step 5200 | val_ppl=11.43

README.md

@@ -25,9 +25,9 @@ Small language model (3.0M parameters) trained from scratch.
 | Total parameters | 3.000M |
 
 ## Training
-- Tokens seen: 5,028,249,600
-- Val loss: 2.3678
-- Val PPL: 10.67
+- Tokens seen: 5,447,270,400
+- Val loss: 2.4364
+- Val PPL: 11.43
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:63e63d628a147cb4ffff72146b03426eccf085d0a28e4313bf2d0688becc0026
+oid sha256:ef60ce0e7b95e95ceec2f1c6a7f56a499327c8a73b4e7b7a84bfb047773dff94
 size 12006544

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa1ad1b163b6975f2fb39bdd92abf70d5668fc655773bbeeac93af7be93696ee
-size 36728047
+oid sha256:292fb02c548905ef73a9f1bad1a0c4f4546f7fc4efcc2cfeb8b9693b60930360
+size 36728175

resume/latest_step.txt

@@ -1 +1 @@
-4800
+5200

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 4800,
-  "val_loss": 2.3678415299224853,
-  "val_ppl": 10.674327184576963,
+  "step": 5200,
+  "val_loss": 2.436439939956665,
+  "val_ppl": 11.432268647486762,
   "params_M": 3.0,
-  "pushed_at": "2026-07-01T19:21:24.117966",
-  "tokens_seen": 5028249600
+  "pushed_at": "2026-07-01T20:44:21.534655",
+  "tokens_seen": 5447270400
 }