Upload folder using huggingface_hub

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_0/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e194665ab0ab01e0dbba0257d197c349196d38431ef466161b0e18c8c022860
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_0/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.012,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_0",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_0/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_1/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fc8b01942eb029e6d5ab1d4dd6b5717196e97e461226e7bf222b6648d2256fc
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_1/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.018,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_1",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_1/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_10/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5734865ba376c04ad8fa700a60978556b2743d4ebea7969a3b2037880bf70de5
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_10/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 520,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_10",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_10/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_11/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d116fc3fdf485924febb5f1f28a037488b8cc5e07763026d00880e4bb47f481d
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_11/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 820,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_11",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_11/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_2/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73b4c7661d318aaa92245db9036f3cd12e000a2f53ee7177e5d176def40f7c4c
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_2/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.024,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_2",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_2/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_3/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a65475230842ce2c280241d7a8cb2124ec4b7cdfdf20533810ae8c71553cc596
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_3/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.04,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_3",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_3/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_4/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7627d04102202588f7c59ae6fac0db0ab9e6eb1f14b1ae927ec28eb1c0cbb762
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_4/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.06,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_4",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_4/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_5/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2cb8f9cad0579a773fc0a48025a89fe47205445aad4da293da7b51f553735c3
+size 469975062

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_5/config.json

@@ -0,0 +1,28 @@
+{
+    "trainer": {
+        "dict_class": "GatedAutoEncoder",
+        "trainer_class": "GatedSAETrainer",
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "lr": 0.0003,
+        "l1_penalty": 0.08,
+        "warmup_steps": 1000,
+        "sparsity_warmup_steps": 5000,
+        "decay_start": 195312,
+        "seed": 3407,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "GatedTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_5",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_5/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_6/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:817f7e909f973725af379f1c57af2596b5595793726141bae8b48eee83a34c72
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_6/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 50,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_6",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_6/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_7/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ada562dfd95422f45268afa181b0db3627356234c443d0880d926120ed829886
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_7/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 80,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_7",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_7/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_8/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbc4ffeffa5ed600ad2ca933dd473b8375c48ad8dae2421fac74ce7154e8a3c2
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_8/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 160,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_8",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_8/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_9/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51722fe186f504360a1d44c5669f558c3a4b7b5c747c9c399ec5d27e5e6dcec0
+size 469843990

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_9/config.json

@@ -0,0 +1,31 @@
+{
+    "trainer": {
+        "trainer_class": "TopKTrainer",
+        "dict_class": "AutoEncoderTopK",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 320,
+        "device": "cuda",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "TopKTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_9",
+        "submodule_name": "resid_post_layer_10"
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_gated_top_k/resid_post_layer_10/trainer_9/eval_results.json

@@ -0,0 +1 @@
+{"info": "Diffusion LM detected. AR-only CE/loss-recovered metrics are skipped.", "hyperparameters": {"n_inputs": 200, "context_length": 2048}}