Update experiment measurements and status
Browse files
reports/runs/control-2027/resolved.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"vocab": 12288,
|
| 4 |
+
"n_layer": 19,
|
| 5 |
+
"n_embd": 768,
|
| 6 |
+
"n_head": 12,
|
| 7 |
+
"block": 1024,
|
| 8 |
+
"ffn_mult": 2.8125,
|
| 9 |
+
"norm": "rmsnorm",
|
| 10 |
+
"norm_eps": 1e-06,
|
| 11 |
+
"pos": "rope",
|
| 12 |
+
"rope_theta": 100000.0,
|
| 13 |
+
"ffn": "swiglu",
|
| 14 |
+
"value_residual": true,
|
| 15 |
+
"qk_norm": true,
|
| 16 |
+
"logit_cap": 0.0,
|
| 17 |
+
"z_loss": 0.0
|
| 18 |
+
},
|
| 19 |
+
"train": {
|
| 20 |
+
"seed": 2027,
|
| 21 |
+
"micro_batch": 64,
|
| 22 |
+
"global_batch": 512,
|
| 23 |
+
"steps": 1000,
|
| 24 |
+
"warmup": 50,
|
| 25 |
+
"lr": 0.0001,
|
| 26 |
+
"min_lr": 1e-05,
|
| 27 |
+
"muon_lr": 0.0033333333333333335,
|
| 28 |
+
"checkpoint_every": 500,
|
| 29 |
+
"validate_every": 500,
|
| 30 |
+
"log_every": 20,
|
| 31 |
+
"compile": true
|
| 32 |
+
},
|
| 33 |
+
"data": {
|
| 34 |
+
"train": "/data/ablation/mixtures/control/train.bin",
|
| 35 |
+
"validation": "/data/final/val.bin",
|
| 36 |
+
"sha256": "b7f7f59462cac9092bd301e742008413ddfbbd81cb1b2e901bea0d9f9e275ecf"
|
| 37 |
+
}
|
| 38 |
+
}
|