step 4000 | val_ppl=11.43

README.md

@@ -25,9 +25,9 @@ Small language model (9.9M parameters) trained from scratch.
 | Total parameters | 9.853M |
 
 ## Training
-- Tokens seen: 3,771,187,200
-- Val loss: 2.4536
-- Val PPL: 11.63
+- Tokens seen: 4,190,208,000
+- Val loss: 2.4361
+- Val PPL: 11.43
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7c8fd1a940963dad4d80891d35dacbaea76b3bd45eeb3aca7fef7970e6170a5
+oid sha256:3a623c1f904bdb9f46ed0718c33f1996a9db6692de446fea9b322554bff206d3
 size 39421616

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:626d3eef635b7be8b1a6348cd6df2a8f73cd81cf03415b62cd8ed08a00d166cd
+oid sha256:445874cf6b1f4aafe7ac6cc1436c8664c25eae2577357c64d852ca181d1add5e
 size 119812327

resume/latest_step.txt

@@ -1 +1 @@
-3600
+4000

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 3600,
-  "val_loss": 2.4536261505126955,
-  "val_ppl": 11.630444088190558,
+  "step": 4000,
+  "val_loss": 2.4361114742279053,
+  "val_ppl": 11.428514155678789,
   "params_M": 9.853,
-  "pushed_at": "2026-06-18T00:27:20.366594",
-  "tokens_seen": 3771187200
+  "pushed_at": "2026-06-18T02:22:12.443456",
+  "tokens_seen": 4190208000
 }