update benchmark and README

BENCHMARK.md

@@ -32,10 +32,10 @@ The specs for these are contained in the [`slm_lab/spec/benchmark`](https://gith
 
 | Env. \ Alg. | A2C (GAE) | A2C (n-step) | PPO | DQN | DDQN+PER |
 |:---|---|---|---|---|---|
-| Breakout | 389.99 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62019989-0171c000-b176-11e9-94da-017b146afe65.png"></details> | 391.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020340-6c6fc680-b177-11e9-8aa1-9ac5c2001783.png"></details> | **425.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067085-c0b28f00-b1e7-11e9-9dd5-c52b6104878f.png"></details> | 65.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100441-9ba13900-b246-11e9-9373-95c6063915ab.png"></details> |  |
-| Pong | 20.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020247-10a53d80-b177-11e9-9f0d-1433d4d87210.png"></details> | 19.66 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020342-6f6ab700-b177-11e9-824e-75f431dc14ec.png"></details> | **20.09** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067100-c6a87000-b1e7-11e9-919e-ad68e4166213.png"></details> | 18.34 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100450-9fcd5680-b246-11e9-8170-2ad4473e8294.png"></details> |  |
-| Qbert | 13,328.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020263-261a6780-b177-11e9-8936-22a74d2405d3.png"></details> | 13,259.19 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020347-742f6b00-b177-11e9-8bfb-edfcfd44c8b7.png"></details> | **13,691.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067104-cb6d2400-b1e7-11e9-9c4f-9eaac265d7d6.png"></details> | 4,787.79 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100455-a4920a80-b246-11e9-8ca5-d4dc1ce3d76f.png"></details> |  |
-| Seaquest | 892.68 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020266-29adee80-b177-11e9-83c2-fafbdbb982b9.png"></details> | **1,686.08** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020350-772a5b80-b177-11e9-8917-e3c8a745cd08.png"></details> | 1,583.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067113-cf994180-b1e7-11e9-870b-b9bba71f2a7e.png"></details> | 1,118.50 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100462-a9ef5500-b246-11e9-8699-9356ff81ff93.png"></details> |  |
+| Breakout <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232119-554cf680-b37a-11e9-9059-3e49bbb799d2.png"><img src="https://user-images.githubusercontent.com/8209263/62232118-554cf680-b37a-11e9-9d5b-dd2ddf527305.png"></details> | 389.99 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62019989-0171c000-b176-11e9-94da-017b146afe65.png"></details> | 391.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020340-6c6fc680-b177-11e9-8aa1-9ac5c2001783.png"></details> | **425.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067085-c0b28f00-b1e7-11e9-9dd5-c52b6104878f.png"></details> | 65.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100441-9ba13900-b246-11e9-9373-95c6063915ab.png"></details> | 181.72 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230967-dd7dcc80-b377-11e9-965b-60a9f3d5a7a1.png"></details> |
+| Pong <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232135-5b42d780-b37a-11e9-9454-ff2d109ef4f4.png"><img src="https://user-images.githubusercontent.com/8209263/62232134-5b42d780-b37a-11e9-892f-a84ea8881e78.png"></details> | 20.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020247-10a53d80-b177-11e9-9f0d-1433d4d87210.png"></details> | 19.66 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020342-6f6ab700-b177-11e9-824e-75f431dc14ec.png"></details> | 20.09 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067100-c6a87000-b1e7-11e9-919e-ad68e4166213.png"></details> | 18.34 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100450-9fcd5680-b246-11e9-8170-2ad4473e8294.png"></details> | **20.44** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230975-e2428080-b377-11e9-8970-6917ae80c0b4.png"></details> |
+| Qbert <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232149-60078b80-b37a-11e9-99bb-cedc9fe064d5.png"><img src="https://user-images.githubusercontent.com/8209263/62232148-60078b80-b37a-11e9-9610-17ac447a479f.png"></details> | 13,328.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020263-261a6780-b177-11e9-8936-22a74d2405d3.png"></details> | 13,259.19 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020347-742f6b00-b177-11e9-8bfb-edfcfd44c8b7.png"></details> | **13,691.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067104-cb6d2400-b1e7-11e9-9c4f-9eaac265d7d6.png"></details> | 4,787.79 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100455-a4920a80-b246-11e9-8ca5-d4dc1ce3d76f.png"></details> | 11,673.52 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230986-e79fcb00-b377-11e9-8861-3686954b7e1a.png"></details> |
+| Seaquest <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232168-6bf34d80-b37a-11e9-9564-fa3609dc5c75.png"><img src="https://user-images.githubusercontent.com/8209263/62232167-6bf34d80-b37a-11e9-8db3-c79a0e78292b.png"></details> | 892.68 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020266-29adee80-b177-11e9-83c2-fafbdbb982b9.png"></details> | 1,686.08 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020350-772a5b80-b177-11e9-8917-e3c8a745cd08.png"></details> | 1,583.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067113-cf994180-b1e7-11e9-870b-b9bba71f2a7e.png"></details> | 1,118.50 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100462-a9ef5500-b246-11e9-8699-9356ff81ff93.png"></details> | **3,751.34** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230991-ebcbe880-b377-11e9-8de4-a01379d1d61c.png"></details> |
 
 
 

README.md

@@ -39,6 +39,17 @@ Below shows the latest benchmark status. See the full [benchmark results here](h
 
 Due to their standardized design, all the algorithms can be parallelized asynchronously using Hogwild. Hence, SLM Lab also includes A3C, distributed-DQN, distributed-PPO.
 
+#### Atari benchmark
+
+This benchmark table is pulled from [PR396](https://github.com/kengz/SLM-Lab/pull/396). See the full [benchmark results here](https://github.com/kengz/SLM-Lab/blob/master/BENCHMARK.md).
+
+| Env. \ Alg. | A2C (GAE) | A2C (n-step) | PPO | DQN | DDQN+PER |
+|:---|---|---|---|---|---|
+| Breakout <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232119-554cf680-b37a-11e9-9059-3e49bbb799d2.png"><img src="https://user-images.githubusercontent.com/8209263/62232118-554cf680-b37a-11e9-9d5b-dd2ddf527305.png"></details> | 389.99 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62019989-0171c000-b176-11e9-94da-017b146afe65.png"></details> | 391.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020340-6c6fc680-b177-11e9-8aa1-9ac5c2001783.png"></details> | **425.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067085-c0b28f00-b1e7-11e9-9dd5-c52b6104878f.png"></details> | 65.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100441-9ba13900-b246-11e9-9373-95c6063915ab.png"></details> | 181.72 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230967-dd7dcc80-b377-11e9-965b-60a9f3d5a7a1.png"></details> |
+| Pong <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232135-5b42d780-b37a-11e9-9454-ff2d109ef4f4.png"><img src="https://user-images.githubusercontent.com/8209263/62232134-5b42d780-b37a-11e9-892f-a84ea8881e78.png"></details> | 20.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020247-10a53d80-b177-11e9-9f0d-1433d4d87210.png"></details> | 19.66 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020342-6f6ab700-b177-11e9-824e-75f431dc14ec.png"></details> | 20.09 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067100-c6a87000-b1e7-11e9-919e-ad68e4166213.png"></details> | 18.34 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100450-9fcd5680-b246-11e9-8170-2ad4473e8294.png"></details> | **20.44** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230975-e2428080-b377-11e9-8970-6917ae80c0b4.png"></details> |
+| Qbert <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232149-60078b80-b37a-11e9-99bb-cedc9fe064d5.png"><img src="https://user-images.githubusercontent.com/8209263/62232148-60078b80-b37a-11e9-9610-17ac447a479f.png"></details> | 13,328.32 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020263-261a6780-b177-11e9-8936-22a74d2405d3.png"></details> | 13,259.19 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020347-742f6b00-b177-11e9-8bfb-edfcfd44c8b7.png"></details> | **13,691.89** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067104-cb6d2400-b1e7-11e9-9c4f-9eaac265d7d6.png"></details> | 4,787.79 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100455-a4920a80-b246-11e9-8ca5-d4dc1ce3d76f.png"></details> | 11,673.52 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230986-e79fcb00-b377-11e9-8861-3686954b7e1a.png"></details> |
+| Seaquest <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62232168-6bf34d80-b37a-11e9-9564-fa3609dc5c75.png"><img src="https://user-images.githubusercontent.com/8209263/62232167-6bf34d80-b37a-11e9-8db3-c79a0e78292b.png"></details> | 892.68 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020266-29adee80-b177-11e9-83c2-fafbdbb982b9.png"></details> | 1,686.08 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62020350-772a5b80-b177-11e9-8917-e3c8a745cd08.png"></details> | 1,583.04 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62067113-cf994180-b1e7-11e9-870b-b9bba71f2a7e.png"></details> | 1,118.50 <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62100462-a9ef5500-b246-11e9-8699-9356ff81ff93.png"></details> | **3,751.34** <details><summary><i>graph</i></summary><img src="https://user-images.githubusercontent.com/8209263/62230991-ebcbe880-b377-11e9-8de4-a01379d1d61c.png"></details> |
+
 ### Environments
 
 SLM Lab integrates with multiple environment offerings: