Add files using upload-large-folder tool

latents/layers.0.mlp/13107_16383.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f89d2d95b3a76092cb5986a851860156b052592f41b6fc3535e24332b5c6bd55
+size 142114144

latents/layers.0.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.1.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.10.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.11.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.12.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecddb550d3df068c52629e89ff7c16b74f6ea2b4ce592c88cae777b4e92a6404
+size 95995360

latents/layers.12.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.13.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.14.mlp/6553_9829.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:565056fde3ff5de8dff12eb9d0e486769e8fc81df45054b55b0ad7dd8df5f41c
+size 77526256

latents/layers.14.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.15.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:412ac4c5b74545a6019d31b557e66cef33f9b292cbb13bd34f6b1c8990c5abfa
+size 82303816

latents/layers.15.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.16.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.17.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.18.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2025f812cf713d573083724009d27f7ab22f5e1f6ffe0b7df10c0a3bd940f4c
+size 110648904

latents/layers.18.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.19.mlp/13107_16383.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07727cba5df920f4e241b776e365bc5500ed6731709f0625584a9f1e79a5a6c3
+size 97362792

latents/layers.19.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.2.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.20.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.21.mlp/0_3275.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9584a1711fbd97737639744df06acf81514f5d3283117367929f1d5f9a3b32eb
+size 70023648

latents/layers.21.mlp/13107_16383.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9beef344d54cf9dab6424dfa42a8ee5e6482bc20c140e5bdc492547557e75fd
+size 68286288

latents/layers.21.mlp/3276_6552.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa13fb4746c6e281d9fac54011cd2969a4dd5d5371a7e4cbf7ae06e2781807fb
+size 68418576

latents/layers.21.mlp/6553_9829.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6cc72522cb5295dff979f7164798174621778f5cc680eecf601da10861aadc2
+size 68586504

latents/layers.21.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33316bf9b94a56ace92800858a87af0169b50b365e09996a8e6cfcfe0c88ab37
+size 76055176

latents/layers.21.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.22.mlp/0_3275.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:449aeecb93927bcc20cd7da3343482eeab1b55688d3612348e3226882896e177
+size 76803312

latents/layers.22.mlp/3276_6552.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef638ec137246ceb78db1e7f101f7d89639bf45f10460a25502e482047144c7
+size 80527400

latents/layers.22.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.23.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.24.mlp/0_3275.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f8a4d18de008b583f8c719c4d4e60a1bd3fadbb8b10ba8e2e68120bf4b0d47b
+size 63471248

latents/layers.24.mlp/3276_6552.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:767bb0f63fb3846350761f7682f94faddc4bc546ad24122ac2e960c81bd923e2
+size 69611080

latents/layers.24.mlp/6553_9829.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc772c7a8d3a47fbda077d154a82f5ba83a8a43f219fee2c56677ba829a9ba04
+size 62302368

latents/layers.24.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:037125ae58de5c1e8a08dd2b0d9326d11e220c9e7265bd661e5f44cfb8e8e42e
+size 68635816

latents/layers.24.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.25.mlp/13107_16383.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:428c37ed6b657259068360cfda963652aeed6e84591cc8eb7f8e4d1b2cccbb57
+size 77973200

latents/layers.25.mlp/6553_9829.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9447886d66ac6821af328f37941b79f9f7ad4695fc18e3d894a38d9ca37a72b
+size 69532872

latents/layers.25.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5adbc1a3fd8aadfbc0ca39c33857b2a34cd8f7686615ad1c7ecb066f44ddc7cc
+size 85392664

latents/layers.25.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.3.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.4.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.5.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.6.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.7.mlp/6553_9829.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acdf5e15e0b8982c40ae3f500020af9a81598c22e5db8d02ac3e9ab0a9e379bd
+size 122482800

latents/layers.7.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.8.mlp/3276_6552.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37f8890cb031e3ecb8ab2d10578dcb66037e3fe9833469b549d4e07a98a92cd4
+size 94163624

latents/layers.8.mlp/9830_13106.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8c283edee97b8d17bdbbc67ac41ae2aa3612d1f716be7775671f17876aec27e
+size 99608256

latents/layers.8.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

latents/layers.9.mlp/config.json

@@ -0,0 +1,11 @@
+{
+    "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+    "dataset_split": "train",
+    "dataset_name": "",
+    "dataset_column": "text",
+    "batch_size": 16,
+    "cache_ctx_len": 256,
+    "n_tokens": 1000000,
+    "n_splits": 5,
+    "model_name": "google/gemma-2-2b"
+}

run_config.json

@@ -0,0 +1,81 @@
+{
+    "cache_cfg": {
+        "dataset_repo": "EleutherAI/fineweb-edu-dedup-10b",
+        "dataset_split": "train",
+        "dataset_name": "",
+        "dataset_column": "text",
+        "batch_size": 16,
+        "cache_ctx_len": 256,
+        "n_tokens": 1000000,
+        "n_splits": 5
+    },
+    "constructor_cfg": {
+        "faiss_embedding_model": "sentence-transformers/all-MiniLM-L6-v2",
+        "faiss_embedding_cache_dir": ".embedding_cache",
+        "faiss_embedding_cache_enabled": true,
+        "example_ctx_len": 32,
+        "min_examples": 200,
+        "n_non_activating": 50,
+        "center_examples": true,
+        "non_activating_source": "random",
+        "neighbours_type": "co-occurrence"
+    },
+    "sampler_cfg": {
+        "n_examples_train": 40,
+        "n_examples_test": 50,
+        "n_quantiles": 10,
+        "train_type": "quantiles",
+        "test_type": "quantiles",
+        "ratio_top": 0.2
+    },
+    "model": "google/gemma-2-2b",
+    "sparse_model": "EleutherAI/gemmascope-transcoders-sparsify",
+    "hookpoints": [
+        "layers.0.mlp",
+        "layers.1.mlp",
+        "layers.2.mlp",
+        "layers.3.mlp",
+        "layers.4.mlp",
+        "layers.5.mlp",
+        "layers.6.mlp",
+        "layers.7.mlp",
+        "layers.8.mlp",
+        "layers.9.mlp",
+        "layers.10.mlp",
+        "layers.11.mlp",
+        "layers.12.mlp",
+        "layers.13.mlp",
+        "layers.14.mlp",
+        "layers.15.mlp",
+        "layers.16.mlp",
+        "layers.17.mlp",
+        "layers.18.mlp",
+        "layers.19.mlp",
+        "layers.20.mlp",
+        "layers.21.mlp",
+        "layers.22.mlp",
+        "layers.23.mlp",
+        "layers.24.mlp",
+        "layers.25.mlp"
+    ],
+    "explainer_model": "hugging-quants/Meta-Llama-3.1-70B-Instruct-AWQ-INT4",
+    "explainer_model_max_len": 5120,
+    "explainer_provider": "offline",
+    "explainer": "default",
+    "scorers": [
+        "fuzz",
+        "detection"
+    ],
+    "name": "gemmascope-transcoders-sparsify-1m",
+    "max_latents": null,
+    "filter_bos": false,
+    "log_probs": false,
+    "load_in_8bit": false,
+    "hf_token": null,
+    "pipeline_num_proc": 48,
+    "num_gpus": 1,
+    "seed": 22,
+    "verbose": true,
+    "num_examples_per_scorer_prompt": 5,
+    "overwrite": []
+}