Update README.md

README.md

@@ -31,7 +31,7 @@ Custom decoder-only transformer:
 - **Tokens seen:** ~4B
 - **Steps:** 30,000
 - **Optimizer:** AdamW (lr=3e-4, cosine decay to 3e-5)
-- **Hardware:** Single A100 80GB
+- **Hardware:** Single A100 40GB
 
 ## Installation
 
@@ -57,4 +57,5 @@ print(tokenizer.decode(out[0], skip_special_tokens=True))
 
 ## License
 
-Model weights: MIT. Training data: ODC-By 1.0.
+Model weights: MIT. 
+Training data: This work uses the FineWeb-Edu dataset, available under the Open Data Commons Attribution License (ODC-By 1.0).