diff --git a/single_128x/config.json b/single_128x/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ec7dfcc080ba5906c69b9268fb78f6dc0b0dbcce --- /dev/null +++ b/single_128x/config.json @@ -0,0 +1 @@ +{"sae": {"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false}, "batch_size": 8, "grad_acc_steps": 4, "micro_acc_steps": 1, "optimizer": "signum", "lr": null, "lr_warmup_steps": 1000, "k_decay_steps": 0, "auxk_alpha": 0.0, "dead_feature_threshold": 10000000, "hookpoints": ["model.layers.17.mlp", "model.layers.18.mlp", "model.layers.19.mlp", "model.layers.20.mlp", "model.layers.21.mlp", "model.layers.22.mlp", "model.layers.24.mlp", "model.layers.29.mlp"], "init_seeds": [0], "layers": [], "layer_stride": 1, "distribute_modules": false, "save_every": 1000, "log_to_wandb": true, "run_name": "single_128x", "wandb_log_frequency": 1, "model": "HuggingFaceTB/SmolLM2-135M", "dataset": "EleutherAI/fineweb-edu-dedup-10b", "split": "train", "ctx_len": 2048, "hf_token": null, "revision": null, "load_in_8bit": false, "max_examples": 500000, "resume": false, "text_column": "text", "finetune": null, "shuffle_seed": 42, "data_preprocessing_num_proc": 48} \ No newline at end of file diff --git a/single_128x/model.layers.0.mlp/cfg.json b/single_128x/model.layers.0.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.0.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.0.mlp/sae.safetensors b/single_128x/model.layers.0.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ee2636eda18fa0c2b9d0d5357dbf67ccb1f816e5 --- /dev/null +++ b/single_128x/model.layers.0.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92299d41037bb97d4819c4bb29d917db2ae7a8420bae7803994fd42a23235795 +size 341363360 diff --git a/single_128x/model.layers.1.mlp/cfg.json b/single_128x/model.layers.1.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.1.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.1.mlp/sae.safetensors b/single_128x/model.layers.1.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8c6baa49bd8d3e3e170d721c61c93f8193c0a732 --- /dev/null +++ b/single_128x/model.layers.1.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c75b340c8f2e77ee66073cef10e67e12e80f1d156ed165b7b1b4ecd7e0522242 +size 341363360 diff --git a/single_128x/model.layers.10.mlp/cfg.json b/single_128x/model.layers.10.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.10.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.10.mlp/sae.safetensors b/single_128x/model.layers.10.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..aba2dac6eeaea64f9961d63d1fe8f4bcf263613a --- /dev/null +++ b/single_128x/model.layers.10.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4cb18c5d7da8cadcaeaa4e54ab89be53a6eacb6f73999df2b6726c17b2d440b +size 341363360 diff --git a/single_128x/model.layers.11.mlp/cfg.json b/single_128x/model.layers.11.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.11.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.11.mlp/sae.safetensors b/single_128x/model.layers.11.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c9a02dd0a39e969bd3660e88de712806d2394f9b --- /dev/null +++ b/single_128x/model.layers.11.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa865413536b0df777441d0c41a7d72d57ede77a3d24c838a9fcdb713064671e +size 341363360 diff --git a/single_128x/model.layers.12.mlp/cfg.json b/single_128x/model.layers.12.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.12.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.12.mlp/sae.safetensors b/single_128x/model.layers.12.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8a9bec731b2eda833a39d0eb1975011c510cb386 --- /dev/null +++ b/single_128x/model.layers.12.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:088dcf999f7c73f87b86fb48fb72aa74ac4937768fc0ef28f53a6b7c1c5d11b6 +size 341363360 diff --git a/single_128x/model.layers.13.mlp/cfg.json b/single_128x/model.layers.13.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.13.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.13.mlp/sae.safetensors b/single_128x/model.layers.13.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c4efc7755b9dee33ffc67c48c538697dd4049064 --- /dev/null +++ b/single_128x/model.layers.13.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76d9722130bc757ac0c047bc44f4d86cbe25f48e8387d62451fd4dd2423cf012 +size 341363360 diff --git a/single_128x/model.layers.14.mlp/cfg.json b/single_128x/model.layers.14.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.14.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.14.mlp/sae.safetensors b/single_128x/model.layers.14.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1f592fdb049791c50e081460ead364e68e36f51d --- /dev/null +++ b/single_128x/model.layers.14.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbe02cf1183e889d8d6f61e5f9c65d25283e9183e8df8dcdabbc8f4f0e3d9948 +size 341363360 diff --git a/single_128x/model.layers.15.mlp/cfg.json b/single_128x/model.layers.15.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.15.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.15.mlp/sae.safetensors b/single_128x/model.layers.15.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ec9224e493d62cfae3d15534867f9e538b27d106 --- /dev/null +++ b/single_128x/model.layers.15.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5a5790d95e1e2f846143f9bf395a479e542b6aaaf6c98b100dadf439e040551 +size 341363360 diff --git a/single_128x/model.layers.16.mlp/cfg.json b/single_128x/model.layers.16.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..0ad40bc4d1716b507aa3b6f9fb99b1b715c3b25e --- /dev/null +++ b/single_128x/model.layers.16.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.16.mlp/sae.safetensors b/single_128x/model.layers.16.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..063b41ffca9a971737996104e31f50e194248da5 --- /dev/null +++ b/single_128x/model.layers.16.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b4207e91baeacb6094a314b995b29886fcbf82fbf90dfbe5effc90fe5bf3cfb +size 341363360 diff --git a/single_128x/model.layers.17.mlp/cfg.json b/single_128x/model.layers.17.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.17.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.17.mlp/sae.safetensors b/single_128x/model.layers.17.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e19268343617652fb0884a7176ee02ce6388f2c8 --- /dev/null +++ b/single_128x/model.layers.17.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22389bfbea332b1d241ba12334a82e2b700d7df9cf466a83ec15f31be471fd5c +size 341363360 diff --git a/single_128x/model.layers.18.mlp/cfg.json b/single_128x/model.layers.18.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.18.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.18.mlp/sae.safetensors b/single_128x/model.layers.18.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6053a4a4796a834a2e46a5b92604df69aae7defc --- /dev/null +++ b/single_128x/model.layers.18.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c28b7dbb9a081701cee94bf8c584c25f9032e998b39e0eaa8611b018da114615 +size 341363360 diff --git a/single_128x/model.layers.19.mlp/cfg.json b/single_128x/model.layers.19.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.19.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.19.mlp/sae.safetensors b/single_128x/model.layers.19.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..95fe706e1403bef9eb84087b0ed9945ba04dc935 --- /dev/null +++ b/single_128x/model.layers.19.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63548cd5e1697979e13355c107127a5548c89033fc087ffd903bad7e9eb9b033 +size 341363360 diff --git a/single_128x/model.layers.2.mlp/cfg.json b/single_128x/model.layers.2.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.2.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.2.mlp/sae.safetensors b/single_128x/model.layers.2.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6ef1470252e3c28b8d1d16ff11c66ab288383e29 --- /dev/null +++ b/single_128x/model.layers.2.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afa2ed5574d04b4b2ecfb53bcdc25650198d1117f9cbeaea089d244a094055b1 +size 341363360 diff --git a/single_128x/model.layers.20.mlp/cfg.json b/single_128x/model.layers.20.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.20.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.20.mlp/sae.safetensors b/single_128x/model.layers.20.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..19e8e1eb94f4cb0c90488ddc4e396486374bffcf --- /dev/null +++ b/single_128x/model.layers.20.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e51ecad8c9c1145f85f5887643d102ba807c7bb65c63a00f34b0e0b7519a8476 +size 341363360 diff --git a/single_128x/model.layers.20.mlp_old/cfg.json b/single_128x/model.layers.20.mlp_old/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..38dcbb010e5653dc79cb70afd8da41db56e8c115 --- /dev/null +++ b/single_128x/model.layers.20.mlp_old/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.20.mlp_old/sae.safetensors b/single_128x/model.layers.20.mlp_old/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9cf5e80157e2284f825596fee9e9f55d81ca9757 --- /dev/null +++ b/single_128x/model.layers.20.mlp_old/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bbf6f1dc5377a17525fb999c4b10897c86ced205579c37121c8d6dc735b06ac +size 341363360 diff --git a/single_128x/model.layers.21.mlp/cfg.json b/single_128x/model.layers.21.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.21.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.21.mlp/sae.safetensors b/single_128x/model.layers.21.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1cd6bf2fc8f811d19d2ffad024ee7dca3cc22cab --- /dev/null +++ b/single_128x/model.layers.21.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43fe17a75aeb346df081a2bd5d49ab16178623c04f02a99855d0b4aa84b7dd08 +size 341363360 diff --git a/single_128x/model.layers.21.mlp_old/cfg.json b/single_128x/model.layers.21.mlp_old/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..38dcbb010e5653dc79cb70afd8da41db56e8c115 --- /dev/null +++ b/single_128x/model.layers.21.mlp_old/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.21.mlp_old/sae.safetensors b/single_128x/model.layers.21.mlp_old/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..baad6d2560ae2ce86bb1768a4d3fa5caee4eb0fe --- /dev/null +++ b/single_128x/model.layers.21.mlp_old/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a89aa886e34339841f1d18aca73a4d07800d3628680f894ad5c322d924be86b0 +size 341363360 diff --git a/single_128x/model.layers.22.mlp/cfg.json b/single_128x/model.layers.22.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.22.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.22.mlp/sae.safetensors b/single_128x/model.layers.22.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fe13b86789f42a9887c25315852ab821bd5d28ca --- /dev/null +++ b/single_128x/model.layers.22.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f73dd558709203804e1b2526fc5cb67ab83ca0bbffaf29ad8322daba8e70def0 +size 341363360 diff --git a/single_128x/model.layers.22.mlp_old/cfg.json b/single_128x/model.layers.22.mlp_old/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..38dcbb010e5653dc79cb70afd8da41db56e8c115 --- /dev/null +++ b/single_128x/model.layers.22.mlp_old/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.22.mlp_old/sae.safetensors b/single_128x/model.layers.22.mlp_old/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c68aff46f9b2a2335c4bffba11f072e256d8754c --- /dev/null +++ b/single_128x/model.layers.22.mlp_old/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d176fb7a5a2688cad50a9c22cfcebb8cd1e0972639dc79f9e77a2fa49aed2750 +size 341363360 diff --git a/single_128x/model.layers.23.mlp/cfg.json b/single_128x/model.layers.23.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.23.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.23.mlp/sae.safetensors b/single_128x/model.layers.23.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e1ecc95ce3d653ca491fbf439346eb3bf990a5ed --- /dev/null +++ b/single_128x/model.layers.23.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1218d33e4b8bba5f4b067206a24693164208f6191f3885d1a86fd8ecbc0a0623 +size 341363360 diff --git a/single_128x/model.layers.24.mlp/cfg.json b/single_128x/model.layers.24.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.24.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.24.mlp/sae.safetensors b/single_128x/model.layers.24.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e6daf05beae27ca89bb202189b99b9ab7667dc21 --- /dev/null +++ b/single_128x/model.layers.24.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9703d9323baafaeab74d7357e6e2f3385b94c742bd873bcbeae009a8a7bb736d +size 341363360 diff --git a/single_128x/model.layers.25.mlp/cfg.json b/single_128x/model.layers.25.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.25.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.25.mlp/sae.safetensors b/single_128x/model.layers.25.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..067cb67465e6be0019811287673ad3d4842e96ab --- /dev/null +++ b/single_128x/model.layers.25.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e56d01b7c5cd5e6ca031d5b3c43ecf34c3d184f8b9a7cdeb3d2fb86d7a2f745e +size 341363360 diff --git a/single_128x/model.layers.26.mlp/cfg.json b/single_128x/model.layers.26.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..acaab9e38dd921b9acd7d395481ed3ff5eef6dec --- /dev/null +++ b/single_128x/model.layers.26.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true,"d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.26.mlp/sae.safetensors b/single_128x/model.layers.26.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2652fe4349c09a1117a3ef3a222d385af672392e --- /dev/null +++ b/single_128x/model.layers.26.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:430a72cf5fa68b97b887b93b289ee007353db785d04971fe64059edc02b703ab +size 341363360 diff --git a/single_128x/model.layers.27.mlp/cfg.json b/single_128x/model.layers.27.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..0ad40bc4d1716b507aa3b6f9fb99b1b715c3b25e --- /dev/null +++ b/single_128x/model.layers.27.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.27.mlp/sae.safetensors b/single_128x/model.layers.27.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..655aea11237f47d6a60356a0f8242273744ade41 --- /dev/null +++ b/single_128x/model.layers.27.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8317d1b216779c8667832b55664fabfe08fab6c3444856ce096c980d1b3bddb1 +size 341363360 diff --git a/single_128x/model.layers.28.mlp/cfg.json b/single_128x/model.layers.28.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..38dcbb010e5653dc79cb70afd8da41db56e8c115 --- /dev/null +++ b/single_128x/model.layers.28.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.28.mlp/sae.safetensors b/single_128x/model.layers.28.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a4f2d00ffbe8dc7ecc26317e9d9be73c41d7870f --- /dev/null +++ b/single_128x/model.layers.28.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb59358c254fbebfba816bdb06a6fc508b83c189d4a3d3eb7569b7d3f9f19cea +size 341363360 diff --git a/single_128x/model.layers.29.mlp/cfg.json b/single_128x/model.layers.29.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.29.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.29.mlp/sae.safetensors b/single_128x/model.layers.29.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..368a720eb1fdd4d5636634bba62c8c295a2af318 --- /dev/null +++ b/single_128x/model.layers.29.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff05370f49260565718486209371f0646e7c30fdc1f130d07af87194fbc9842a +size 341363360 diff --git a/single_128x/model.layers.3.mlp/cfg.json b/single_128x/model.layers.3.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.3.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.3.mlp/sae.safetensors b/single_128x/model.layers.3.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6bbe018d35ffdf9f9002c487ae0a4d8136871b3f --- /dev/null +++ b/single_128x/model.layers.3.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ead9b2a42a4cea47123f0f7638e7b4790cc35aff2ad4bfdc6547ea351af87c9 +size 341363360 diff --git a/single_128x/model.layers.4.mlp/cfg.json b/single_128x/model.layers.4.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.4.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.4.mlp/sae.safetensors b/single_128x/model.layers.4.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..091bf0d17dc58e8ee30f4a9c3e4f38eedf624e0e --- /dev/null +++ b/single_128x/model.layers.4.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6a4647edce663de11eb1075eb37914c96e76cd0c3890438d09fe3a97c68bea0 +size 341363360 diff --git a/single_128x/model.layers.5.mlp/cfg.json b/single_128x/model.layers.5.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.5.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.5.mlp/sae.safetensors b/single_128x/model.layers.5.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6e637759962b20d553efbdc8741eeef7b73ae714 --- /dev/null +++ b/single_128x/model.layers.5.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e4f2f151d6f721f66ef78f0e970f9066cdaafab9f40273c19c428330ca0683f +size 341363360 diff --git a/single_128x/model.layers.6.mlp/cfg.json b/single_128x/model.layers.6.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.6.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.6.mlp/sae.safetensors b/single_128x/model.layers.6.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..20bb8a746b3acd1e0b0d4fea729fc2082fac4ab9 --- /dev/null +++ b/single_128x/model.layers.6.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5ccb76dd69a5f41c71c3bd0b173c307e02cad6674fd367f1b9dc125c7332b81 +size 341363360 diff --git a/single_128x/model.layers.7.mlp/cfg.json b/single_128x/model.layers.7.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.7.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.7.mlp/sae.safetensors b/single_128x/model.layers.7.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ee70cb4e9d48a3286b3a3d4594b4216544c8d5ec --- /dev/null +++ b/single_128x/model.layers.7.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87a3f9e1d7c7e88cb44d6224404e7b3ea9d7cc4424859036d7f36eaf2732823d +size 341363360 diff --git a/single_128x/model.layers.8.mlp/cfg.json b/single_128x/model.layers.8.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..0ad40bc4d1716b507aa3b6f9fb99b1b715c3b25e --- /dev/null +++ b/single_128x/model.layers.8.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.8.mlp/sae.safetensors b/single_128x/model.layers.8.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bdffa51f8485cf02c3051f7b61ca28d948fc4b29 --- /dev/null +++ b/single_128x/model.layers.8.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07f811a32b7c0deb5cb15ad3f1e8529bb38265768eafba0713b3c208b4bc02e5 +size 341363360 diff --git a/single_128x/model.layers.9.mlp/cfg.json b/single_128x/model.layers.9.mlp/cfg.json new file mode 100644 index 0000000000000000000000000000000000000000..5e352b0700fe744562769a38fbb325acb93e81dd --- /dev/null +++ b/single_128x/model.layers.9.mlp/cfg.json @@ -0,0 +1 @@ +{"activation": "topk", "expansion_factor": 128, "normalize_decoder": true, "num_latents": 0, "k": 128, "multi_topk": false, "skip_connection": true, "transcode": true, "multi_layer": false, "lens": false, "d_in": 576} \ No newline at end of file diff --git a/single_128x/model.layers.9.mlp/sae.safetensors b/single_128x/model.layers.9.mlp/sae.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d8a4810852c4e5d1de6131125709e9234c74fb3e --- /dev/null +++ b/single_128x/model.layers.9.mlp/sae.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc031e7cac0942cf22c409c7cd01eac2d603e65f08981941b23074b183d833be +size 341363360 diff --git a/single_128x/optimizer_0.pt b/single_128x/optimizer_0.pt new file mode 100644 index 0000000000000000000000000000000000000000..826cf9065f190a239d98623b285c9baed5b42885 --- /dev/null +++ b/single_128x/optimizer_0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7834718242c4c77e82befd77135a48ac20cfdf7d8624453b7f281b9ad9017fc6 +size 2730916366 diff --git a/single_128x/rank_0_state.pt b/single_128x/rank_0_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..f3a6b77a29b906fecf8bc59ca4adef2b9b9fd15e --- /dev/null +++ b/single_128x/rank_0_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4eec7099ee48f214c44251cddd6819c67f4a14ed77b0f948bc7703b738b050d8 +size 4721852 diff --git a/single_128x/state.pt b/single_128x/state.pt new file mode 100644 index 0000000000000000000000000000000000000000..4eff9ae8a896157c1a0553106682e0d7d6128d55 --- /dev/null +++ b/single_128x/state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44d738715009486e24de8c546db27158058643accac7eb9d7303ad52a9515bf5 +size 856