nova819/sole-r1-v3-sft
Image-Text-to-Text • 770k • Updated • 10
Datasets:
image stringlengths 78 78 | problem stringclasses 990
values | solution_orig stringclasses 101
values | solution stringlengths 741 952 | prompt stringlengths 2.09k 2.37k | image_name stringlengths 63 75 | from_annotation int64 0 0 | completion stringlengths 741 952 | data_source stringclasses 1
value | progress float64 0 100 | metadata dict |
|---|---|---|---|---|---|---|---|---|---|---|
images/f0/f012e4138a60e3e072f37248e22b8345888ae5e10a83946d9bdc4a65e3295af9.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_0.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 0 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/f0/f012e4138a60e3e072f37248e22b8345888ae5e10a83946d9bdc4a65e3295af9.png",
"timestep": 0,
"previous_timestep": 0,
"previous_source": "sampled_sequence",
"from_zero"... |
images/bc/bc7ca265553fb2ed90945d356b2931db9301fa500b30f8715e1ec6284efc96e2.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 4 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_4.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 4.406872 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/bc/bc7ca265553fb2ed90945d356b2931db9301fa500b30f8715e1ec6284efc96e2.png",
"timestep": 4,
"previous_timestep": 0,
"previous_source": "sampled_sequence",
"from_zero"... |
images/49/4925b70183c04e93fd0ffbe8803e11d7c9e9e9deb4778e1503ac0ec82ea49665.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 4 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact with... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_4_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact with... | libero_spatial_rlinf_sole_cot_v6 | 4.406872 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/49/4925b70183c04e93fd0ffbe8803e11d7c9e9e9deb4778e1503ac0ec82ea49665.png",
"timestep": 4,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true,... |
images/fc/fcdb4d21e1ea37b15aa59c0fdba0c67502278eec224858bf0f9a0685ed51dd53.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 11 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_8.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 10.633302 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/fc/fcdb4d21e1ea37b15aa59c0fdba0c67502278eec224858bf0f9a0685ed51dd53.png",
"timestep": 8,
"previous_timestep": 4,
"previous_source": "sampled_sequence",
"from_zero"... |
images/22/222ac42395bc1fc8c99639308e084cfcb99913a50cfc794cdd2ae3f8ea0ae31e.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 15 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_11.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 14.664356 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/22/222ac42395bc1fc8c99639308e084cfcb99913a50cfc794cdd2ae3f8ea0ae31e.png",
"timestep": 11,
"previous_timestep": 8,
"previous_source": "sampled_sequence",
"from_zero... |
images/4b/4b805fda1ac11411a38ffbb15293189c7dda9acdfbe306e3ad07fbb2e88123e6.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 19 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_15.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 18.671391 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/4b/4b805fda1ac11411a38ffbb15293189c7dda9acdfbe306e3ad07fbb2e88123e6.png",
"timestep": 15,
"previous_timestep": 11,
"previous_source": "sampled_sequence",
"from_zer... |
images/41/41778d1020b6004da4c69cdb5ac56a7e433bfcd6afcebd66ed778ab8b46d513a.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 24 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_19.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 24.361193 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/41/41778d1020b6004da4c69cdb5ac56a7e433bfcd6afcebd66ed778ab8b46d513a.png",
"timestep": 19,
"previous_timestep": 15,
"previous_source": "sampled_sequence",
"from_zer... |
images/fa/fa51d8dcf88c8a06b864bcc6ad24d282252ce63925386f87168b4fc9bd4d65ed.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 32 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_23.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 31.756559 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/fa/fa51d8dcf88c8a06b864bcc6ad24d282252ce63925386f87168b4fc9bd4d65ed.png",
"timestep": 23,
"previous_timestep": 19,
"previous_source": "sampled_sequence",
"from_zer... |
images/52/52cf85541cc25a4609856548a6e03a6a74ed1ff572e43453222240bb2ad7abec.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 41 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_27.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 40.901409 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/52/52cf85541cc25a4609856548a6e03a6a74ed1ff572e43453222240bb2ad7abec.png",
"timestep": 27,
"previous_timestep": 23,
"previous_source": "sampled_sequence",
"from_zer... |
images/6a/6a2ee81d42b7b04a26fc841523b519ac9a6f5d237a2db096930c0cf1762fb82b.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 41 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact with... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_27_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact with... | libero_spatial_rlinf_sole_cot_v6 | 40.901409 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/6a/6a2ee81d42b7b04a26fc841523b519ac9a6f5d237a2db096930c0cf1762fb82b.png",
"timestep": 27,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/b7/b7fcbd15f502800d9e43ea5e4410bfb223d6470055e5bf2e97e826a2a318123b.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 48 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_30.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 48.158512 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/b7/b7fcbd15f502800d9e43ea5e4410bfb223d6470055e5bf2e97e826a2a318123b.png",
"timestep": 30,
"previous_timestep": 27,
"previous_source": "sampled_sequence",
"from_zer... |
images/8f/8fcc054ed3b6a12c1ca9e339ae9ff765bb484966cc48d9327804a64e99b9ae41.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 58 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_34.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 58.034519 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/8f/8fcc054ed3b6a12c1ca9e339ae9ff765bb484966cc48d9327804a64e99b9ae41.png",
"timestep": 34,
"previous_timestep": 30,
"previous_source": "sampled_sequence",
"from_zer... |
images/79/79730471f817f5c03fd609ee719ce777fd85fc18da522ff42dd5974d7ce8967c.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 65 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_38.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 64.544411 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/79/79730471f817f5c03fd609ee719ce777fd85fc18da522ff42dd5974d7ce8967c.png",
"timestep": 38,
"previous_timestep": 34,
"previous_source": "sampled_sequence",
"from_zer... |
images/c9/c9cb8dcb31c528086672f6fdffc1ec4d9c21848d726ede8eaa7c513439b511bc.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 67 | <think>At the previous timestep, the gripper was near the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object. Compared wi... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_41.png | 0 | <think>At the previous timestep, the gripper was near the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object. Compared wi... | libero_spatial_rlinf_sole_cot_v6 | 66.546631 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/c9/c9cb8dcb31c528086672f6fdffc1ec4d9c21848d726ede8eaa7c513439b511bc.png",
"timestep": 41,
"previous_timestep": 38,
"previous_source": "sampled_sequence",
"from_zer... |
images/3f/3f4f332bcd867e66afa07bf4e01f82b9db4abeb1aa449870d50dd7b1fc1431dd.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 67 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object. Compared ... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_41_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object. Compared ... | libero_spatial_rlinf_sole_cot_v6 | 66.546631 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/3f/3f4f332bcd867e66afa07bf4e01f82b9db4abeb1aa449870d50dd7b1fc1431dd.png",
"timestep": 41,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/a9/a9aac8c9ba25d91c6510109816e0a4926db058f6a0afa6a53023661cb22883bf.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 67 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_42.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 67.172142 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/a9/a9aac8c9ba25d91c6510109816e0a4926db058f6a0afa6a53023661cb22883bf.png",
"timestep": 42,
"previous_timestep": 41,
"previous_source": "sampled_sequence",
"from_zer... |
images/06/064d452fbeebf056b1eb6da7aa3758b261aa31c005d1b50aa4c0a5d92a31b4c7.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 68 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_46.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 68.021996 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/06/064d452fbeebf056b1eb6da7aa3758b261aa31c005d1b50aa4c0a5d92a31b4c7.png",
"timestep": 46,
"previous_timestep": 42,
"previous_source": "sampled_sequence",
"from_zer... |
images/1a/1aa359a0d1aa53fa89e8673ca45e4b919911cd21a280220a0199b98196255c57.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 71 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_49.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 70.879272 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/1a/1aa359a0d1aa53fa89e8673ca45e4b919911cd21a280220a0199b98196255c57.png",
"timestep": 49,
"previous_timestep": 46,
"previous_source": "sampled_sequence",
"from_zer... |
images/eb/eb1a23713a375730c76bc6365e2337a21d8ea664ca4686f6bd4cf70231ccf7d3.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 73 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_53.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 72.664543 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/eb/eb1a23713a375730c76bc6365e2337a21d8ea664ca4686f6bd4cf70231ccf7d3.png",
"timestep": 53,
"previous_timestep": 49,
"previous_source": "sampled_sequence",
"from_zer... |
images/fd/fd6cfff1839b64e8c40fbd40711b738ca13d1fc245888f3ca985dc16b25a221d.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 73 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object and the ob... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_53_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object and the ob... | libero_spatial_rlinf_sole_cot_v6 | 72.664543 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/fd/fd6cfff1839b64e8c40fbd40711b738ca13d1fc245888f3ca985dc16b25a221d.png",
"timestep": 53,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/b7/b7460426e35c79158d9df524f9af29e451d2a6a8129c44aea2d22aa43a0eb0a0.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 74 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_57.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 73.709038 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/b7/b7460426e35c79158d9df524f9af29e451d2a6a8129c44aea2d22aa43a0eb0a0.png",
"timestep": 57,
"previous_timestep": 53,
"previous_source": "sampled_sequence",
"from_zer... |
images/ce/ce73535513386e8304a271ebdb598b97aeb01ecd3d3de76867b586707f56639f.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 74 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object and the ob... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_57_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot makes first useful contact with the object. The transition also shows the gripper establishes contact with the object and the ob... | libero_spatial_rlinf_sole_cot_v6 | 73.709038 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/ce/ce73535513386e8304a271ebdb598b97aeb01ecd3d3de76867b586707f56639f.png",
"timestep": 57,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/62/6207e0fa2a9bdd196c3170fd1f86f237cae4479018746f6b50c1a32529a5a859.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 74 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_61.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 74.380447 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/62/6207e0fa2a9bdd196c3170fd1f86f237cae4479018746f6b50c1a32529a5a859.png",
"timestep": 61,
"previous_timestep": 57,
"previous_source": "sampled_sequence",
"from_zer... |
images/bb/bb42dd238ff1218b9dc1d79b106aee5c5d043ea3581c53128a94844b2869f0a5.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 74 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_64.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps working to secure the object before lifting. The transition also shows the gripper maintains contact with the obj... | libero_spatial_rlinf_sole_cot_v6 | 74.498596 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/bb/bb42dd238ff1218b9dc1d79b106aee5c5d043ea3581c53128a94844b2869f0a5.png",
"timestep": 64,
"previous_timestep": 61,
"previous_source": "sampled_sequence",
"from_zer... |
images/69/695b77db5bf3724e80e93d5d743f746fa18885e713f927111a2912727d10a48f.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 75 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the objec... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_66.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the objec... | libero_spatial_rlinf_sole_cot_v6 | 74.641594 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/69/695b77db5bf3724e80e93d5d743f746fa18885e713f927111a2912727d10a48f.png",
"timestep": 66,
"previous_timestep": 64,
"previous_source": "sampled_sequence",
"from_zer... |
images/8a/8aa1c842c15c8a0a42329dfe8c6e38ff64071c838ef7fa2f74db485d9b523dfc.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 75 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_68.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | libero_spatial_rlinf_sole_cot_v6 | 75.46048 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/8a/8aa1c842c15c8a0a42329dfe8c6e38ff64071c838ef7fa2f74db485d9b523dfc.png",
"timestep": 68,
"previous_timestep": 66,
"previous_source": "sampled_sequence",
"from_zer... |
images/c3/c3f8131ec1f894869343a0e2f0b482286c74b8c44c6c9fbae4ff1b218d1943a3.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 75 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_68_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | libero_spatial_rlinf_sole_cot_v6 | 75.46048 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/c3/c3f8131ec1f894869343a0e2f0b482286c74b8c44c6c9fbae4ff1b218d1943a3.png",
"timestep": 68,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/c6/c64228387bc941b99cb6030f4deb339e7ef7e81f0cea55c7f53336ec4a0794e4.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 76 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_72.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | libero_spatial_rlinf_sole_cot_v6 | 76.049332 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/c6/c64228387bc941b99cb6030f4deb339e7ef7e81f0cea55c7f53336ec4a0794e4.png",
"timestep": 72,
"previous_timestep": 68,
"previous_source": "sampled_sequence",
"from_zer... |
images/c9/c96ec1dfcd01440fa60d9d8af7adaff8a1e690d0e62b93d7ec1271d717090bad.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 76 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_76.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot keeps hold of the object while repositioning it. The transition also shows the gripper maintains contact with the object, the... | libero_spatial_rlinf_sole_cot_v6 | 76.049332 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/c9/c96ec1dfcd01440fa60d9d8af7adaff8a1e690d0e62b93d7ec1271d717090bad.png",
"timestep": 76,
"previous_timestep": 72,
"previous_source": "sampled_sequence",
"from_zer... |
images/b6/b63b2a4b8f4ac5f935730f975a6abff94c4b66fb6c3060e62a5dca8bc9c3a188.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 78 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object ... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_80.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object ... | libero_spatial_rlinf_sole_cot_v6 | 77.933586 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/b6/b63b2a4b8f4ac5f935730f975a6abff94c4b66fb6c3060e62a5dca8bc9c3a188.png",
"timestep": 80,
"previous_timestep": 76,
"previous_source": "sampled_sequence",
"from_zer... |
images/22/224d504cd35694749b5089964de7463744d644bddcfdd7acdea4664a25ba7d2d.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 78 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_80_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | libero_spatial_rlinf_sole_cot_v6 | 77.933586 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/22/224d504cd35694749b5089964de7463744d644bddcfdd7acdea4664a25ba7d2d.png",
"timestep": 80,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/d0/d0defdfd1762541889fed7fffa140ad0f1aaed25081ba85ae72fe7559be2a62e.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 83 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object ... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_83.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object ... | libero_spatial_rlinf_sole_cot_v6 | 82.640572 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/d0/d0defdfd1762541889fed7fffa140ad0f1aaed25081ba85ae72fe7559be2a62e.png",
"timestep": 83,
"previous_timestep": 80,
"previous_source": "sampled_sequence",
"from_zer... |
images/d2/d2e765a395a2da72e606e624aedf77be64ce15c454bfd536522459e99fba9464.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 90 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and near the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object remain... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_87.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and near the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper maintains contact with the object, the object remain... | libero_spatial_rlinf_sole_cot_v6 | 89.81147 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/d2/d2e765a395a2da72e606e624aedf77be64ce15c454bfd536522459e99fba9464.png",
"timestep": 87,
"previous_timestep": 83,
"previous_source": "sampled_sequence",
"from_zer... |
images/e7/e7460bdcc1be1018b06f40a7623bed7f484bff8fc9cbb3d675b822dbc2a75bc9.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the current timestep are shown to the righ... | 90 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_87_from_zero.png | 0 | <think>At the first timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot carries the object closer to the target. The transition also shows the gripper establishes contact with the object, the object i... | libero_spatial_rlinf_sole_cot_v6 | 89.81147 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/e7/e7460bdcc1be1018b06f40a7623bed7f484bff8fc9cbb3d675b822dbc2a75bc9.png",
"timestep": 87,
"previous_timestep": 0,
"previous_source": "from_zero",
"from_zero": true... |
images/8e/8e513d0a39eb2722373d906b71a612e428cdecaa6e42fd7dcdd76b50380abbfb.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 100 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and near the intended target. At the current timestep, the object moves closer to the target. The transition also shows the gripper maintains contact with the object, the object was lowered, and t... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000000_rank0000_env0000_id000000_success_timestep_91.png | 0 | <think>At the previous timestep, the gripper was in contact with the object. The object was already raised from the surface and near the intended target. At the current timestep, the object moves closer to the target. The transition also shows the gripper maintains contact with the object, the object was lowered, and t... | libero_spatial_rlinf_sole_cot_v6 | 100 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000000_rank0000_env0000_id000000_success",
"image_path": "images/8e/8e513d0a39eb2722373d906b71a612e428cdecaa6e42fd7dcdd76b50380abbfb.png",
"timestep": 91,
"previous_timestep": 87,
"previous_source": "sampled_sequence",
"from_zer... |
images/74/74f8d2070491c0cc5012e4b55622aa1f311884fe820144d5f9bbe0572741c4f1.png | Here is an image containing multiple camera views of a robot attempting to complete a task. The views on the top are from an external camera. The views on the bottom are from the robot's wrist camera. The views from the very first timestep are shown to the left. The views from the previous timestep are shown in the mid... | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | [{"role": "system", "content": "You are an expert roboticist with the goal of predicting task progress percentages given frames from a video of a robot attempting to complete a task. You first think, in the form of an internal monologue, before providing your final answer. Your reasoning process MUST BE enclosed within... | episode_000001_rank0000_env0004_id000000_success_timestep_0.png | 0 | <think>At the previous timestep, the gripper was away from the object. The object was still on or near its starting surface and not yet at the intended target. At the current timestep, the robot approaches the object but has not started manipulating it. The transition also shows the gripper still has not made contact w... | libero_spatial_rlinf_sole_cot_v6 | 0 | {
"dataset_version": "v3_beta075_direction_from_zero",
"episode_id": "episode_000001_rank0000_env0004_id000000_success",
"image_path": "images/74/74f8d2070491c0cc5012e4b55622aa1f311884fe820144d5f9bbe0572741c4f1.png",
"timestep": 0,
"previous_timestep": 0,
"previous_source": "sampled_sequence",
"from_zero"... |
End of preview. Expand in Data Studio
sole-r1-v3-data
LIBERO Spatial progress reasoning 数据。非官方 SOLE-R1 数据发布。原始 CoT、进度标签和图像内容保持不变；只整理路径和发布格式。
划分与样本数：{"train": 14046}。图片位于 images/，按内容去重。
请下载完整数据仓库（包括 images），不要仅下载 JSONL。进入下载目录后运行：
python resolve_paths.py --split train --output train.local.jsonl
此命令生成可供原训练loader读取的绝对图像路径；不改变原JSONL。其他划分替换 split 参数即可。训练实现将在配套代码仓库提供，本数据包本身不是完整训练程序。
v3原数据未在此重新划分；train代表原训练数据。image为已有多视角/多时刻拼图，problem与completion为原训练文本。
来源：LIBERO仿真与本项目构造的进度/CoT；请保留LIBERO和Qwen等上游项目署名。未在此擅自替上游资源声明新的许可。
LIBERO: https://github.com/Lifelong-Robot-Learning/LIBERO 。仿真资产遵循各自上游许可；LIBERO代码的MIT许可附于LIBERO_LICENSE.txt。
- Downloads last month
- 108