I wonder if the dip is more about LLama3 70b training and data than a change in sentiment. The data cut-off was Dec 2023 for 70b. That looks to coincide with the reversal of the dip.
We don't try to retrieve articles/topics from the model, which would be affected by the cutoff, just asking it to analyze the sentiment or summarize the content provided in a prompt
edit I realized too late I had the years off. It is pure coincidence of month, not a real data bias. Sorry! I still think it would be interesting to see a 7B comparison but that is just to see how well a small model could spot big trends compared to a bigger one.