9d1a6a8d7beb6a2eb5c6226e --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 97, + "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..723abc922331ce119b1b05402afcc333e27caa4c --- /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-3-4b-it", + "architecture": "jump_relu", + "l0": 16, + "affine_connection": false, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_8_width_16k_l0_big_affine/config.json b/transcoder_all/layer_8_width_16k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..db1d95577ed3c8fba0e3a3a968efbd2ba50f4c85 --- /dev/null +++ b/transcoder_all/layer_8_width_16k_l0_big_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.8.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.8.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 102, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_8_width_16k_l0_small_affine/config.json b/transcoder_all/layer_8_width_16k_l0_small_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c751b67693f9e6d5c683e0fe86742026e70d3ce --- /dev/null +++ b/transcoder_all/layer_8_width_16k_l0_small_affine/config.json @@ -0,0 +1,10 @@ +{ + "hf_hook_point_in": "model.layers.8.pre_feedforward_layernorm.output", + "hf_hook_point_out": "model.layers.8.post_feedforward_layernorm.output", + "width": 16384, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 17, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file diff --git a/transcoder_all/layer_9_width_262k_l0_big_affine/config.json b/transcoder_all/layer_9_width_262k_l0_big_affine/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3bbb88f32f87d8ef6549280b3250a6842a7dc52c --- /dev/null +++ b/transcoder_all/layer_9_width_262k_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": 262144, + "model_name": "google/gemma-3-4b-it", + "architecture": "jump_relu", + "l0": 107, + "affine_connection": true, + "type": "transcoder" +} \ No newline at end of file