Using GNU `sed`, and assuming your file is small enough to fit in memory, you can do this:

```
$ sed -z 's/.*jum//' file 
ps over
the lazy
dog
```

The `-z` tells `sed` to treat its input as NUL-separated, meaning that it expects a NUL (`\0`) to signidfy the end of each line. Since your file will have no NUL characters, this has the effect of sed treating the entire file as a single line so we can do the whole thing in a single `s///` operation. 

Note that this will find the _last_ `jum` in the file and delete everything before that. If there are other occurrences of `jum` earlier, those will be removed.