You can't calibrate models using the data it's supposed to predict. That's not how modeling is supposed to work.
"...GCMs are not sufficiently reliable to distinguish between natural and man-made causes of the temperature increase in the 20th century. Some of the predictions from GCMs are accompanied by standard errors, as in statistical analysis. But since the GCMs are deterministic models one cannot interpret these standard errors in the same way as in statistics. GCMs are typically evaluated applying the same observations used to calibrate the model parameters. In an article in Science, Voosen (2016) writes; “Indeed, whether climate scientists like to admit it or not, nearly every model has been calibrated precisely to the 20th century climate records – otherwise it would have ended up in the trash”. Unfortunately,models that match 20th century data as a result of calibration using the same 20th century data are of dubious quality for determining the causes of the 20th century temperature variability. The problem is that some of the variables representing sources of climate variability other than greenhouse gases are not properly controlled for during the calibrations. The resulting calibration of the climate sensitivity may therefore be biased. Further critical evaluations are given by several authors, such as Essex (2022)..."