Upload folder using huggingface_hub

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/ae.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89c6d2be19b13385a36f5a8e2d6a7471614471c640d912a18be137868429b903
+size 75525712

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24404e9d6d1285146667c5289c3d2b32823b23aa3926aa9219756018d3556ef0
+size 75525930

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_11718.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b78ed08c1a38fa629c2cf803569eabaaed89dd615b0fcc8a7fac92df2f637d6c
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_15624.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c4fbf2857b1e5bfaaf9c57dc65829e2157c44eb479c56a128917250efe61745
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_1953.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eef3da5fd0966a2f4961bd67d30009a594878929bfb43b1466bf73e6e07bc42d
+size 75525969

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_19531.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de8a84487bdbeb048734671ff4cd0e16f305ecb91cf3d59f446c9f2421850dea
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_23437.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:587d5a41bc364c2b639f8cdd5ccccd337cfd47a907f441321d50a02d5e2d858d
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_27343.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f7070263a351d8435ba175629eb5d6fe697ddc6672bd2bb93acc885eec8b76b
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_31249.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd9bcaf4a624abeb02e56aeb15674bbde2b1c748d48b8581ad02ce971ae8d4d8
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_35155.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc4bd2cab4fd3680dc3de0c4b8ba2aedcd4c87254a2eb179a7bda18ec877aa4a
+size 75526238

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_390.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2117b5fab3a7389998d91dd2f43fe775d5f5c7d1f6add633e50b9dff41e3d26b
+size 75525956

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_3906.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae5b05b7886c9997e592f71d41cbba3389305af124fa2ed083b870debf4d4011
+size 75525969

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/checkpoints/ae_7812.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75b621a30963432c556bf13b5944f77c644b42971cefee0f2911988d1f0db9dd
+size 75525969

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/config.json

@@ -0,0 +1,59 @@
+{
+    "trainer": {
+        "trainer_class": "MatryoshkaBatchTopKTrainer",
+        "dict_class": "MatryoshkaBatchTopKSAE",
+        "lr": 0.0003,
+        "steps": 39062,
+        "auxk_alpha": 0.03125,
+        "warmup_steps": 1000,
+        "decay_start": 31249,
+        "threshold_beta": 0.999,
+        "threshold_start_step": 1000,
+        "top_k_aux": 1152,
+        "seed": 2,
+        "activation_dim": 2304,
+        "dict_size": 4096,
+        "group_fractions": [
+            0.03125,
+            0.0625,
+            0.125,
+            0.25,
+            0.53125
+        ],
+        "group_weights": [
+            0.2,
+            0.2,
+            0.2,
+            0.2,
+            0.2
+        ],
+        "group_sizes": [
+            128,
+            256,
+            512,
+            1024,
+            2176
+        ],
+        "ortho_alpha": 0.1,
+        "ortho_warmup_start_step": 0,
+        "ortho_warmup_steps": 0,
+        "k": 80,
+        "device": "cuda:0",
+        "layer": 12,
+        "lm_name": "google/gemma-2-2b",
+        "wandb_name": "M-4096-80-2_trainer_0",
+        "submodule_name": "resid_post_layer_12",
+        "dynamic_loss_boundaries": false,
+        "num_dynamic_loss_samples": 5,
+        "dynamic_loss_pareto_alpha": 0.5
+    },
+    "buffer": {
+        "d_submodule": 2304,
+        "io": "out",
+        "n_ctxs": 132,
+        "ctx_len": 1024,
+        "refresh_batch_size": 4,
+        "out_batch_size": 128,
+        "device": "cuda:0"
+    }
+}

._fast_run_4096w_80k_128bz_ft5m_s2_l40s_50m_newortho0.1_google_gemma-2-2b_matryoshka_batch_top_k/resid_post_layer_12/trainer_0/eval_results.json

@@ -0,0 +1 @@
+{"l2_loss": 74.495, "l1_loss": 851.8, "l0": 80.26638671875, "frac_variance_explained": 0.72560546875, "cossim": 0.85169921875, "l2_ratio": 0.85904296875, "relative_reconstruction_bias": 1.008125, "loss_original": 2.152158203125, "loss_reconstructed": 2.57166015625, "loss_zero": 12.4375, "frac_recovered": 0.95931640625, "frac_alive": 0.226318359375, "hyperparameters": {"n_inputs": 200, "context_length": 1024}}