step 800 | val_ppl=18.87

README.md

@@ -25,9 +25,9 @@ Small language model (9.9M parameters) trained from scratch.
 | Total parameters | 9.853M |
 
 ## Training
-- Tokens seen: 419,020,800
-- Val loss: 3.4701
-- Val PPL: 32.14
+- Tokens seen: 838,041,600
+- Val loss: 2.9375
+- Val PPL: 18.87
 
 ## Usage
 ```python

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b687f62f3f58daccffe6c0697badab591a4ecb542386537822f220a8ceefab5a
+oid sha256:0aa439042ac467021b8680b0cc1a440ca060e9d316df6edd36c34617b2c8aa38
 size 39421616

resume/ckpt.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20e45cc79bac0b46a1db8188779d2ca1528781617101b6cbde5cf6d1f4fba09a
+oid sha256:c8b846794e1ec343ec39fdbe203f94103c67d9a658c88c769590edaf29584d77
 size 119811943

resume/latest_step.txt

@@ -1 +1 @@
-400
+800

training_meta.json

@@ -1,8 +1,8 @@
 {
-  "step": 400,
-  "val_loss": 3.4701391414642333,
-  "val_ppl": 32.14121430925628,
+  "step": 800,
+  "val_loss": 2.9375060623168947,
+  "val_ppl": 18.868730147139708,
   "params_M": 9.853,
-  "pushed_at": "2026-06-17T09:38:56.823655",
-  "tokens_seen": 419020800
+  "pushed_at": "2026-06-17T11:23:42.505200",
+  "tokens_seen": 838041600
 }