Pretrained models for data-efficient Rainbow. Reported scores matched for most games (sometimes models are a bit worse, sometimes a bit better).
Alien
| Reward | Q-values |
|---|---|
 |
 |
Amidar
| Reward | Q-values |
|---|---|
 |
 |
Assault
| Reward | Q-values |
|---|---|
 |
 |
Asterix
| Reward | Q-values |
|---|---|
 |
 |
Bank Heist
| Reward | Q-values |
|---|---|
 |
 |
Battlezone
| Reward | Q-values |
|---|---|
 |
 |
Boxing
| Reward | Q-values |
|---|---|
 |
 |
Breakout
| Reward | Q-values |
|---|---|
 |
 |
Chopper Command
| Reward | Q-values |
|---|---|
 |
 |
Crazy Climber
| Reward | Q-values |
|---|---|
 |
 |
Demon Attack
| Reward | Q-values |
|---|---|
 |
 |
Freeway
| Reward | Q-values |
|---|---|
 |
 |
Frostbite
| Reward | Q-values |
|---|---|
 |
 |
Gopher
| Reward | Q-values |
|---|---|
 |
 |
H.E.R.O.
| Reward | Q-values |
|---|---|
 |
 |
James Bond 007
| Reward | Q-values |
|---|---|
 |
 |
Kangaroo
| Reward | Q-values |
|---|---|
 |
 |
Krull
| Reward | Q-values |
|---|---|
 |
 |
Kung-Fu Master
| Reward | Q-values |
|---|---|
 |
 |
Ms. Pac-Man
| Reward | Q-values |
|---|---|
 |
 |
Pong
| Reward | Q-values |
|---|---|
 |
 |
Private Eye
| Reward | Q-values |
|---|---|
 |
 |
Q*bert
| Reward | Q-values |
|---|---|
 |
 |
Road Runner
| Reward | Q-values |
|---|---|
 |
 |
Seaquest
| Reward | Q-values |
|---|---|
 |
 |
Up'n Down
| Reward | Q-values |
|---|---|
 |
 |