Upload Case 7
#11
by
iarcuschin
- opened
- 7/edges.pkl +3 -0
- 7/ll_model.pth +3 -0
- 7/ll_model_cfg.pkl +3 -0
- 7/meta.json +1 -0
7/edges.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3156b2e517b8c45e76ea28f8c2ccbcda226f636cd44a809c908e167871d2cdab
|
| 3 |
+
size 546
|
7/ll_model.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ee0cfc16d28ec4133f303fa1bdac3be6c19be72c10a23205956151582cc075a
|
| 3 |
+
size 42474
|
7/ll_model_cfg.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:630fa961d2d30927dfe575b0625d4bfaefa14bc0b751a0ee7224206705b939dc
|
| 3 |
+
size 1048
|
7/meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"atol": 0.05, "lr": 0.01, "use_single_loss": false, "iit_weight": 1.0, "behavior_weight": 1.0, "strict_weight": 0.5, "epochs": 2000, "act_fn": "gelu", "wandb_suffix": "", "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "", "model_pair": "strict", "same_size": false, "seed": 1234, "batch_size": 256, "include_mlp": false, "next_token": false, "detach_while_caching": true}
|