step 6800 | val_ppl=11.39

README.md

@@ -25,9 +25,9 @@ Small language model (3.0M parameters) trained from scratch.
 | Total parameters | 3.000M |
 
 ## Training
-- Tokens seen: 6,704,332,800
-- Val loss: 2.4364
-- Val PPL: 11.43
+- Tokens seen: 7,123,353,600
+- Val loss: 2.4326
+- Val PPL: 11.39
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe30b0d5cdc879fe5be64e852b78a36b10c659c2f41c14380e7a83a0121a5b21
+oid sha256:2237626f41d2306f569ac4637339815059fbc7fd3a201191793712e87f12960b
 size 12006544

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ee11d620694a09c17fafe818ce3043e07d0f3d597e8b4018367819ffea7f3a7
+oid sha256:a57133e1eec4d59963839d2650ca5f1802401b235f5240b6fb6719dfffbdc2cf
 size 36728175

resume/latest_step.txt

@@ -1 +1 @@
-6400
+6800

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 6400,
-  "val_loss": 2.436439939956665,
-  "val_ppl": 11.432268647486762,
+  "step": 6800,
+  "val_loss": 2.43262853767395,
+  "val_ppl": 11.388778604339521,
   "params_M": 3.0,
-  "pushed_at": "2026-07-01T23:02:00.771009",
-  "tokens_seen": 6704332800
+  "pushed_at": "2026-07-01T23:47:55.024852",
+  "tokens_seen": 7123353600
 }