Update README.md

README.md

@@ -5,13 +5,14 @@ license: apache-2.0
 
 # Model Card: LlavaOLMoBitnet1B
 
-Multimodal Large Language Models (MM-LLMs) have seen significant advancements in the last year, demonstrating impressive performance across tasks. However, to truly democratize AI, models must exhibit strong capabilities and be able to run efficiently on small compute footprints accessible by most. Part of this quest, we introduce LLaVaOLMoBitnet1B - the first Ternary Multimodal LLM capable of accepting Image(s)+Text inputs to produce coherent textual responses. The model is fully open-sourced along with training scripts to encourage further research in this space. We also release a technical report highlighting the training proecss, eval details, challenges associated with ternary models and future oppurtunities.   
+Multimodal Large Language Models (MM-LLMs) have seen significant advancements in the last year, demonstrating impressive performance across tasks. However, to truly democratize AI, models must exhibit strong capabilities and be able to run efficiently on small compute footprints accessible by most. Part of this quest, we introduce LLaVaOLMoBitnet1B - the first Ternary Multimodal LLM capable of accepting Image(s)+Text inputs to produce coherent textual responses. The model is fully open-sourced along with training scripts to encourage further research in this space. We also release a technical report highlighting the training process, eval details, challenges associated with ternary models and future opportunities.   
 
 Authors: Jainaveen Sundaram, Ravishankar Iyer 
 
 
 ### Training Data
-TODO - Add info on Training data (if applicable)
+Two step training pipeline outlined in the LLaVa1.5 paper, consisting of two phases: (1) A Pre-training phase for feature alignment followed by an (2) End-to-end instruction fine-tuning
+The pre-training phase involves 1 epoch on a filtered subset of 595K Conceptual Captions [2], with only the projection layer weights updated. For instruction fine-tuning, we use 1 epoch of the LLaVa-Instruct-150K dataset, with both projection layer and LLM weights updated.
 
 ### Evaluation Data
 TODO - Add info on Eval Data (if applicable)