Which of these datasets were not utilized in Llama3.3-70B training?

[vthakkarumn]

I am trying to find a dataset (or a set of clinical notes) that were not utilized during Llama3.3 (specifically 70B model) training. Which of the below datasets were not utilized in Llama3.3 training? Can anyone help please?

1. MIMIC-III (Medical Information Mart for Intensive Care III)
2. i2b2 (Informatics for Integrating Biology & the Bedside)
3. n2c2 (National NLP Clinical Challenges)
4. SHARE (Stanford Health AI Research and Evaluation)
5. PhysioNet
6. CLEF eHealth
7. TREC Medical Records Track
8. OpenNotes
9. eICU Collaborative Research Database
10. PubMed Central (PMC) Open Access Subset

Thanks!