Hi,
A high Rfree can indicate that the refinement strategy is not appropriate.
Several suggestions:
1) At 2.5 A resolution, you probably don't need reference model restraints.
2 ) Simulated annealing (SA) is most useful early in refinement, when the model is far from convergence. Manually built models, or MR solutions involving significant local conformational changes, are common inputs where SA can improve over simple gradient-driven refinement. It is generally less helpful later in refinement, and/or at high resolution.
3) Individual B-factors and TLS are mutually exclusive, so did you define different strategies for different areas? If yes, you could try to revisit the parameterization.
4) It might be also worthwhile to double check the secondary structure assignment for secondary structure restraints.
Weight optimization is typically done at the end of refinement. So while you are trying to figure out the strategy, you might want to disable it, as it is time-consuming.
Best wishes,
Dorothee