diff --git a/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cc590919b3664ef649943f730f68f13226e15522
--- /dev/null
+++ b/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d31762b7cabb032794cc894a130279f89bafe42b6283786bf6a6d3fda6a0ff8
+size 1060520
diff --git a/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c06c5ca79ea0e35918dd768ce0c2e36eebea0d50
--- /dev/null
+++ b/gpt2-small_L0_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..461c67a3f23898bc1e5c4a55262f03efd72aa76b
--- /dev/null
+++ b/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aed3581387beb655ac2fdadbffab7eef8d091a60d5785079d0dd68ab16efcac4
+size 1060528
diff --git a/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab5b2d623a1fc136e0a2f60610a3babce3e8adf3
--- /dev/null
+++ b/gpt2-small_L0_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..23b15196d0b11316f84ea9885eeabd55ca933b81
--- /dev/null
+++ b/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0179b44141b53ff885a30b3bef30e6ecae0515c1cf2674173fe92fdc551c7671
+size 1060528
diff --git a/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9dd95d3aadea73dc690e7bc50c5367046f5cb89f
--- /dev/null
+++ b/gpt2-small_L0_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bbd6e4adae9c433a3779a76ee905b7336648212d
--- /dev/null
+++ b/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b997b5b0e91cad93a03167f28fbd207b0958f3efc02fdd2d6c3dbdc71684287f
+size 1060520
diff --git a/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..87520b37f32a733a943205e499e040936b31cfc1
--- /dev/null
+++ b/gpt2-small_L0_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f9381bac67bd64c21323b0f8b3450301e047e084
--- /dev/null
+++ b/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e181ec24c0a5386c297aaebc06f8ec7658aa98cfeb0b9e710cfa72bc8f35566
+size 1060520
diff --git a/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..eafe31252322fbbc5adf598287e546b1a846009e
--- /dev/null
+++ b/gpt2-small_L0_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..29cf246f42c863874b84a61c9707fbdb156a29ac
--- /dev/null
+++ b/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42e96ccc94619475787fe0fc1ff0218dd7e725edcbe4f6326fe17a325d6c00ff
+size 1060520
diff --git a/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..f3906fd189c8aeed9e1a6ec45106a712c1edcf82
--- /dev/null
+++ b/gpt2-small_L0_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62bdf57336e3125aa00a88477b6952a0d9c6dc05
--- /dev/null
+++ b/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c78d5dca4516076a0d72f103b721181389823e84d1a7ecaf4900c26426fcf629
+size 1060520
diff --git a/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..428fd1c8d644373f0f1dc2dc2c22404f3ce4672a
--- /dev/null
+++ b/gpt2-small_L0_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3b08e95c892dc1294b1b01b417786ebda5b8dfde
--- /dev/null
+++ b/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8737d63fcecbc34cbb50ff0e28bb10dad9db601eb40d8ed02d7adf8aaf7d7bbc
+size 1060520
diff --git a/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..e29f67fc03cd6040eb57e41141cd9e595f077e88
--- /dev/null
+++ b/gpt2-small_L0_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..777323fb7b9597e04ad63dc143f47a1df67ce900
--- /dev/null
+++ b/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cbd4b3165531eeb2e10e29b7f59a758badf742863d9691c307e712dd21d26aa
+size 1060520
diff --git a/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..da9c2ba0c8d09dda8867ab908c63487478c2304b
--- /dev/null
+++ b/gpt2-small_L0_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a761fb072c3e501a1b62ffbda773ed21e47f2506
--- /dev/null
+++ b/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b30fede14c9eb20665f1db7b90d654cfde8caaec5f97402f55863386a459483b
+size 1060520
diff --git a/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1cdb07a6f4f55f3dc9887f49ce4b747b08785b42
--- /dev/null
+++ b/gpt2-small_L0_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c36044151fd85269a7448cf38539bee9c0c8731e
--- /dev/null
+++ b/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25632dee721b4a2f39c7e3dc33a6feab1f9a8801ab96250772b602937003e60c
+size 1060520
diff --git a/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..60a648fd8df49a51523d1ea5876da108a262ce24
--- /dev/null
+++ b/gpt2-small_L0_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..84cd5e83ed7de62e7efcdf61b5159217e591391b
--- /dev/null
+++ b/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ad42723234a1e2005fdc5cdbae3eafca6863741b7fcce0d877655ffa6b1c96f
+size 1060520
diff --git a/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..fbbe1263bd6187fa918cbb217b6294ae868ce750
--- /dev/null
+++ b/gpt2-small_L0_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 0, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.0.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L0H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cdfdad2388fc8bf6d025ecd9ed8f5e1d7ea07926
--- /dev/null
+++ b/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:537d6163e03194e60b64e8fe02a9280bb4edc70736e995e038a0a68565571d01
+size 1060528
diff --git a/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..553eda13ce661322750a388abcb158c39e8cbba7
--- /dev/null
+++ b/gpt2-small_L10_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..54f7d33c1cc3bd4cbecc06ee21a25d5912b9b16f
--- /dev/null
+++ b/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fd536192b2c6271f73c5bdc81ede4a4adaff06636b678d4100b5ac446b94404
+size 1060536
diff --git a/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..220e2bf070a6bc5e224a2923f04e04ad6c41a5f9
--- /dev/null
+++ b/gpt2-small_L10_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a32c1752e42eea2c5be00ce68e2c2c31ddd7a13b
--- /dev/null
+++ b/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2baf79a674a1c7a7b3a7ea3730b7bd920c9e5c89270fc57aa86804155958324
+size 1060536
diff --git a/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9607506ffd9c5932c38f00c98c9946907d3f5c96
--- /dev/null
+++ b/gpt2-small_L10_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3295cd2e95388165bdcb187166b4e45c22e9e4ff
--- /dev/null
+++ b/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ba5a8676df520ac811b2128a527b25b321606430727a7c2fdf6e8debb8505f1
+size 1060528
diff --git a/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..0913dba40fd806b57332d33d33ebe28ca76c1e28
--- /dev/null
+++ b/gpt2-small_L10_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b9d603d3bf5fef45ab9ef8ca0ffc6eeeb32d5569
--- /dev/null
+++ b/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4759273ca0b794e7828e06f4ac7da60a7b118a730cc3e37deeb1d02168d35c7d
+size 1060528
diff --git a/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8b6621439bced80fb23220b47b7383e133bea45a
--- /dev/null
+++ b/gpt2-small_L10_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8e985e6894a04130cba867e3770035bfe1ba7351
--- /dev/null
+++ b/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5d79a1f0e5d25d997267b8a2cbd8699a25f97ebfebd592fd75180b82c729cf1
+size 1060528
diff --git a/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3ba02c42a7f9a893f0b640239f4b8d1ac129f179
--- /dev/null
+++ b/gpt2-small_L10_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bae2828af70a44b865b7aad996c679d18d5943b9
--- /dev/null
+++ b/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03ed8d7ae732dd63c462f0fd5f17b6da98b2bc5cfec5edc409e58d6883ee1086
+size 1060528
diff --git a/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..fa8f1238580554c4dcd7934163b03fb7356c7585
--- /dev/null
+++ b/gpt2-small_L10_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..43d41bef11cd6620bfdc4f3120e9f8fded42c2f6
--- /dev/null
+++ b/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af1e9c9e8a5d64714afefb5a9aa474f26936e9331d95338f1e3761214855b1cb
+size 1060528
diff --git a/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..160074f62919529d4fc8e7b04380e5272a241a62
--- /dev/null
+++ b/gpt2-small_L10_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..588b0f516c1c8719c823c326e0eab5dbb1d666f4
--- /dev/null
+++ b/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7367a54a98f921ef1e87314b688f477c46fe9589d5a7821eedbb1fe056d0926f
+size 1060528
diff --git a/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ed1d729cd71f6b3faafd7a6be4f66a5a9d9cd075
--- /dev/null
+++ b/gpt2-small_L10_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4ff085c95442bd61cb1a6a0b01350b8af789a386
--- /dev/null
+++ b/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8daaeda74a44c0b4428e043e2d7f87b106297be9d5bb4e97f6bea230f770ae4e
+size 1060528
diff --git a/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3abe8c5b99144f59efedc825c78d0d68fda1503f
--- /dev/null
+++ b/gpt2-small_L10_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6eaf90308860f8fd69a850e9ed533d7c43bd9f98
--- /dev/null
+++ b/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c757d4bf89a3193632b27750631bdb62553b26a7da48d68ad3c4da074d5dd55a
+size 1060528
diff --git a/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..88c06ab99e96dc13a33d14c7f435e1b00726c8f9
--- /dev/null
+++ b/gpt2-small_L10_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7f3081b3372152ce4d6483c6aaea008176c1aba6
--- /dev/null
+++ b/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:447595578a4d51319837d6fb88799de201d65a8a6e24a96c68e254d34a011a17
+size 1060528
diff --git a/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..e2cc6ccfac21684f22f3148e35fe3f33d3acb676
--- /dev/null
+++ b/gpt2-small_L10_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 10, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.10.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L10H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..86fcb1e6738907322570b3ad2b85633ed12e1cf6
--- /dev/null
+++ b/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04153d14ce3cce9c7bb1efc4841732143d760916b3d37bd9ac35ddccf8dc6c14
+size 1060528
diff --git a/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..763cdd082fc1c6a13e2f5c956f541023eccf9b2c
--- /dev/null
+++ b/gpt2-small_L11_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2b3a9effd8f929b5f7b821f1e844869a489c7274
--- /dev/null
+++ b/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b622a922d5e0cc20e525100f024a14bb9f9f06d41d738c43b65b9221b422b782
+size 1060536
diff --git a/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..26158923a9204660e87f7653d59fb386fdc9c74e
--- /dev/null
+++ b/gpt2-small_L11_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7731fe8e29382c42b27481f09b077bb2dba1d9c7
--- /dev/null
+++ b/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a357a9f26a43d580386c79e20c42953f01c3e76895e8d3fbb95478933677e57d
+size 1060536
diff --git a/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c058c6442384d266a65644c65f396ae8cc59b339
--- /dev/null
+++ b/gpt2-small_L11_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b1ad7fadba24828d2953a6027a63698af1b5f367
--- /dev/null
+++ b/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3117e0b221aaa1b5c6ceb4b587488d313313df417826356ecc86eb1d3da4c710
+size 1060528
diff --git a/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1e4e539210c8d1b998775d7203a75a9bec7b491e
--- /dev/null
+++ b/gpt2-small_L11_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4744a867f2e4b744fffa9a4dae96c23c90f3f63d
--- /dev/null
+++ b/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cfa65d57fc70f03b47bde1dad456788046e8d973dabfc40931cf8c37bcff8d6
+size 1060528
diff --git a/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..aba7b635c4e168239f90fc23cc93122d7a45b07a
--- /dev/null
+++ b/gpt2-small_L11_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9187f52c384808a80042ef61c615b094ce138b85
--- /dev/null
+++ b/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91b923c41009fedbff5497aa15932fb2bbddae324668742bae7fa0d2c3753449
+size 1060528
diff --git a/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..20b245edbe451db1ab3f3455fcadfec7659a2f9a
--- /dev/null
+++ b/gpt2-small_L11_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c4b7926c13f6931d90f9da6e8faabfe14d9f8727
--- /dev/null
+++ b/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89138bf38176eded904902bb345827285ccdd083b401bfadd82dde22b0155bc7
+size 1060528
diff --git a/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..304dcec57ab5bfb8b8c531cb20598aa7aed8e808
--- /dev/null
+++ b/gpt2-small_L11_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d5cd8059e7c9c840a3db1b18c44524973b09d9bd
--- /dev/null
+++ b/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5728d5ccec9413570ba87cc327796096e20d29050812d690e92171386ea1bfe7
+size 1060528
diff --git a/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1effbd9d7c9d5e83598b72a7095c9e8728309f31
--- /dev/null
+++ b/gpt2-small_L11_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e38a92f94ddfc0b71a3462dd4de2e117ee9f560c
--- /dev/null
+++ b/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f0c0dfd9cbc9e8e70bb823d65a97cc4819ac3147248e455a1ecc6d06cf25b0f
+size 1060528
diff --git a/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..f79e740fa8c2925041eff53b5bcd37927d296e54
--- /dev/null
+++ b/gpt2-small_L11_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..866f8fd30f32b8f27c4d9dae35295879fdf8685f
--- /dev/null
+++ b/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74500c4eb5c5cd8595e0b8a41866cb9d2df8ad330412ce5b9d58997bc31b3161
+size 1060528
diff --git a/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..dce48e36f3774fcd3b726f9b910f1a8ee56507c8
--- /dev/null
+++ b/gpt2-small_L11_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..60041825e5dea15474d74b5cc4c0bfb8fe891984
--- /dev/null
+++ b/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9725899c3fdc9bb50c4eb2f3b4d00cadc6940c92c6ce9812cc84afe19c29a6c6
+size 1060528
diff --git a/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..44318e8658412e0f241046bd903274a234d9877a
--- /dev/null
+++ b/gpt2-small_L11_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0b5c827c68bf880e6fbb08fc5bd96a8fa52332ad
--- /dev/null
+++ b/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:039562af3addabe4ffa75e92fdaf8869244291fd1e0b29f80caade9fe9553879
+size 1060528
diff --git a/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..99e5cdb96f4850e08ecd631b37f9fbdd7a7d45db
--- /dev/null
+++ b/gpt2-small_L11_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 11, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.11.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L11H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d2c6b6de46d5f68627d238c72c9bc503b2da0581
--- /dev/null
+++ b/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ac819268ed9e3b9172701ba5ae8c1f2e6ce918e8670f932b57c2d200a7e05d3
+size 1060520
diff --git a/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3c12f72acbc367dd26b80978d8d324618c560f39
--- /dev/null
+++ b/gpt2-small_L1_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..16795a6aa4f14321fd9f3af28e3df4cbdc1f2707
--- /dev/null
+++ b/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3b1ae4826ff6aa6cd264b250dffde4b8b2eb765412754fbd61d8f57a016c74b
+size 1060528
diff --git a/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..985861fa282ca89e80f94560a110cb74fb95be2b
--- /dev/null
+++ b/gpt2-small_L1_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f3cea9bbe84a054c0be9af570dc8d8a14a3f0ecf
--- /dev/null
+++ b/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aa2f12d9228093b58826c45bb14e5036ce0054540bc4cb8994e62b942b1e67f
+size 1060528
diff --git a/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..6338a493d00be4c9ba6ff5f39f9736552d680b74
--- /dev/null
+++ b/gpt2-small_L1_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..09f6404fddf5613d7c4eedb3a974773f366bb857
--- /dev/null
+++ b/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62652464555264e5bb656ad44f86c263a08c96bdd420824539a0434a7059b5d6
+size 1060520
diff --git a/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..f330822eea449b9f1ff83b08918ce057abeab01b
--- /dev/null
+++ b/gpt2-small_L1_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..61ee0f37d269b37cecc21bd878e63290f7360a3c
--- /dev/null
+++ b/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b4c830311464a73d27809076f4ca515212ac4c7bdefef878a16a47e79b2587f
+size 1060520
diff --git a/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..24666f60ff74502ece28ebb113d0bda32af31478
--- /dev/null
+++ b/gpt2-small_L1_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c91de3c980c3c542c65a12c0a58f04ecdd0a124e
--- /dev/null
+++ b/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4d72549c2714ac42bcb338603602fe505f2d2d5985ab5ebe70d64b4c3f7ac77
+size 1060520
diff --git a/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..de79d5a662cd57779212b54adff4fb9a0905a712
--- /dev/null
+++ b/gpt2-small_L1_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f34835ad5b06a555b276253be85e5d5e59aa1028
--- /dev/null
+++ b/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:876d13a75fe9b50507209f4e1da5dee007589fbcbaf426cb08b4f425ff6e3db4
+size 1060520
diff --git a/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..aee20c0bf5932b5721d2ec7a833bbd9a468dc6eb
--- /dev/null
+++ b/gpt2-small_L1_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2ef36f5337d07cff9ad0e9ae5692ba29ac60fa8a
--- /dev/null
+++ b/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44e8c40c0991f80b73d730349e46c83c662f4bc5b0072bf0271fbb567b4415b3
+size 1060520
diff --git a/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3aca055313d571a14d54e1c1842d8d3f2d9ef6ce
--- /dev/null
+++ b/gpt2-small_L1_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a1a90447b66627166b2ef9073be21a72970c6f10
--- /dev/null
+++ b/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f10c84aec99b7ffe5beb66398b9d3269d046934557fbdf1351d6de96df00d28
+size 1060520
diff --git a/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..6db7c834b19892c05cfb38bc35b7a41684e7fc84
--- /dev/null
+++ b/gpt2-small_L1_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7dbf44c8636038cce6f5095f5af9635c273ca947
--- /dev/null
+++ b/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e002fc30a9db9a767da3ae5c95e7ac2ea18c4bf5b9292630cef57b2f2471b5bb
+size 1060520
diff --git a/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..bf76e9fed6d4c11084e6ee59a80edbee4cc8398b
--- /dev/null
+++ b/gpt2-small_L1_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c54d4c3ebc7f7209822758de3deaa263961dd277
--- /dev/null
+++ b/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66d0237b593b7ffee569422d859e8dc475e029f5a5fa886867262984b5d9274e
+size 1060520
diff --git a/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..afe0f39ec377420e6dc92a726a3418af9ea74185
--- /dev/null
+++ b/gpt2-small_L1_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4f13de59fc66a1afeaeaf6b1c4fc1cf103985d98
--- /dev/null
+++ b/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ed1567ac9480fffb4e67dcbba083ac442d39328ed20b259e895186ab524f550
+size 1060520
diff --git a/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..040b5785f517745c729bbeb7299c323e7a776df7
--- /dev/null
+++ b/gpt2-small_L1_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 1, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.1.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L1H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..878fe9c608eb7466b7d087e8b8ab68a3dae81694
--- /dev/null
+++ b/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42fe335b5427540f7f27d3dc220ee7814e7db4f3e63cc3743d6169f55df1820a
+size 1060520
diff --git a/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef8a917ed7d162ebd725e6e39657777ead7d8394
--- /dev/null
+++ b/gpt2-small_L2_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c482f86b0e20d35dd68dec6326999dda308bb1a9
--- /dev/null
+++ b/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baa732e5e4c62ffaa26148933c29d260e3326ec6784fad825425de36d935396b
+size 1060528
diff --git a/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..84c1e5baf9b1201360138afe99f487ddce5d19dd
--- /dev/null
+++ b/gpt2-small_L2_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e64b9a6b4a6dcca9b85492aea3f341f94998c6f
--- /dev/null
+++ b/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3474c02f7fee286aee586e30a3cccbf016ffe8165df7a504f9c7915205a9d4bb
+size 1060528
diff --git a/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..217ffd35490b6b0d075d5d9c63a284165ccc29ce
--- /dev/null
+++ b/gpt2-small_L2_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2ff85f6f2b71069ec4ed410ffe4673002eb01b89
--- /dev/null
+++ b/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f30098a4a288d46177c6069f9a1a60e4efa8a7da8031328991677cfa8ba55f77
+size 1060520
diff --git a/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ec2c5c8ca9c1a919ae5872b7e8bdb489ec852c7
--- /dev/null
+++ b/gpt2-small_L2_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..70d8198ae61e43a64735aa61b5942e5e4ea00055
--- /dev/null
+++ b/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f4380c52ae5336fdf5d8618261ab3577d1ea736f88db4cb618dcd4ed6ada91f
+size 1060520
diff --git a/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..14e59e21b065730223d97ee07298e62d85e416b9
--- /dev/null
+++ b/gpt2-small_L2_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..30ca6a6d0c06b873ec98212d95726b58e07e10e7
--- /dev/null
+++ b/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:620faee6fd30ecf8e849d293e81d5b8bf33ce53c999229e2999a72302785f924
+size 1060520
diff --git a/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..19e9181a049a790aa337b32b925ab0a08df633f0
--- /dev/null
+++ b/gpt2-small_L2_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..40fc5b49f8c6f30232ce8f2a64879e5c979279b1
--- /dev/null
+++ b/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f51a164d6f94fba3a228c22c0dc488a7402e022225cfcb60b630cfc631d658c
+size 1060520
diff --git a/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..61c9be26fe60e4867f229238bb288afcd17039ee
--- /dev/null
+++ b/gpt2-small_L2_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2bbf5496abcda9c274f4755ac1776ed57b526660
--- /dev/null
+++ b/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79c8014329dcc71cf237c93443b831590863aee2c18cb2a9d082b436a62be4dd
+size 1060520
diff --git a/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8df1dc0e784723347bbf0853907a7e64f97fe77d
--- /dev/null
+++ b/gpt2-small_L2_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a37296935f86faabf60feeb9f93fa224e2d869c6
--- /dev/null
+++ b/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:058c0576516f86ad8e72d2e5b89b96c62997f9f880446245140c008d826d68d0
+size 1060520
diff --git a/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..276eb8a330daaf1ac11b65ce0eb6c1a83e7262ea
--- /dev/null
+++ b/gpt2-small_L2_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..45512c4307cd3de3bbe1a7fbf3690fb798293f5d
--- /dev/null
+++ b/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8d54daf20d54cdd0d72eb7f2bce4ac79422e3bd188aa832538357b69e95e93e
+size 1060520
diff --git a/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b62d4b6430be22b4f2854d97042fb62a9f614183
--- /dev/null
+++ b/gpt2-small_L2_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4690cec5f1873965ce8e4fdf52f2d5bf87ee5217
--- /dev/null
+++ b/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fa7a3f684c111a7788a5a00940a54ce0b056c3cb2885b10b44a31b48ba246f5
+size 1060520
diff --git a/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b3aa2c0bd34525b6b62ab4659a00bef38066cc26
--- /dev/null
+++ b/gpt2-small_L2_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2220a2bcfdf729dd1f3c7c4ed3cafb0c7e30346b
--- /dev/null
+++ b/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad3d7ea84ec6b2e43dd2bc33fd3329c55b2d82ad3f68271939364c7c10c53f47
+size 1060520
diff --git a/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..74b76950e54bdc3733fb1bcea943cb6fb8c2993c
--- /dev/null
+++ b/gpt2-small_L2_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 2, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.2.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L2H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..51ca999ee4ceb228f93f923c911fe0210ff4fe14
--- /dev/null
+++ b/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b801bada30ef0b52295f7279db7572c5a85dacce213cc920ae3de178ba9d35b
+size 1060520
diff --git a/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..14c39e73f87f47392265e95f58ea0ed02155e743
--- /dev/null
+++ b/gpt2-small_L3_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..16e10df464acb21743400d6f8d490bc144794aa3
--- /dev/null
+++ b/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1ce2c799e901071372cd5be8b4f39222ef048f524fdbd38f298f2b3cf9c52de
+size 1060528
diff --git a/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef8d1228a108b8d5e18d216045365e90b56a45c3
--- /dev/null
+++ b/gpt2-small_L3_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..03e8690fa25d239b3d0902330b5f58bc5d7c7e70
--- /dev/null
+++ b/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:398cb6d32f6aab53ffd38dc2fec572b248f84a31e9924dcfc9383a61231af8f3
+size 1060528
diff --git a/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9cf828d19a3e0f9bda90daf314ab11af9b389b1e
--- /dev/null
+++ b/gpt2-small_L3_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2989508d917eaf701ab859b7684940df6e33970b
--- /dev/null
+++ b/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4d105aeda43a4b0fbb1d93a5162451b40e757954efe825e28a85c8e24a63d45
+size 1060520
diff --git a/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..a972387b073266a0cc43e9deabc6ded92d4d1b4c
--- /dev/null
+++ b/gpt2-small_L3_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c58d526bd2256fae130eeb66b75cb6623c8df112
--- /dev/null
+++ b/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed281ba5bce0a37410e292c341ac2ba6c812d971dec9788cc764246196b4ef72
+size 1060520
diff --git a/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..fc513d6ea161624c9e861ff240f7823ec3452c4c
--- /dev/null
+++ b/gpt2-small_L3_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2afa50fc6c82dd9db00e810db20211393d49871c
--- /dev/null
+++ b/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14d87401a187f417a6477c8bb72031450d9e30feab6079bfc8bc2e1c5efa01e9
+size 1060520
diff --git a/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..faaf9b9874b8e5dd35fa1da56d86d13a7e7b182c
--- /dev/null
+++ b/gpt2-small_L3_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4530ccfb228cad0f65949f3136af1c6bf515731b
--- /dev/null
+++ b/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22bc2c200c1ca82c2525ff8c12dd62bbec4519460fc33cd369cdf4434d373e9d
+size 1060520
diff --git a/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..fbd6bb7ce17a10c80823c0f1aa7f075685452e48
--- /dev/null
+++ b/gpt2-small_L3_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fd7f887d9aa7c36e63b05388b63056354df247a5
--- /dev/null
+++ b/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7247e85e054eea507c2604a6991b311bc2757730123062d4bf545ed1c1ced8cc
+size 1060520
diff --git a/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..d7b1a41155e9f30fdcdd65823b709de22a034738
--- /dev/null
+++ b/gpt2-small_L3_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..28860a9bbfd6ff8772e8f9d79b281fa4eaefbbf4
--- /dev/null
+++ b/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a0f1cfb384c30acb9b0d9521e48ecf95541a70351a03f34dc33d0c7192fe009
+size 1060520
diff --git a/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..2673dda4c9875f4e46e7c6420f32623c326dac81
--- /dev/null
+++ b/gpt2-small_L3_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ca2a4a572e130afe76eaa706b2f6743e94fe2e69
--- /dev/null
+++ b/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ae954ab2d8dd84969cfe6cce4c61fd53f0affc61777b9f0f175143c9253233b
+size 1060520
diff --git a/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..0226d436321768371849cd8ada6aa1f1083c00bf
--- /dev/null
+++ b/gpt2-small_L3_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d7360f1821f79bd580abc8ae9f07cd3668f7850a
--- /dev/null
+++ b/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10ab00948b84fbe9a05781a1cd3cfa3d14a8006e93fc425fd5d91ff65188a515
+size 1060520
diff --git a/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8bfcdba9fd5be811c992ab70d3873aded0c99613
--- /dev/null
+++ b/gpt2-small_L3_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0b05b61c44eebdd9bc86214bcbc21733e72b93c5
--- /dev/null
+++ b/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:595035a29b70ad2c2b6e1f30dd2734a5402383300807ad0443ff686201228df8
+size 1060520
diff --git a/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b75dfd7d04951d4c232aabbe3bda9879b02f5b5f
--- /dev/null
+++ b/gpt2-small_L3_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 3, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.3.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L3H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e35a384d94b3360c5f784784ce93a6e849372925
--- /dev/null
+++ b/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2abd1c8ff824003654158a9d1fbf1f67ff8f471aa0a7450118600c60e60ce9ac
+size 1060520
diff --git a/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3e49967a9739781504f7f60c3d4b2ee12665bb9a
--- /dev/null
+++ b/gpt2-small_L4_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f44a205c785d9058bcec4d2a752c592b20a13339
--- /dev/null
+++ b/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1be902129c875d051dcb48ab126497fa8a484bf3cf79659b9ea90872c52c8f51
+size 1060528
diff --git a/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..dd113ff502f3806daf65aa6bcbbd55dde77a5424
--- /dev/null
+++ b/gpt2-small_L4_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7d05dccbad60ea59f0406e35f1eb8dfa7c68e373
--- /dev/null
+++ b/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5392bfe24155a1873ddc84aed9a21e2bc11704e92be6b4f73c97cbd384d99bdc
+size 1060528
diff --git a/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..0d074bfe0f48ab4a8d51d7775a9a9a28e3050edd
--- /dev/null
+++ b/gpt2-small_L4_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..48d73fb4b93638e561a1c0b0ea96a3e2b32c21bf
--- /dev/null
+++ b/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfb83e46c00cb12341bcfddb01bcbeebc9e3a9785f02039282cff8533e02752f
+size 1060520
diff --git a/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1e6f5a2821c001e9fd38ee6e41f4fddc68b81cb7
--- /dev/null
+++ b/gpt2-small_L4_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8bec0d0a085619babe440ec0043e063110e24d85
--- /dev/null
+++ b/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40456350c7e6a6964c3943f280bc2addab4a1a0ce858b19c59818a41d59db2be
+size 1060520
diff --git a/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..57be268e5092f6671fa93c7238c3dca17f1877f9
--- /dev/null
+++ b/gpt2-small_L4_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2dbb546b568725817d16b896f32612124fc47a38
--- /dev/null
+++ b/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cc499b8adf48f7ec53ad99f3e8da51405cd07a63158c183308380f4f767cbd0
+size 1060520
diff --git a/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..d945224b105ad36defe21d2b0b9c513a78196c2a
--- /dev/null
+++ b/gpt2-small_L4_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..17e1f504a731bd3778461a27cdfa7f264bd50abd
--- /dev/null
+++ b/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08c3e6be48c8faf54aea923a7488c3b33e22b66366788e92e4ff61449289d2b7
+size 1060520
diff --git a/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8522be1b658c6e678f14d9677e447310ec39f4b9
--- /dev/null
+++ b/gpt2-small_L4_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eef26ad34dacf4e16a0e7171b48f73f4353c3bf5
--- /dev/null
+++ b/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57d1a3671dc2f988f5466e29da57e3e8e51379f4f22246fff539633ca2c30889
+size 1060520
diff --git a/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..aafc5d4610c0e9b25a5fd4d197c8c80801577845
--- /dev/null
+++ b/gpt2-small_L4_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..70a04b1348f6fcc8d079931a1f978f8f5f2bf983
--- /dev/null
+++ b/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f348695ab4e3c7b69268fc06dee4d3d6a72c86b86eb93d706fb9710da5cc9e73
+size 1060520
diff --git a/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..f7bc9fcef3124d14a252182e1146c3735e38daf0
--- /dev/null
+++ b/gpt2-small_L4_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f85576a1c2f915e4e36b1ec2797cd71132e8c2e0
--- /dev/null
+++ b/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48f724d2cd350d7dd325bd96369d5f01ddcaf376bf4bd71bdf94bd56633f2ab6
+size 1060520
diff --git a/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..612176743fe9aa8642c5551f4db7bd50d5ab552c
--- /dev/null
+++ b/gpt2-small_L4_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1a15c425f59300975b80a012c6723af5c1557b3f
--- /dev/null
+++ b/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a43d84d1f99f9c9f1684021821c55dfb50ecb6172694aac64c7dc18a564011b
+size 1060520
diff --git a/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9115ab25da1ae471e2e99db7b0c5993e80c6b5dc
--- /dev/null
+++ b/gpt2-small_L4_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..decbac97f5dd17ebca8415927d5fd19bbd174ca0
--- /dev/null
+++ b/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ef5426d9196b7c71e37804e6185010692de36ed03cba8779047dcbc01550e66
+size 1060520
diff --git a/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..6a51928fa1df18de04f60eec4b8dd001467f5527
--- /dev/null
+++ b/gpt2-small_L4_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 4, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.4.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L4H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bfe745afcf1fba5dbc63d476f5b26cc9f4a51e0b
--- /dev/null
+++ b/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36b32230b1262632a746208069657b2547fd88a867f2f5565c194d6ea6d7f6de
+size 1060520
diff --git a/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c9e46332fae8f38e02169c2c560fdddb7ff15a0b
--- /dev/null
+++ b/gpt2-small_L5_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bc06d19d0cbb342d5446b43fe192f760b0ac2281
--- /dev/null
+++ b/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ecb2304f1a9753f03fd8b7186a97ae9abbc12231e19546777da9c9b7076bc4b
+size 1060528
diff --git a/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f1caa0866ec860082e6ee6f9ca3b325575f8bc
--- /dev/null
+++ b/gpt2-small_L5_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..003224e12409ab7514820f8e9de81e75a223ea21
--- /dev/null
+++ b/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecf499997fd8f369da654b2faba297b3129c63b9e4c217ddc5476cf0c0feebc3
+size 1060528
diff --git a/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8ba296aea8604f00398907d0b0fa0fb92a9fa050
--- /dev/null
+++ b/gpt2-small_L5_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2dce085d54ef8f87a78b345aca8db1e4cf67f2ab
--- /dev/null
+++ b/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d8c7942be1e0411e7126ecb182063eef55d65450a38488c765bad88f523aa2d
+size 1060520
diff --git a/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8f6428a62dccc1a277927c375a44be56b4faef96
--- /dev/null
+++ b/gpt2-small_L5_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1656a32442b129105e7cd12f537123fd4d718d66
--- /dev/null
+++ b/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:685dd19a4b1d27eef9d3f72265d5a047f31d9bb50229a38cbcd24ebc01463a65
+size 1060520
diff --git a/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..7b90bcae58b1693c12241382a93d49767db4ca4c
--- /dev/null
+++ b/gpt2-small_L5_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a24a3aad1766adfee9545fbcc06c07904ff5513a
--- /dev/null
+++ b/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:713a591535c5b516e56f225b6fe8ca9f6a460188d5e70e2c0adfbf5882047032
+size 1060520
diff --git a/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..7896542be58f1c35568af31c37a357c9d19fb05e
--- /dev/null
+++ b/gpt2-small_L5_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..de3d9a26ebcc4d6de83a62eabab95b13d280e1ae
--- /dev/null
+++ b/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f93b0316a328b3061a90ba48f293c74ecacb6a674e4faaf82ce150900fa83e0d
+size 1060520
diff --git a/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..381c306e76e585dfd84199117fa4d67ffed6c853
--- /dev/null
+++ b/gpt2-small_L5_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5657611239bbc64293792132bcf8916fb6b6f086
--- /dev/null
+++ b/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:848a6910cce04e1195e91e0da0cecc353be6d5b2358b8ff6158f564c693e58aa
+size 1060520
diff --git a/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef691ff2e3316155eeeaf195ea814fe624d538eb
--- /dev/null
+++ b/gpt2-small_L5_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e4516afab796b8ee794d7a397551593345d564d
--- /dev/null
+++ b/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21f2fc4c3abbb46e4aacfe052132f36ca0054ca40fd8570d19f91e072a8f853d
+size 1060520
diff --git a/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c6f3bcae13a544a75853c90f3321e054652ab9af
--- /dev/null
+++ b/gpt2-small_L5_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..89caac9b815d9d2b791623d22cb578b40190de5d
--- /dev/null
+++ b/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a24362a813e9239884a6aa82e6bda5e296620f37d39b3aeadb3f9511bb08c036
+size 1060520
diff --git a/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..73d50e534cf4dbab08d9b9b893314ea4be2043d0
--- /dev/null
+++ b/gpt2-small_L5_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..92d1642df47661c2ed0c85c1ff127219ed72e8f7
--- /dev/null
+++ b/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73d929de4c9ec553d3d1cb222bec16eeae30383f6bf2f4fbc7d1204b692e87a1
+size 1060520
diff --git a/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..85d39dccbe1515bea16812ddad92e71700ea49d6
--- /dev/null
+++ b/gpt2-small_L5_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4fc71310302aa820378e92a3176bb8bf077e121c
--- /dev/null
+++ b/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b940e0a26fddfa171353b42b457c4f9ccd2eaeffbc42cc4df0bea38bd185320
+size 1060520
diff --git a/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..740f050385703632c57630d53dc2561c9e3817d3
--- /dev/null
+++ b/gpt2-small_L5_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 5, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.5.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L5H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..314c571187db89f29a049a01adb23c4052e5c01a
--- /dev/null
+++ b/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2414994e516ff754b34a78f9aa8615775f8f4b4447584c7c0f8fbd8a1bdef04a
+size 1060520
diff --git a/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9e3af7d19eeacea93b5bd1be446cc73837142233
--- /dev/null
+++ b/gpt2-small_L6_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d7595b4c11ec94ad001caaa94c3033b351b6cd9a
--- /dev/null
+++ b/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ce19c3939ce8b5f88608cb263232dea4082108eb37b6360d73c70b57f2813d7
+size 1060528
diff --git a/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b74b1b7de445256e72cf746c5aa96370146020ea
--- /dev/null
+++ b/gpt2-small_L6_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..525784bb80f9fbabb6d493ff0e87d5a5204c5bc4
--- /dev/null
+++ b/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49fcf42ebe2e35319a29819f5476f3e9efa6d5f26cb563f3f706af73c93bd568
+size 1060528
diff --git a/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..65d69f4082cf4e84b05421cce20b413fb3029e37
--- /dev/null
+++ b/gpt2-small_L6_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eb66785f9a9807ce7c885b400360605b4bc210ee
--- /dev/null
+++ b/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c905d0bca270fba5533bb2922caf561de7ca0cc564df44a5cb3baacfe81cf64d
+size 1060520
diff --git a/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..522e9d63ed6b71522cefe7966dabea4ae97c9d34
--- /dev/null
+++ b/gpt2-small_L6_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e47b6a19ab813c587b0f0238468996f0f84b326d
--- /dev/null
+++ b/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf67bddfed9822edaf737ea37c1202f40f0d5955b4d62c1cc70d2efd69cd17e7
+size 1060520
diff --git a/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8f7b303ae7c9c5b610c15ca2450c7c5d7707a717
--- /dev/null
+++ b/gpt2-small_L6_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..09e3bec130d3952a401fbce898360db83678bc3e
--- /dev/null
+++ b/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13fb663a8b8b070dd6f8075151d0f3d78c2c5d05cc21bcd4728dec66c440a1da
+size 1060520
diff --git a/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..a5c3d42a729443621196f53e649ebac44cf72f38
--- /dev/null
+++ b/gpt2-small_L6_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..05458be8f6e1ccdc560bf7c65b124e53a7600e1b
--- /dev/null
+++ b/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:380ca21a6c8c8e28c7ef4b9cecef6041365d1f8e6f37a3ca899ea5a3fcb4e0d4
+size 1060520
diff --git a/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..d23153160c1465668e26a8143c0f6a33cf181d8f
--- /dev/null
+++ b/gpt2-small_L6_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2de8c6a910ebd59b4a54ec0f1283733333e4d987
--- /dev/null
+++ b/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d86c20f86db7d384720c61ce5daf14648a11a98b3f8d553f317eb02c4d6e5a85
+size 1060520
diff --git a/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..fc295470fedc48c7efbe9d3fc8129b71e248d9e3
--- /dev/null
+++ b/gpt2-small_L6_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..01f6557e0cea9bee955e0ca07ddb8d0b13637971
--- /dev/null
+++ b/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4aa161e31077f754e8ed1d6a43bff404f1a39b850e80e021f69eba436230b71f
+size 1060520
diff --git a/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7d90a405ad13af06ab43723fce108437a508090
--- /dev/null
+++ b/gpt2-small_L6_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0d4410c474594929cc3769a285de853d3011a814
--- /dev/null
+++ b/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0d8b38462e8a23d6b5ed061134f438012619ff546348c7817953be88ad59bf9
+size 1060520
diff --git a/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..51c090255bb2c03473c24c2f812a6efd5b30755f
--- /dev/null
+++ b/gpt2-small_L6_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5c23aecde16f6b6109eac5d5108846ae22af9a25
--- /dev/null
+++ b/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f2360a4cc867577f88f584043f7258e574ab5514bb68c3f429a69b1274c1f95
+size 1060520
diff --git a/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d327188f0e9234bf7ed3caeedd17b4da317f184
--- /dev/null
+++ b/gpt2-small_L6_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4c098a5eacf00c389b2d38afbd8ca4a69911acfd
--- /dev/null
+++ b/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:012070923e74bcf7fdb6cc6f49a4b1c3a86306ea54b360df380f6c21b341d9ee
+size 1060520
diff --git a/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c65092743ad56d28179daeb605554a18bcb1cb61
--- /dev/null
+++ b/gpt2-small_L6_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 6, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.6.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L6H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..24e924ccd6c6ccfd9e094fe5f80a8368d1d57fbd
--- /dev/null
+++ b/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92059252f571b72f20a9eee1078aa66a7b0cc547a408ceb264823c129b555a9f
+size 1060520
diff --git a/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1efd879de1cb8fd0e22e96ca700c13484994a860
--- /dev/null
+++ b/gpt2-small_L7_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..42b3e6053337d58144c64cf4016477f4c5ef93b7
--- /dev/null
+++ b/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc48892cadc84d35b454ac81a2ebdf164854bb66a84dc699fa2528938c6a28e5
+size 1060528
diff --git a/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..74e4d5311a017707054e690c44bde37af040f419
--- /dev/null
+++ b/gpt2-small_L7_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0c0d87e7193132c24c392d755db60ca13143f7e7
--- /dev/null
+++ b/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cc577a99981a702f89cef86c5771ec4b6d0aedbbe101f06e39c956ba0c68524
+size 1060528
diff --git a/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..06defc152e9870502befaca83d183c96f942c801
--- /dev/null
+++ b/gpt2-small_L7_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7837573c6fd448839be811c32c7cd4b086c29c80
--- /dev/null
+++ b/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef914986a5679bb292cc1ff7a1906c2dcca8989f68954bb3f1cbc369a2928cca
+size 1060520
diff --git a/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..46e8e0c12d3985f91332d92394dbf4e7b9bf4f92
--- /dev/null
+++ b/gpt2-small_L7_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..04ab7aaa018855f3f36570147f54967fceb2e00c
--- /dev/null
+++ b/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a79d8c4f24b95501faac6fa28ebc5355f3ee2c1ee3102fe276ba4fe551f2f51
+size 1060520
diff --git a/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..a931bf6bcd2d51ce24c1d32f266222c894b94f03
--- /dev/null
+++ b/gpt2-small_L7_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3f51da3a2a1ed9252cd3136ab19d689425768bbd
--- /dev/null
+++ b/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:061b10814d65a11e4232fcc8c87a4ecbfd8f85575fd7a02b34dce2fa04160d48
+size 1060520
diff --git a/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8b857f6c4c0cc73f950762e92ca4b04ddd71d049
--- /dev/null
+++ b/gpt2-small_L7_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9dbefb772485e357c5fcaeb78d71066f909fb3f3
--- /dev/null
+++ b/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4deef9e9206d10e2860df6486b966ba4cbeb23099016cf7c96e9af84bd3f29db
+size 1060520
diff --git a/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..a12d8520f56c65f7402df0298b7d5b02e32f50a6
--- /dev/null
+++ b/gpt2-small_L7_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bfdfc246e0955d7fc5bff0a29df17660629a27f6
--- /dev/null
+++ b/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:546e3562ad7424485d071ccfa61d472173b7ccb6cfb4c91cac9e9dba8c4e10a4
+size 1060520
diff --git a/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c2aa3e7d500688efa204b54d00d29310afce67d4
--- /dev/null
+++ b/gpt2-small_L7_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0233b5a51aaeb7b564ade37a0c9602da844befd4
--- /dev/null
+++ b/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ee323f1a142ecaf7e791ae04d3efa6125c18bcf75b61ff090a94fb75af85036
+size 1060520
diff --git a/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..66a11f50e520dfc5ca80ca28ce0ca5390a632beb
--- /dev/null
+++ b/gpt2-small_L7_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2294452c2c28bc6de067ad91e8120d1de84f34ef
--- /dev/null
+++ b/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b08b33c9abe8bd2b20244526b0111eff10e99c131578db03721d5c493cd86fe7
+size 1060520
diff --git a/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8e71fd6b83255d6ba2972d648971677828aea4a5
--- /dev/null
+++ b/gpt2-small_L7_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..007894f0e19e46e461a21ae8d493b08e48011a0a
--- /dev/null
+++ b/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c3961387247778efb3b0d511890b8d211546c103c5a7970d0a75ca14d5425db
+size 1060520
diff --git a/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..12304d83860c0faf6e311f474abce49022a02289
--- /dev/null
+++ b/gpt2-small_L7_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0641b6a52e65c51e830a40ee8f04dbb52fbb2bad
--- /dev/null
+++ b/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:754f787aa0d8ecde3b16fc54ff34bf41b18eb88af0cf59510ffba638cc772182
+size 1060520
diff --git a/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..2c667627650991cfc2719f2481153494b3249467
--- /dev/null
+++ b/gpt2-small_L7_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 7, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.7.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L7H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3b133232f0711a411acb2da4feebf9ab12c0b10b
--- /dev/null
+++ b/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ab172617c517bcd82edde6f6bfbd07ecb73bf76abb3636603a9a825a281fa98
+size 1060520
diff --git a/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..d06010a056b498b76cd6ac3f0183465155f59f4a
--- /dev/null
+++ b/gpt2-small_L8_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..30bd9f246b0279b702be70f0d4b1f5f28e244fd6
--- /dev/null
+++ b/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba71063a0e543a871b179c6eca89d5ca21dfd26e6024d5f8a4d18759fc12c402
+size 1060528
diff --git a/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3c5b096de2949419d1d55bb7e382e035d884fde5
--- /dev/null
+++ b/gpt2-small_L8_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7dbd6032e9ccb41a411ff65c2baa60a596023cfa
--- /dev/null
+++ b/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f85951edaba0624fe65e3a19d1ac5deca865976aca3427aed6db855ad0f57f1a
+size 1060528
diff --git a/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..de155111981fce0b6484c757a6100b8df18e8de1
--- /dev/null
+++ b/gpt2-small_L8_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3b6ddae72244826b74777a4df43142b907537513
--- /dev/null
+++ b/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cbb3fb2da6750427a0d73592d662720684ce0e125095f06cdf200a3d3d18242
+size 1060520
diff --git a/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1f38516dbfe598de2f1ba9c88d60947c4b3beaf0
--- /dev/null
+++ b/gpt2-small_L8_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7d743c606da1e0f08b2519cf986969ef3a3a9eec
--- /dev/null
+++ b/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8eaaf53d2378ef9159b4bf6eb30a45e51fbb321fad072bb5f223fe17e62b8da2
+size 1060520
diff --git a/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..84b44ebe6f9cd34d3f29d78653a7f748ad4e751b
--- /dev/null
+++ b/gpt2-small_L8_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..919a61c57eac29a3fe44318dd362d9d6cee6469f
--- /dev/null
+++ b/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55a6fd142426c2087e91a8188d863963220e838f54237cc40f432f46dd29be6a
+size 1060520
diff --git a/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..9634111bccfeb2546ad6fac15868431158bcb047
--- /dev/null
+++ b/gpt2-small_L8_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f3d97c634cf3cc8f3954dde96b92fd1caa3bf253
--- /dev/null
+++ b/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca9852acead1ae2e12da7985ed1002e39ed7fad6e8666dcd62b8b6de8f513058
+size 1060520
diff --git a/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1339a0de29fb27a70f2c08f2c5c7a07bacdbfea5
--- /dev/null
+++ b/gpt2-small_L8_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f406e567486d02ad324f587de8fd6885b27c955f
--- /dev/null
+++ b/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e6e20524e3ee5e4766b3115432d7bec67661967e305b75eadb220451da6e66e
+size 1060520
diff --git a/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..d42bae054a68bbfb98520442de3ea8d10fdcde93
--- /dev/null
+++ b/gpt2-small_L8_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..620b305127cfb812501c9807fc8b8a422e693010
--- /dev/null
+++ b/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff0cd39f407b77ce86698c6e89d726426a56246c2c95fc1766b92130da666802
+size 1060520
diff --git a/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..71a161973a389e72b42c6dc376c2ca42f301edfd
--- /dev/null
+++ b/gpt2-small_L8_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..133c7f14e19b8f14a034d6b9fc6516bc800d1a83
--- /dev/null
+++ b/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:508e7fdaff6bb54cd4baa077294b838641bd2647fa6f9d5b9f8808247731d1cd
+size 1060520
diff --git a/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..86309ff396bcda03ab6ab5c5f9eb6f1692b2c16f
--- /dev/null
+++ b/gpt2-small_L8_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0aeced5417b19dc1c0075571cdd1916d97c818ee
--- /dev/null
+++ b/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c360bfb20c406a2cfb4c73ef55b390c599ebccc0f6ba883cdf816abe7f3db5da
+size 1060520
diff --git a/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..c71837936013a5316b08bee160b7f0ae9c2486d7
--- /dev/null
+++ b/gpt2-small_L8_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5498779b26d7fd1a51ac7119e331635287753060
--- /dev/null
+++ b/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9204f9003cbb10eafef82f6b11009ad29e1dc6ac045cdc54932fc916627aea24
+size 1060520
diff --git a/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..e1bf7459dc39badcad0e6d18ffdcc453dee5a136
--- /dev/null
+++ b/gpt2-small_L8_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 8, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.8.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L8H9_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..101d33411157935c8eb56a36631492d876185ab7
--- /dev/null
+++ b/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:645f6e849290d6a66e01db33bf76caeab14c233edaa4ae241ea4813b17adf93c
+size 1060520
diff --git a/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3954c8b357ec6973cc6d7a212326c7afc82c74e7
--- /dev/null
+++ b/gpt2-small_L9_H0_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 0, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H0_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c108d3ea5f1b573bcc3cb37420eb4a8b75d49bf8
--- /dev/null
+++ b/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:780c35601bbd6e1370c728789a57c11bd88bd14b394ef6b5da534ea2fd749c7c
+size 1060528
diff --git a/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..be8028ef4a7ff6ff0fc29ec381806d37bcb70cbe
--- /dev/null
+++ b/gpt2-small_L9_H10_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 10, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H10_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a8b30e45326041bca32d3b446a9bbe701cf28ff7
--- /dev/null
+++ b/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89e50dde5c810ab1cfc3546e0cc1b581ffa32335ccdfba988d6116619f5b7068
+size 1060528
diff --git a/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..43a8f4a4408d3d4dbafc78b13fc4b1bbbb0b132d
--- /dev/null
+++ b/gpt2-small_L9_H11_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 11, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H11_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f38fed7f3e534cc4e838bf010a418d6d793b7260
--- /dev/null
+++ b/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0497929433abeebcecbf8a6518b9ba1314772686a78e1016b01604bd8f9a5df3
+size 1060520
diff --git a/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..2bd783168ae30e55823cdfcd9a0c82f7d9bd2dab
--- /dev/null
+++ b/gpt2-small_L9_H1_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 1, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H1_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8e5be0c4ae176b33f1c197d41badf45843b3ed43
--- /dev/null
+++ b/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0214d6353290c70c90e5c1baa510265dd12e5ed647a326fda5cf3c36d04889a1
+size 1060520
diff --git a/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..8a3af52d068df5536d82a005d2a09406c64be1f7
--- /dev/null
+++ b/gpt2-small_L9_H2_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 2, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H2_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b3e8ae78948c071e26e55f9a01f422d74456b31f
--- /dev/null
+++ b/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b42c5ccee0c4a291f7eb5b3fde7d069009c8221bf5d053488b66735c6a6f9e4b
+size 1060520
diff --git a/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..7f59e7fafaefb0e7992403ee100137a545dd2dff
--- /dev/null
+++ b/gpt2-small_L9_H3_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 3, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H3_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..975aa171fc09f790aaadb78748f0dda3643f9ee0
--- /dev/null
+++ b/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f0c80e885e87fbb2bdf70c558ca21bf1fefdd44d8513367c67e7dbc7cd1000f
+size 1060520
diff --git a/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b7ebc760f87be10ace6507407ed92fbeb7165bbf
--- /dev/null
+++ b/gpt2-small_L9_H4_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 4, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H4_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ae996a236fcb2380fff0fdf7df6fdd60551825c4
--- /dev/null
+++ b/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2629e94ecbf89090a850b37ff115072cc6240f66c4eb303a0dc8ff5de47302c7
+size 1060520
diff --git a/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..114bec8d4eb0c1eb0011c1273513b968dc0016d5
--- /dev/null
+++ b/gpt2-small_L9_H5_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 5, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H5_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4f63985ff1293ff9e22eee2d235936b4dd030bbc
--- /dev/null
+++ b/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c5ea56e5a7979c056675549855e2a7f61aab57fbf1ae24fca6678cc18365e54
+size 1060520
diff --git a/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3b55e2126d892dd004ced8f7ac069d4dce1c329b
--- /dev/null
+++ b/gpt2-small_L9_H6_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 6, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H6_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e4b30b6f7a5a3d4053b19123e348df32c08d638f
--- /dev/null
+++ b/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99e01c39c693d23621922be468e84897ac00394bc61a9a745f1bcb3d869d9385
+size 1060520
diff --git a/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3f3e0e25cb89993ef4533d38db5fefea4cc1ce19
--- /dev/null
+++ b/gpt2-small_L9_H7_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 7, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H7_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7e6a9f20280f5e79a1a1b1b6edbeb228f5d398dc
--- /dev/null
+++ b/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03ebe69e8482d641b53cb4c1f9618040fca9f42b349ca3bffe77e910f0638ec4
+size 1060520
diff --git a/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..211b672c21a615384ac9d20a75d45f9b5df80722
--- /dev/null
+++ b/gpt2-small_L9_H8_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 8, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H8_2048_z"}
\ No newline at end of file
diff --git a/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt b/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0fe3cacd63773624f96915aca2fc165c60e74b76
--- /dev/null
+++ b/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61410b7dd93d39136bbf62bf0d1770e3a9c8751fdf1db8142b1dc0adc3fe6420
+size 1060520
diff --git a/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json b/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d2544611ed204c7ade0dae2112120d00750c785
--- /dev/null
+++ b/gpt2-small_L9_H9_z_lr1.20e-03_l11.00e+00_ds2048_nt2000000000_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_sl64_c1_v9_cfg.json
@@ -0,0 +1 @@
+{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0012, "num_tokens": 2000000000, "l1_coeff": 1.0, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 64, "contiguous": 1, "enc_dtype": "fp32", "model_name": "gpt2-small", "site": "z", "device": "cuda", "reinit": "reinit", "concat_heads": false, "per_head": true, "layer": 9, "head": 9, "resample_scheme": "anthropic", "anthropic_neuron_resample_scale": 0.2, "dead_direction_cutoff": 1e-06, "re_init_every": 25000, "anthropic_resample_last": 12500, "resample_factor": 0.01, "num_resamples": 4, "wandb_project_name": "attention-head-saes", "wandb_entity": "danf", "save_state_dict_every": 50000, "b_dec_init": "zeros", "sched_type": "cosine_warmup", "sched_epochs": 1000, "sched_lr_factor": 0.1, "sched_warmup_epochs": 1000, "sched_finish": true, "anthropic_resample_batches": 100, "eval_every": 1000, "test": false, "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 24576, "act_name": "blocks.9.attn.hook_z", "act_size": 64, "dict_size": 2048, "name": "gpt2-small_L9H9_2048_z"}
\ No newline at end of file