Upload trainer_0/config.json with huggingface_hub
Browse files- trainer_0/config.json +5 -5
trainer_0/config.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
"trainer": {
|
| 3 |
"trainer_class": "TopKTrainer",
|
| 4 |
"dict_class": "AutoEncoderTopK",
|
| 5 |
-
"lr": 0.
|
| 6 |
"steps": 12,
|
| 7 |
"auxk_alpha": 0.03125,
|
| 8 |
"warmup_steps": 0,
|
|
@@ -10,8 +10,8 @@
|
|
| 10 |
"threshold_beta": 0.999,
|
| 11 |
"threshold_start_step": 1000,
|
| 12 |
"seed": null,
|
| 13 |
-
"activation_dim":
|
| 14 |
-
"dict_size":
|
| 15 |
"k": 128,
|
| 16 |
"device": "cuda:2",
|
| 17 |
"layer": 4,
|
|
@@ -20,8 +20,8 @@
|
|
| 20 |
"submodule_name": null
|
| 21 |
},
|
| 22 |
"buffer": {
|
| 23 |
-
"n_models":
|
| 24 |
-
"d_submodule":
|
| 25 |
"io": "out",
|
| 26 |
"n_ctxs": 1024,
|
| 27 |
"ctx_len": 128,
|
|
|
|
| 2 |
"trainer": {
|
| 3 |
"trainer_class": "TopKTrainer",
|
| 4 |
"dict_class": "AutoEncoderTopK",
|
| 5 |
+
"lr": 0.00010690449676496975,
|
| 6 |
"steps": 12,
|
| 7 |
"auxk_alpha": 0.03125,
|
| 8 |
"warmup_steps": 0,
|
|
|
|
| 10 |
"threshold_beta": 0.999,
|
| 11 |
"threshold_start_step": 1000,
|
| 12 |
"seed": null,
|
| 13 |
+
"activation_dim": 896,
|
| 14 |
+
"dict_size": 57344,
|
| 15 |
"k": 128,
|
| 16 |
"device": "cuda:2",
|
| 17 |
"layer": 4,
|
|
|
|
| 20 |
"submodule_name": null
|
| 21 |
},
|
| 22 |
"buffer": {
|
| 23 |
+
"n_models": 1,
|
| 24 |
+
"d_submodule": 896,
|
| 25 |
"io": "out",
|
| 26 |
"n_ctxs": 1024,
|
| 27 |
"ctx_len": 128,
|