A good summary in The Week.
To summarise their summary, the possible reasons, all of them disputed by experts and statisticians, are as follows:
1. More cases are among younger people, who are far less vulnerable. Or to put it more crudely, the most vulnerable died in the first wave.
2. The death rate = deaths ÷ diagnosed cases. The number of diagnosed cases depends on how many people are tested, so for a given number of deaths, the apparent death rate falls if there is more testing.*
3. Hospitals have got much better at treating the disease. Medicine is about trial and error, and learning from your mistakes.
4. The virus is getting less deadly. This has been observed with most viruses - a virus' best strategy is not to kill its hosts. Normally this takes years or decades, so it would be surprising if it happened in less than a year.
5. Smaller doses. An infectious person with a face mask spreads smaller doses, which are easier to fight off.
To sum up, nobody knows. Some experts even dispute that the death rate is going down at all.
* Best of luck trying to make any sense of the statistics, you can try comparing anything with anything for European countries and the coefficient of correlation is close to zero. The coefficient for number of cases-v-number of tests is about 0.2. The coefficient for death rate-v-tests per million population is also about 0.2.
Also, why is the number of deaths per million in Switzerland three times as high as in neighbouring Austria? Why is the number of deaths per million population in Belgium a hundred times as high as in Slovakia?
Why are the death rates lowest in the east (Slovakia, Latvia, Greece) and highest in the west (Belgium, Spain, UK)? Why is there also a slight gradient from north (fewer deaths) to south (more deaths)? The UK and Sweden are outliers in the north; Greece is an outlier in the south.