exp010: ctx_512w
Browse files
experiments/exp_010_component_potentials/runs/ctx_512w/checkpoint.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:56126435030e58b8c7e830ac356d1721ed02335953a24647b75f2064f8b43518
|
| 3 |
+
size 27697006
|
experiments/exp_010_component_potentials/runs/ctx_512w/result.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run": "ctx_512w",
|
| 3 |
+
"family": "ctx",
|
| 4 |
+
"dev_final": 0.011899590492248535,
|
| 5 |
+
"rank_final": 3.965059518814087,
|
| 6 |
+
"statute_final": "uniform",
|
| 7 |
+
"params": 6851394,
|
| 8 |
+
"minutes": 8.3,
|
| 9 |
+
"bpb": 3.0744691839977056,
|
| 10 |
+
"bpb_eval": NaN,
|
| 11 |
+
"coverage": 0.944091796875,
|
| 12 |
+
"gate_acc": 0.9466552734375,
|
| 13 |
+
"mode_spread_deg": NaN,
|
| 14 |
+
"mix_entropy": NaN,
|
| 15 |
+
"write_kl": 0.14493660628795624,
|
| 16 |
+
"kernel_aux": NaN,
|
| 17 |
+
"loss": 6.40767240524292,
|
| 18 |
+
"condition": {
|
| 19 |
+
"tag": "ctx_512w",
|
| 20 |
+
"dim": "384",
|
| 21 |
+
"n_layers": "4",
|
| 22 |
+
"n_heads": "6",
|
| 23 |
+
"K": "64",
|
| 24 |
+
"seq_len": "512",
|
| 25 |
+
"batch_size": "16",
|
| 26 |
+
"accum_steps": "1",
|
| 27 |
+
"d_point": "48",
|
| 28 |
+
"head": "hybrid",
|
| 29 |
+
"steps": "5000",
|
| 30 |
+
"log_every": "500"
|
| 31 |
+
}
|
| 32 |
+
}
|
experiments/exp_010_component_potentials/runs/ctx_512w/snapshots.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62b3aebfb3c4dff893ab0f80a8255465e05765429ff2cd8469d14f7fc5c29471
|
| 3 |
+
size 17188
|