Update README.md

README.md

@@ -39,6 +39,10 @@ Fine-tuning the base GALACTICA models on the 52k instruction-response pairs in t
 The GALPACA weights are made available for use with the `transformers` library.
 TODO: add example inference usage.
 
+## Training Resources
+
+GALPACA 30B was fine-tuned in about 8 hours using 16 A100 80GB GPUS at an effective batch-size of 1024 and with a maximum context window of 384 tokens. This model was trained using DeepSpeed Stage 3 optimizations.
+
 ## Performance and Limitations
 
 More information about the performance and limitations of this family of models can be found on the original GALACTICA model card.