masatof commited on
Commit
3ec9219
·
verified ·
1 Parent(s): 21f9211

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -1
README.md CHANGED
@@ -37,7 +37,9 @@ The model is trained with autoregressive next-token prediction.
37
 
38
  Training summary:
39
 
40
- * pretraining data: approximately **1T tokens**
 
 
41
  * optimizer: **ROCA Optimizer**
42
  * dtype: bfloat16
43
 
 
37
 
38
  Training summary:
39
 
40
+ * pretraining data: approximately **1T tokens** with 8k context.
41
+ * midtraining data: approximately **50B tokens** with 32k context.
42
+ * posttraining data (SFT & modified-SLIME): approximately **5B tokens**.
43
  * optimizer: **ROCA Optimizer**
44
  * dtype: bfloat16
45