step 14897 | val_ppl=10.15

README.md

@@ -25,9 +25,9 @@ Small language model (3.0M parameters) trained from scratch.
 | Total parameters | 3.000M |
 
 ## Training
-- Tokens seen: 15,503,769,600
-- Val loss: 2.2386
-- Val PPL: 9.38
+- Tokens seen: 15,605,382,144
+- Val loss: 2.3172
+- Val PPL: 10.15
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:10ba5f82778819c357260dcef2053c3487a28bd52dd845c1fb87082ecba3bd53
+oid sha256:91307b17086acedaed8a5d3648dd8a1d17ad72e8b7810389c1295e7b41fee3eb
 size 12006544

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c445f823200776bd90b6f8dd3b33313413ee48d055cb1370a36719115334b71b
+oid sha256:53a6eeec25e5d8a70752743422e0441ec837db41d7985b759a22b795704f9c9b
 size 36728175

resume/latest_step.txt

@@ -1 +1 @@
-14800
+14897

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 14800,
-  "val_loss": 2.238574444885254,
-  "val_ppl": 9.379950116046235,
+  "step": 14897,
+  "val_loss": 2.3171666892242433,
+  "val_ppl": 10.14688426324034,
   "params_M": 3.0,
-  "pushed_at": "2026-07-03T23:17:32.204747",
-  "tokens_seen": 15503769600
+  "pushed_at": "2026-07-03T23:34:28.908239",
+  "tokens_seen": 15605382144
 }