Upload trainer_0/config.json with huggingface_hub
Browse files- trainer_0/config.json +8 -5
trainer_0/config.json
CHANGED
|
@@ -7,15 +7,18 @@
|
|
| 7 |
"lr": 1e-06,
|
| 8 |
"l1_penalty": 0.01,
|
| 9 |
"warmup_steps": 0,
|
| 10 |
-
"sparsity_warmup_steps":
|
| 11 |
-
"steps":
|
| 12 |
"decay_start": null,
|
| 13 |
"seed": null,
|
| 14 |
"device": "cuda:2",
|
| 15 |
"layer": 7,
|
| 16 |
"lm_name": "blah",
|
| 17 |
"wandb_name": "StandardTrainerAprilUpdate",
|
| 18 |
-
"submodule_name": null
|
|
|
|
|
|
|
|
|
|
| 19 |
},
|
| 20 |
"buffer": {
|
| 21 |
"n_models": 2,
|
|
@@ -23,8 +26,8 @@
|
|
| 23 |
"io": "out",
|
| 24 |
"n_ctxs": 512,
|
| 25 |
"ctx_len": 256,
|
| 26 |
-
"refresh_batch_size":
|
| 27 |
-
"out_batch_size":
|
| 28 |
"device": "cuda:2",
|
| 29 |
"rescale_acts": false
|
| 30 |
}
|
|
|
|
| 7 |
"lr": 1e-06,
|
| 8 |
"l1_penalty": 0.01,
|
| 9 |
"warmup_steps": 0,
|
| 10 |
+
"sparsity_warmup_steps": 0,
|
| 11 |
+
"steps": 30517,
|
| 12 |
"decay_start": null,
|
| 13 |
"seed": null,
|
| 14 |
"device": "cuda:2",
|
| 15 |
"layer": 7,
|
| 16 |
"lm_name": "blah",
|
| 17 |
"wandb_name": "StandardTrainerAprilUpdate",
|
| 18 |
+
"submodule_name": null,
|
| 19 |
+
"frac_features_shared": 0.1,
|
| 20 |
+
"shared_l1_penalty": 0.002,
|
| 21 |
+
"num_shared_features": 5734
|
| 22 |
},
|
| 23 |
"buffer": {
|
| 24 |
"n_models": 2,
|
|
|
|
| 26 |
"io": "out",
|
| 27 |
"n_ctxs": 512,
|
| 28 |
"ctx_len": 256,
|
| 29 |
+
"refresh_batch_size": 1024,
|
| 30 |
+
"out_batch_size": 16384,
|
| 31 |
"device": "cuda:2",
|
| 32 |
"rescale_acts": false
|
| 33 |
}
|