Upload folder using huggingface_hub

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e8b2c805dcb2624db548b3987fa59042e9029194505769965d44077f8ab265c
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 50,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aefc5e90ba2af846b106376389e0c802ba24cdc75273678b940f240fb2e3c2b7
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 80,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:841af7950d72f221689709124de504b853640f2bd337740731097421a7453088
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_10",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 520
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eaba84a008af5f3135234220470846628d0bf8463b46d58b64d31c8eb434d03
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_11",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 820
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:907185638dfee25ddc34a4b7a8fda39757959dff7ffa1ce9b84211f583bb858c
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 160,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63c6138830e253cf3e868300bdef8cc499fd200cb749f283199e231f38682f7f
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 320,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c1c852686087048fe41969b1fb228971140af1d6155bfe7bfb6079a6e176f52
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 520,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f30bd7aacac1c0d1f70b57a19668e6ec55641d1b360c44dc534247914f0eda46
+size 469843990

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

@@ -0,0 +1,32 @@
+{
+    "trainer": {
+        "trainer_class": "BatchTopKTrainer",
+        "dict_class": "BatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 244140,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 195312,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1792,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "k": 820,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "BatchTopKTrainer-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:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a2f7199ad9f7f5f218b5005ba25affd09ba17b960b3388f9ccde1af262ffc17
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_6",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 50
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05ba58d9c5a518b0848e524392722ec4ce303b7e21255a6c19456921cfdb9631
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_7",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 80
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2612c829726f8d3e8368eaeab22b52b1f78b81bbac78337a11e800a8363a9a84
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_8",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 160
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc6d33cc28ef78470e10c322b2770944c21e39f1a8d3377b3c11c5287cbc1672
+size 469909279

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

@@ -0,0 +1,29 @@
+{
+    "trainer": {
+        "trainer_class": "JumpReluTrainer",
+        "dict_class": "JumpReluAutoEncoder",
+        "lr": 0.0003,
+        "steps": 244140,
+        "seed": 3407,
+        "activation_dim": 3584,
+        "dict_size": 16384,
+        "device": "cuda:0",
+        "layer": 10,
+        "lm_name": "Dream-org/Dream-v0-Base-7B",
+        "wandb_name": "JumpReluTrainer-Dream-org/Dream-v0-Base-7B-resid_post_layer_10_trainer_9",
+        "submodule_name": "resid_post_layer_10",
+        "bandwidth": 0.001,
+        "sparsity_penalty": 1.0,
+        "sparsity_warmup_steps": 5000,
+        "target_l0": 320
+    },
+    "buffer": {
+        "d_submodule": 3584,
+        "io": "out",
+        "n_ctxs": 122,
+        "ctx_len": 2048,
+        "refresh_batch_size": 8,
+        "out_batch_size": 2048,
+        "device": "cuda:0"
+    }
+}

saes_Dream-org_Dream-v0-Base-7B_batch_top_k_jump_relu/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}}