belerico commited on
Commit
81dbbed
·
verified ·
1 Parent(s): b13af68

Upload folder using huggingface_hub

Browse files
.DS_Store ADDED
Binary file (6.15 kB). View file
 
base-2/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true, "d_in": 2304}
base-2/sae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0af53457f6eac3b847221e58cd4a0c83f5be8b7083f62cf4ff1394984b46cebd
3
+ size 679781800
base-3/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true, "init_b_dec_from_median": false, "d_in": 2304}
base-3/sae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5d5871f8b208c3aad63dd595d9ae3fd1c456589a4db7868ac1dd646e23c5287
3
+ size 679781800
base-4/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true, "d_in": 2304}
base-4/sae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e193c8b5c8138163baaf37daabd27a9020692356831a0ab5c7f08ad81fc53db
3
+ size 679781800
base-5/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true, "d_in": 2304}
base-5/sae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43d98d310fba38f86cc6c9a7c50059b76afb946118ecc49e77cc93c335cb8259
3
+ size 679781800
config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"sae": {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true}, "batch_size": 4, "max_seq_len": 1024, "num_training_tokens": 1000000000, "cycle_iterator": true, "grad_acc_steps": 1, "micro_acc_steps": 1, "adam_8bit": false, "adam_epsilon": 1e-08, "adam_betas": [0.0, 0.999], "lr": 0.0007, "lr_init": 7e-05, "lr_end": 7e-05, "lr_scheduler_name": "constant", "lr_warmup_steps": 0.01, "lr_decay_steps": 0.2, "l1_coefficient": 0.0003, "l1_warmup_steps": 0.05, "use_l2_loss": true, "auxk_alpha": 0.0, "dead_feature_threshold": 10000000, "hookpoints": ["layers.0", "layers.1", "layers.2", "layers.3", "layers.4", "layers.5"], "layers": [0, 1, 2, 3, 4, 5], "layer_stride": 1, "distribute_modules": false, "save_every": 50000, "normalize_activations": 1, "num_norm_estimation_tokens": 2000000, "clusters": {"k6-c1": [2, 3, 4, 5], "base-2": [2], "base-4": [4], "base-5": [5]}, "cluster_hookpoints": {"k6-c1": ["layers.2", "layers.3", "layers.4", "layers.5"], "base-2": ["layers.2"], "base-4": ["layers.4"], "base-5": ["layers.5"]}, "hook": null, "log_to_wandb": true, "run_name": "checkpoints-clusters/google/gemma-2-2b-1024-lambda-0.0003-target-L0-None-lr-0.0007", "wandb_log_frequency": 1}
k6-c1/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"expansion_factor": 16, "normalize_decoder": true, "num_latents": 0, "k": -1, "multi_topk": false, "jumprelu": true, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "jumprelu_target_l0": null, "init_enc_as_dec_transpose": true, "d_in": 2304}
k6-c1/sae.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b03eb458a9e5cec7adeffb7046939d4b5d0d203f7ac8874993884a18cc24c0f7
3
+ size 679781800
l1_scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abcbfb40a2bca290ba6ca888084f7afefe943927dd2ed362b818074961d7a658
3
+ size 1012
lr_scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0be7660f336849e4ae2950529e4e881c18d25cf4c756b362af200ec0c8e70333
3
+ size 1460
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:341738d58f30699b9a3a604938a21dfbde5f4651ac7c925c98fd976e760a353e
3
+ size 744751104
scaling_factors.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2ddf8de1e3406c4165c9bc8f3b0e1ce79e3ed1c8deb358277292ca90e7e49b2
3
+ size 1024
state.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11014f6b3e1c57b4979d9cb8d3e76bb4eeb8b7a44fc7d22b0a76849d134b82b5
3
+ size 1181440