step 9200 | val_ppl=9.71

README.md

@@ -25,9 +25,9 @@ Small language model (3.0M parameters) trained from scratch.
 | Total parameters | 3.000M |
 
 ## Training
-- Tokens seen: 9,218,457,600
-- Val loss: 2.2903
-- Val PPL: 9.88
+- Tokens seen: 9,637,478,400
+- Val loss: 2.2735
+- Val PPL: 9.71
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8dd87e940cbffb46d64cbe2a21a79b28720d2d4e603b92abf208a04cbf081ee
+oid sha256:327a67e4a7c2514f6802765ff7e1bc337fa6895a44f09f64534eaa713b0f017d
 size 12006544

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d30b2cd5bd3fe4a1cf77c1b1e9083556c97ee27bd1d7d649bb457de1cc8e23b6
+oid sha256:d0e2b261f10fd5f93f5335736792e929019207f29bbdc03d4036645f65cd2b73
 size 36728175

resume/latest_step.txt

@@ -1 +1 @@
-8800
+9200

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 8800,
-  "val_loss": 2.290264854888916,
-  "val_ppl": 9.877553453080482,
+  "step": 9200,
+  "val_loss": 2.2734714436340333,
+  "val_ppl": 9.713060698803686,
   "params_M": 3.0,
-  "pushed_at": "2026-07-02T03:37:31.685146",
-  "tokens_seen": 9218457600
+  "pushed_at": "2026-07-02T04:23:27.563926",
+  "tokens_seen": 9637478400
 }