from collections import OrderedDict


class LRUCache:
    def __init__(self, max_size=50, max_memory_mb=500):
        self.cache = OrderedDict()
        self.max_size = max_size
        self.max_memory_mb = max_memory_mb
        self.current_memory_mb = 0

    def _estimate_size_mb(self, tensor):
        if hasattr(tensor, 'element_size'):
            return tensor.element_size() * tensor.nelement() / (1024 * 1024)
        return 0

    def get(self, key):
        if key in self.cache:
            self.cache.move_to_end(key)
            return self.cache[key]
        return None

    def set(self, key, value):
        size_mb = self._estimate_size_mb(value)
        if size_mb > self.max_memory_mb:
            return
        if key in self.cache:
            self.current_memory_mb -= self._estimate_size_mb(self.cache[key])
            del self.cache[key]
        while (len(self.cache) >= self.max_size or
               self.current_memory_mb + size_mb > self.max_memory_mb):
            if len(self.cache) == 0:
                break
            old_key, old_value = self.cache.popitem(last=False)
            self.current_memory_mb -= self._estimate_size_mb(old_value)
        self.cache[key] = value
        self.cache.move_to_end(key)
        self.current_memory_mb += size_mb

    def clear(self):
        self.cache.clear()
        self.current_memory_mb = 0