last 2026-10-26 04:43:54
Browse files- last/ckpt.pt +1 -1
- last/state.json +1 -1
- logs/train_log.jsonl +27 -0
last/ckpt.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 329499755
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c21c86be926b1832fbd6f8344f872ec8109c07ae1a618a2530ad240a1700f5b7
|
| 3 |
size 329499755
|
last/state.json
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
{"step":
|
|
|
|
| 1 |
+
{"step": 2295, "best_val": 0.91477290856758, "tokens": 29339524, "time": 1792989834.5044096}
|
logs/train_log.jsonl
CHANGED
|
@@ -25,3 +25,30 @@
|
|
| 25 |
{"step": 1050, "loss": 0.8739219456911087, "lr": 0.0001786775661379826, "tok_s": 10377.375157935554}
|
| 26 |
{"step": 1100, "loss": 0.9231078624725342, "lr": 0.00017621515196058186, "tok_s": 12446.257261721274}
|
| 27 |
{"step": 1150, "loss": 0.9487660825252533, "lr": 0.00017363961030678927, "tok_s": 12739.831781641597}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
{"step": 1050, "loss": 0.8739219456911087, "lr": 0.0001786775661379826, "tok_s": 10377.375157935554}
|
| 26 |
{"step": 1100, "loss": 0.9231078624725342, "lr": 0.00017621515196058186, "tok_s": 12446.257261721274}
|
| 27 |
{"step": 1150, "loss": 0.9487660825252533, "lr": 0.00017363961030678927, "tok_s": 12739.831781641597}
|
| 28 |
+
{"step": 1200, "loss": 0.9254888594150543, "lr": 0.00017095534144631673, "tok_s": 11473.308966912637}
|
| 29 |
+
{"step": 1250, "loss": 0.9310972988605499, "lr": 0.00016816693140748467, "tok_s": 12577.751999878736}
|
| 30 |
+
{"step": 1250, "val": 0.9210733401546363}
|
| 31 |
+
{"step": 1300, "loss": 0.8653187304735184, "lr": 0.00016527914414207013, "tok_s": 11284.037147817386}
|
| 32 |
+
{"step": 1350, "loss": 0.875058576464653, "lr": 0.00016229691338617445, "tok_s": 13086.806931577596}
|
| 33 |
+
{"step": 1400, "loss": 0.9580196440219879, "lr": 0.00015922533423101844, "tok_s": 12859.674411952936}
|
| 34 |
+
{"step": 1450, "loss": 0.8628480732440948, "lr": 0.0001560696544180641, "tok_s": 12601.750255798916}
|
| 35 |
+
{"step": 1500, "loss": 0.9511719644069672, "lr": 0.00015283526537333665, "tok_s": 12430.632830505027}
|
| 36 |
+
{"step": 1500, "val": 0.9186573113952553}
|
| 37 |
+
{"step": 1550, "loss": 0.934482753276825, "lr": 0.00014952769299626334, "tok_s": 10073.13633354152}
|
| 38 |
+
{"step": 1600, "loss": 0.958391547203064, "lr": 0.00014615258821876727, "tok_s": 12730.449882366673}
|
| 39 |
+
{"step": 1650, "loss": 0.9036903530359268, "lr": 0.0001427157173507447, "tok_s": 12581.582716001394}
|
| 40 |
+
{"step": 1700, "loss": 0.8833057880401611, "lr": 0.00013922295222842154, "tok_s": 12402.647912565104}
|
| 41 |
+
{"step": 1750, "loss": 0.8902811706066132, "lr": 0.0001356802601824195, "tok_s": 12507.538377892315}
|
| 42 |
+
{"step": 1750, "val": 0.9175305264582957}
|
| 43 |
+
{"step": 1800, "loss": 0.7865059971809387, "lr": 0.00013209369384267194, "tok_s": 10555.245308831361}
|
| 44 |
+
{"step": 1850, "loss": 0.8891026079654694, "lr": 0.0001284693807976071, "tok_s": 12961.75313356928}
|
| 45 |
+
{"step": 1900, "loss": 0.872664138674736, "lr": 0.00012481351312526606, "tok_s": 12441.517379559607}
|
| 46 |
+
{"step": 1950, "loss": 0.8614092469215393, "lr": 0.00012113233681424131, "tok_s": 12457.177008945715}
|
| 47 |
+
{"step": 2000, "loss": 0.951303243637085, "lr": 0.00011743214109250993, "tok_s": 13127.079772197721}
|
| 48 |
+
{"step": 2000, "val": 0.91477290856758}
|
| 49 |
+
{"step": 2050, "loss": 0.9295798391103745, "lr": 0.00011371924768239318, "tok_s": 10412.662935948929}
|
| 50 |
+
{"step": 2100, "loss": 0.9062058478593826, "lr": 0.00011, "tok_s": 12373.786485106477}
|
| 51 |
+
{"step": 2150, "loss": 0.8629401624202728, "lr": 0.0001062807523176068, "tok_s": 12735.462624419339}
|
| 52 |
+
{"step": 2200, "loss": 0.9598381668329239, "lr": 0.0001025678589074901, "tok_s": 12569.673240732529}
|
| 53 |
+
{"step": 2250, "loss": 0.8634618520736694, "lr": 9.886766318575871e-05, "tok_s": 13198.951203106682}
|
| 54 |
+
{"step": 2250, "val": 0.9149101198460234}
|