step 14400 | val_ppl=9.38

README.md

@@ -25,9 +25,9 @@ Small language model (3.0M parameters) trained from scratch.
 | Total parameters | 3.000M |
 
 ## Training
-- Tokens seen: 14,665,728,000
-- Val loss: 2.2449
-- Val PPL: 9.44
+- Tokens seen: 15,084,748,800
+- Val loss: 2.2386
+- Val PPL: 9.38
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f948165e7ca0f91903dec3e603dc28d477a1131d455ca83eaa0193f79f8c46b
+oid sha256:07a90c224bec4ed2c3ba29bf23684621bbbcf9822b17b8beacddff64b8071e7c
 size 12006544

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee6efddbffbe8eda9cb7a680ac6695f7bfaf213862c32f6974f34b567ebd4495
+oid sha256:6dfb97b55641ca5f79fb11130b035e3c74dee002414539e42cf49a5130ff8ea8
 size 36728175

resume/latest_step.txt

@@ -1 +1 @@
-14000
+14400

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 14000,
-  "val_loss": 2.2449165924072267,
-  "val_ppl": 9.439628186960764,
+  "step": 14400,
+  "val_loss": 2.238574444885254,
+  "val_ppl": 9.379950116046235,
   "params_M": 3.0,
-  "pushed_at": "2026-07-03T21:42:54.479680",
-  "tokens_seen": 14665728000
+  "pushed_at": "2026-07-03T22:30:10.707879",
+  "tokens_seen": 15084748800
 }