Upload 2 files
Browse files- ae.pt +3 -0
- config.json +28 -0
ae.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86096c1a43eda4dfa9249b3779ab23a487c5733fdd90052976f1f160805121b4
|
| 3 |
+
size 67131048
|
config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"trainer": {
|
| 3 |
+
"dict_class": "AutoEncoder",
|
| 4 |
+
"trainer_class": "StandardTrainerAprilUpdate",
|
| 5 |
+
"activation_dim": 2048,
|
| 6 |
+
"dict_size": 8192,
|
| 7 |
+
"lr": 0.001,
|
| 8 |
+
"l1_penalty": 0.1,
|
| 9 |
+
"warmup_steps": 1000,
|
| 10 |
+
"sparsity_warmup_steps": 1000,
|
| 11 |
+
"steps": 10000,
|
| 12 |
+
"decay_start": null,
|
| 13 |
+
"seed": 42,
|
| 14 |
+
"device": "cuda",
|
| 15 |
+
"layer": 0,
|
| 16 |
+
"lm_name": "gelu-1l",
|
| 17 |
+
"wandb_name": "StandardSAE_gelu-1l_8192_trainer_0",
|
| 18 |
+
"submodule_name": null
|
| 19 |
+
},
|
| 20 |
+
"buffer": {
|
| 21 |
+
"d_submodule": 2048,
|
| 22 |
+
"n_ctxs": 3000,
|
| 23 |
+
"ctx_len": 128,
|
| 24 |
+
"refresh_batch_size": 32,
|
| 25 |
+
"out_batch_size": 1024,
|
| 26 |
+
"device": "cuda"
|
| 27 |
+
}
|
| 28 |
+
}
|