buffer 0 step 0 actions torch.float32 (64,) 7b924a8f1dd3c16afdd71cc01fdfe2292f0fbf4896a83f0dbbf7a350d5b9cba9
buffer 0 step 0 logprobs torch.bfloat16 (64,) eb69733230eee073323bbad10b8c661defbfd70025d4e5e9fcd0c9b690f890cd
buffer 0 step 0 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 0 step 0 draws torch.int64 (64,) d273174b7107c659a3e4b5ae07a99ce24122b99c1cd33f51ec72f44cec45cafd
buffer 0 step 1 actions torch.float32 (64,) 9a9da7f33ba1312185fd31060aaac8d12c7b70b267a4b727583675579b9d0dee
buffer 0 step 1 logprobs torch.bfloat16 (64,) 6f5a05626c78cad5a4169c502934caff162b989e99b234adaba26a525213ea84
buffer 0 step 1 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 0 step 1 draws torch.int64 (64,) 33eb873a887f08c0ce87bbc9e9e171784c372fcfdc5f76aaaa40c5fdf1ba0ca1
buffer 0 step 2 actions torch.float32 (64,) 97d6574875e2a5e52b127c4a39f3a3e0534c291dd1910010e059b15887646ffc
buffer 0 step 2 logprobs torch.bfloat16 (64,) 445bc46e91be0b849df8f637e901c05057094fd8c8e9613524b2ecc01958cefe
buffer 0 step 2 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 0 step 2 draws torch.int64 (64,) a5b44ead74453f794cef4e12c88de3bddd5c4aa627eec8cb0d5aa1a5dba04eb3
buffer 1 step 0 actions torch.float32 (64,) 50ac487d182a1cc6130e948a0389c32c2793b0fb70351a242e41b6551d3236d1
buffer 1 step 0 logprobs torch.bfloat16 (64,) 55ae6320fa85c2b1923704857e42b292f2dbe34046ff1597ece742a589f81107
buffer 1 step 0 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 1 step 0 draws torch.int64 (64,) 643feb399c68dd6d8aa99927f90aad55e758387235048fa37dd3e72fba37892e
buffer 1 step 1 actions torch.float32 (64,) 7b4774b82a7a3160a25d4d88c091b0ca0217bc08c1103d4b2147bd63d5b07f87
buffer 1 step 1 logprobs torch.bfloat16 (64,) e4029b24a07695d9c0ebc113255c7a6e3f7a9b8f139a2ac8e2620624ddca6f12
buffer 1 step 1 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 1 step 1 draws torch.int64 (64,) b02972804baaa4564de545587e4805e745f83b95e3fef43cc1a6fd7d4e1b9bac
buffer 1 step 2 actions torch.float32 (64,) 27561979c3e0d4c5cf1ae687faab2691cb56e22c778c8ed060acbfcf03201796
buffer 1 step 2 logprobs torch.bfloat16 (64,) 7166d8347bb9926d12a1841043e2ef302e52f3e3f7f7a77cfbf6cc569528b1a4
buffer 1 step 2 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 1 step 2 draws torch.int64 (64,) a185e9d2c45adfe10a5f77ac87ce808162e34e30accd86a4087da2c67a027dfa
buffer 2 step 0 actions torch.float32 (64,) 0df650f10bb2edc8010a0db19a15b9549ecfbcda74b8061eb9a5a522ccae1d65
buffer 2 step 0 logprobs torch.bfloat16 (64,) 3c4065e483366428a5ad61f0cc14762ecbb55d86ef244c2ec60b12999208a994
buffer 2 step 0 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 2 step 0 draws torch.int64 (64,) 97cc2a1f05cbd9c1a82256862cc55f6a84bc343f7bb82224328c6a336fc7fe52
buffer 2 step 1 actions torch.float32 (64,) 7415c6053ce2c2d77f42a5568c3b6e96b49bff644a8101ad9cd2fb5b958826eb
buffer 2 step 1 logprobs torch.bfloat16 (64,) f118d9171c5706235743767b6934d51667b1ae6832660cda3f98a28677b139fb
buffer 2 step 1 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 2 step 1 draws torch.int64 (64,) 63258062a69abeeae3a4dfe75aa4194dc6d316f25da2e0bdd5a045a1936e6d62
buffer 2 step 2 actions torch.float32 (64,) f86f30711f40ba54b3d0c3b3b4ae4f9d74fb0bd7e65ec9f1cff59951cdf1257c
buffer 2 step 2 logprobs torch.bfloat16 (64,) daf60ab27c4df75e107db9b2e40b06475589b8e07102bb8c0c661f7c105b3ef9
buffer 2 step 2 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 2 step 2 draws torch.int64 (64,) a95c1043a42914777cb13754e45a043a432072dfe05e00d860edafce12f58405
buffer 3 step 0 actions torch.float32 (64,) d1d82fbdc99d6471bb77fb899722a641f26bbedbe92fd7772e3f9007705a1b74
buffer 3 step 0 logprobs torch.bfloat16 (64,) 84ed78e1bcb7da946e97829fd3fde2a07f6eb2934a39338de311a5443c57839e
buffer 3 step 0 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 3 step 0 draws torch.int64 (64,) 663bd7fd9e78481b14463e40a31cf6ef1ee0cd7894e4946e92afd2930fab5e43
buffer 3 step 1 actions torch.float32 (64,) cf4078e27a9f3d42980dbec80320d13e7b55691176b1c9dff8e705d5faa4f633
buffer 3 step 1 logprobs torch.bfloat16 (64,) 75454cf301b2672beee3d4ad9a2ded971d05fdf8f39dc15984a7e6fb0202d5f1
buffer 3 step 1 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 3 step 1 draws torch.int64 (64,) 5d471cca17f07ca8279860d2f8bb665b6c293b70a848e01c2d8be1d0c6e181c9
buffer 3 step 2 actions torch.float32 (64,) 9f7e832db011195f667838185658d4915d7e35bd8e6b160cc32d32d124cd5992
buffer 3 step 2 logprobs torch.bfloat16 (64,) e0e68160d8016ab0b82a2cf4f26596b1adf1eb56f55b9116c04ba8c962bafa0a
buffer 3 step 2 values torch.bfloat16 (64,) 52b9dd57f7dc1df1922638995bc4b64c99aabd8120614b9e2f4184d592d64fb5
buffer 3 step 2 draws torch.int64 (64,) ad36df9f8da64eb5c5ede7d727b1458dd47179df7cd5b547f9d4027ea9b4d64b
