support many checkpoints in evaluate

docs/experiment-notes-stablelm.md

@@ -138,4 +138,32 @@
   - 550: 
   - 600: 0.9473684210526315
   - 650: 0.9387651821862348
-  - Final: 0.9463562753036437
+  - Final: 0.9463562753036437
+
+
+## stablelm-2-1_6b-zephyr
+
+# rev3
+- full fine tune
+- epochs: 1
+- 2048 train ctx
+- batch size 32
+- learning rate 1e-5
+- weight decay 0.1
+- gradient clipping 1.0
+- dataset size: medium
++ evaluation results:
+  - 100: 0.35779352226720645
+  - 200: 0.5247975708502024
+  - 300: 0.5339068825910931
+  - 400: 0.6280364372469636
+  - 500: 0.6923076923076923
+  - 600: 0.7064777327935222
+  - 700: 0.7135627530364372
+  - 800: 0.7044534412955465
+  - 900: 0.707995951417004
+  - 1000: 0.718117408906882
+  - Final: 0.7145748987854251
+
+# rev4
+- dataset size: large