Retire the superseded 32-turn comparison

README.md

@@ -15,16 +15,11 @@ authors:
 - name: Tamas Foldi
   affiliation: HCLTech
 configs:
-- config_name: fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020
+- config_name: empty
   default: true
   data_files:
   - split: test
-    path: runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/scores.csv
-- config_name: fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020
-  default: false
-  data_files:
-  - split: test
-    path: runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/scores.csv
+    path: scores.csv
 ---
 # SimReadyBench
 

leaderboard.json

@@ -1,4 +1 @@
-{
-  "schema": "simreadybench-leaderboard/v1",
-  "entries": []
-}
+{"schema":"simreadybench-leaderboard/v1","entries":[]}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/assets/accessory_eyeglasses_a01/reconstruction-1/candidate/manifest.json

@@ -1,14 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "402eff31f89f535216709a98ae2464d016eb5d0791da26bd149413695f567f5f",
-      "size_bytes": 492
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 492,
-  "snapshot_sha256": "6184ea997ffc576379db22ac0b7959bef4a2fb0aa3f3bb3ba2f7d4fcd0e4d657"
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/assets/accessory_eyeglasses_a01/reconstruction-2/candidate/manifest.json

@@ -1,14 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "402eff31f89f535216709a98ae2464d016eb5d0791da26bd149413695f567f5f",
-      "size_bytes": 492
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 492,
-  "snapshot_sha256": "6184ea997ffc576379db22ac0b7959bef4a2fb0aa3f3bb3ba2f7d4fcd0e4d657"
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/accessory_eyeglasses_a01/reconstruction-1/harbor-trajectory-4.json

@@ -1,326 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a2d-9e99-768b-8189-f4283b3d4320",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:38:23.782Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.14458004385068107, 0.1743206977770294, 0.05692390352487564]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "599eb781",
-        "pi_parent_id": "aca0785b"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:38:33.184Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-80fc5d27a06254a1",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-835934a210f83062",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-853705f0384cfc3d",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-80fc5d27a06254a1",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-835934a210f83062",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-853705f0384cfc3d",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 1921,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 1977
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "3597c7f6",
-        "pi_parent_id": "599eb781",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "26ee6eea31cee3e8e2586f568e5f273b",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:38:37.822Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-82bb1186cbf63881",
-          "function_name": "bash",
-          "arguments": {
-            "command": "ls /task"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-82bb1186cbf63881",
-            "content": "contract.md\ninstruction.md\nscoring-policy.json\nsimready-bench-native-output-policy.ts\nthumbnail.png\ntool-docs.md",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3849,
-        "completion_tokens": 19,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 3868
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "1f0c9fcb",
-        "pi_parent_id": "8b1b9607",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "9e6efb2edbe967ff04f01ed4f3d2a34a",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:40:52.700Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-807cdb2cfe80e226",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.Usd as Usd\nimport pxr.UsdGeom as UsdGeom\nimport pxr.UsdPhysics as UsdPhysics\nimport pxr.UsdShade as UsdShade\nimport pxr.Gf as Gf\nimport numpy as np\n\ndef create_material(stage, name, color):\n    material_path = f\"/World/Materials/{name}\"\n    material = UsdShade.Material.Define(stage, material_path)\n    pbr_shader = UsdShade.Shader.Define(stage, f\"{material_path}/PBRShader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(0.2))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(0.0))\n    \n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material\n\ndef create_box(stage, path, size, translation, rotation, material):\n    mesh = UsdGeom.Mesh.Define(stage, path)\n    \n    # Create a simple box\n    w, h, d = size\n    vertices = np.array([\n        [-w/2, -h/2, -d/2], [w/2, -h/2, -d/2], [w/2, h/2, -d/2], [-w/2, h/2, -d/2],\n        [-w/2, -h/2, d/2], [w/2, -h/2, d/2], [w/2, h/2, d/2], [-w/2, h/2, d/2]\n    ], dtype=np.float32)\n    \n    face_vertex_counts = [4, 4, 4, 4, 4, 4]\n    face_vertex_indices = [\n        0, 1, 2, 3,\n        4, 5, 6, 7,\n        0, 1, 5, 4,\n        2, 3, 7, 6,\n        0, 3, 7, 4,\n        1, 2, 6, 5\n    ]\n    \n    mesh.CreatePointsAttr(vertices)\n    mesh.CreateFaceVertexCountsAttr(face_vertex_counts)\n    mesh.CreateFaceVertexIndicesAttr(face_vertex_indices)\n    \n    # Transform\n    xform = UsdGeom.Xform.Get(mesh)\n    # Need to apply translation and rotation. \n    # For simplicity, we'll use a parent Xform for the part.\n    return mesh\n\ndef main():\n    stage = Usd.Stage.CreateNew(\"/output/asset.usd\")\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n    \n    # Nominal dimensions: [0.14458, 0.17432, 0.05692]\n    # X: 14.4cm, Y: 17.4cm, Z: 5.7cm\n    # Based on the image, these are the bounding box of the unfolded glasses?\n    # Or maybe the folded ones? Usually, width is the largest dimension.\n    # Wait, usually width (X) is larger than depth (Z).\n    # Let's assume: Width = 0.144, Height = 0.057, Depth = 0.174 (folded)\n    # Let's re-evaluate: \n    # Width: 14.4cm (standard for glasses)\n    # Height: 5.7cm (standard for glasses)\n    # Depth: 17.4cm (this must be the temple length)\n    # So the bounding box provided is for the whole asset.\n    \n    # Define colors\n    dark_grey = (0.1, 0.1, 0.1)\n    lens_color = (0.05, 0.05, 0.1, 0.8) # Semi-transparent\n    \n    mat_frame = create_material(stage, \"FrameMat\", dark_grey)\n    mat_lens = create_material(stage, \"LensMat\", (0.0, 0.0, 0.0)) # Dark lenses\n    \n    # World root\n    root = UsdGeom.Xform.Define(stage, \"/World/Glasses\")\n    \n    # Frame\n    frame_path = \"/World/Glasses/Frame\"\n    frame_xform = UsdGeom.Xform.Define(stage, frame_path)\n    \n    # Simple representation of the front frame\n    # Front part: a rectangle-ish shape with holes for lenses\n    # For simplicity, use a few boxes\n    # Bridge\n    bridge = UsdGeom.Mesh.Define(stage, f\"{frame_path}/Bridge\")\n    # ... mesh creation ...\n    \n    # Lenses (as separate meshes)\n    # Left lens\n    l_lens = UsdGeom.Mesh.Define(stage, f\"{frame_path}/LeftLens\")\n    # Right lens\n    r_lens = UsdGeom.Mesh.Define(stage, f\"{frame_path}/RightLens\")\n    \n    # Temples (hinged)\n    # Left Temple\n    l_temple_path = \"/World/Glasses/LeftTemple\"\n    l_temple_xform = UsdGeom.Xform.Define(stage, l_temple_path)\n    l_temple_mesh = UsdGeom.Mesh.Define(stage, f\"{l_temple_path}/Mesh\")\n    \n    # Right Temple\n    r_temple_path = \"/World/Glasses/RightTemple\"\n    r_temple_xform = UsdGeom.Xform.Define(stage, r_temple_path)\n    r_temple_mesh = UsdGeom.Mesh.Define(stage, f\"{r_temple_path}/Mesh\")\n    \n    # Add Physics\n    UsdPhysics.RigidBodyAPI.Apply(frame_xform)\n    UsdPhysics.RigidBodyAPI.Apply(l_temple_xform)\n    UsdPhysics.RigidBodyAPI.Apply(r_temple_xform)\n    \n    # Joints\n    # Left Hinge\n    l_joint = UsdPhysics.RevoluteJoint.Define(stage, \"/World/Glasses/LeftHinge\")\n    l_joint.CreateBody0Rel().SetTarget(frame_path)\n    l_joint.CreateBody1Rel().SetTarget(l_temple_path)\n    l_joint.CreateAxisAttr(Gf.Vec3f(0, 0, 1)) # Z-axis hinge\n    l_joint.CreateLocalPos0(Gf.Vec3f(-0.07, 0, 0))\n    l_joint.CreateLocalPos1(Gf.Vec3f(0, 0, 0))\n    \n    # Right Hinge\n    r_joint = UsdPhysics.RevoluteJoint.Define(stage, \"/World/Glasses/RightHinge\")\n    r_joint.CreateBody0Rel().SetTarget(frame_path)\n    r_joint.CreateBody1Rel().SetTarget(r_temple_path)\n    r_joint.CreateAxisAttr(Gf.Vec3f(0, 0, 1)) # Z-axis hinge\n    r_joint.CreateLocalPos0(Gf.Vec3f(0.07, 0, 0))\n    r_joint.CreateLocalPos1(Gf.Vec3f(0, 0, 0))\n    \n    stage.GetRootLayer().Save()\n\nif __name__ == \"__main__\":\n    main()\n",
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-807cdb2cfe80e226",
-            "content": "Successfully wrote to /tmp/author_asset.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3911,
-        "completion_tokens": 1880,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5791
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "659ad733",
-        "pi_parent_id": "93a84704",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "c49884dfe188d9a065486ebb1c7d266e",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:41:34.076Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-8c4dc36da805cc04",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-8c4dc36da805cc04",
-            "content": "Traceback (most recent call last):\n  File \"/tmp/author_asset.py\", line 129, in <module>\n    main()\n  File \"/tmp/author_asset.py\", line 52, in main\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n                                  ^^^^^^^^^^^^^^\nAttributeError: module 'pxr.UsdGeom' has no attribute 'UpAxis'\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5808,
-        "completion_tokens": 26,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5834
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "f49394f3",
-        "pi_parent_id": "e04cacfe",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "39cddc11c3c0f545f5831513b41bd632",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:41:57.676Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-947ce528c15c5f9d",
-          "function_name": "edit",
-          "arguments": {
-            "edits": [
-              {
-                "newText": "    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)",
-                "oldText": "    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)"
-              }
-            ],
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-947ce528c15c5f9d",
-            "content": "No changes made to /tmp/author_asset.py. The replacement produced identical content. This might indicate an issue with special characters or the text not existing as expected.",
-            "extra": {
-              "tool_name": "edit",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5952,
-        "completion_tokens": 86,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 6038
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "628910c7",
-        "pi_parent_id": "c37ec9a9",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "cbc650d6c37aaf5dc94964d611139d8e",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 21441,
-    "total_completion_tokens": 2067,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 8,
-      "pi_total_tokens": 23508
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/accessory_eyeglasses_a01/reconstruction-2/harbor-trajectory-4.json

@@ -1,326 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a31-4e59-7778-bd48-2ac88f3ca9cb",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:42:25.211Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.14458004385068107, 0.1743206977770294, 0.05692390352487564]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "f46810fb",
-        "pi_parent_id": "adcd1716"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:42:32.994Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a1cbcdd58b73ac83",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-b28a0b6b00d1a12b",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-ba8452da9ee7b38a",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a1cbcdd58b73ac83",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-b28a0b6b00d1a12b",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-ba8452da9ee7b38a",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 1921,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 1977
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "c92ccfb5",
-        "pi_parent_id": "f46810fb",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "65934b32769709e6e2376871aadd6de8",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:42:39.279Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a5b1c24dd7504e3c",
-          "function_name": "bash",
-          "arguments": {
-            "command": "ls /task/thumbnail.png"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a5b1c24dd7504e3c",
-            "content": "/task/thumbnail.png",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3849,
-        "completion_tokens": 23,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 3872
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "18df83d7",
-        "pi_parent_id": "0de439be",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "8575b7c6cc525c713e73288f4d00d85d",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:43:29.460Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a2daaa8870e91e31",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.Usd as Usd\nimport pxr.UsdGeom as UsdGeom\nimport pxr.UsdPhysics as UsdPhysics\nimport pxr.Gf as Gf\n\ndef create_asset():\n    stage = Usd.Stage.CreateNew('/output/asset.usd')\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n    \n    # Dimensions: [0.14458, 0.17432, 0.05692] (X, Y, Z)\n    # Wait, the dimensions provided are [X, Y, Z] = [width, height, depth]?\n    # Usually for glasses: Width (X), Depth (Y), Height (Z).\n    # Reference: Nom dimensions XYZ: [0.14458004385068107, 0.1743206977770294, 0.05692390352487564]\n    # Looking at the image, the object is glasses.\n    # Width is ~14.5cm, Temple length is ~17.4cm, Frame height is ~5.7cm.\n    # So X=width, Y=temple length (depth), Z=height. \n    # Canonical meters.\n    \n    # Root Xform\n    root = UsdGeom.Xform.Define(stage, '/asset')\n    \n    # Rigid bodies\n    # 1. Front frame (fixed)\n    frame = UsdGeom.Mesh.Define(stage, '/asset/frame')\n    UsdPhysics.RigidBodyAPI.Apply(frame.GetPrim())\n    \n    # 2. Left Temple (movable)\n    left_temple = UsdGeom.Mesh.Define(stage, '/asset/left_temple')\n    UsdPhysics.RigidBodyAPI.Apply(left_temple.GetPrim())\n    \n    # 3. Right Temple (movable)\n    right_temple = UsdGeom.Mesh.Define(stage, '/asset/right_temple')\n    UsdPhysics.RigidBodyAPI.Apply(right_temple.GetPrim())\n    \n    # Joint positions\n    # Width = 0.14458. Half width = 0.07229\n    # Height = 0.05692. Half height = 0.02846\n    \n    # Left hinge\n    # Pos: (-0.07229, 0, 0.02846) -> Wait, Z is height.\n    # Let's center the glasses.\n    # Front frame: center at (0, 0, 0). \n    # Height is along Z? No, usually Y is up. \n    # Let's use Y as up, X as width, Z as depth (temple direction).\n    # Dimensions provided: [X, Y, Z] = [0.14458, 0.17432, 0.05692]\n    # This suggests: X=width, Y=temple length, Z=height. \n    # This is non-standard. Let's re-evaluate.\n    # Usually, in USD, Y is up.\n    # I'll map:\n    # Real Width -> X\n    # Real Temple Length -> Z\n    # Real Height -> Y\n    # So: X = 0.14458, Y = 0.05692, Z = 0.17432.\n    \n    # Hinge positions (local to root)\n    # L hinge: (-0.07229, 0.02, 0)\n    # R hinge: (0.07229, 0.02, 0)\n    \n    # Left Joint\n    left_joint = UsdPhysics.RevoluteJoint.Define(stage, '/asset/left_joint')\n    left_joint.CreateBody0Rel().SetTargets([frame.GetPath()])\n    left_joint.CreateBody1Rel().SetTargets([left_temple.GetPath()])\n    left_joint.CreateAxisAttr().Set(Gf.Vec3f(0, 1, 0)) # Rotate around Y (up)\n    left_joint.CreateLocalPos0Attr().Set(Gf.Vec3f(-0.07229, 0.02, 0))\n    left_joint.CreateLocalPos1Attr().Set(Gf.Vec3f(0, 0, 0))\n    \n    # Right Joint\n    right_joint = UsdPhysics.RevoluteJoint.Define(stage, '/asset/right_joint')\n    right_joint.CreateBody0Rel().SetTargets([frame.GetPath()])\n    right_joint.CreateBody1Rel().SetTargets([right_temple.GetPath()])\n    right_joint.CreateAxisAttr().Set(Gf.Vec3f(0, 1, 0))\n    right_joint.CreateLocalPos0Attr().Set(Gf.Vec3f(0.07229, 0.02, 0))\n    right_joint.CreateLocalPos1Attr().Set(Gf.Vec3f(0, 0, 0))\n    \n    # Basic Geometry (Boxes for now, as placeholders)\n    # Mapping: \n    # Frame: X=0.14458, Y=0.05692, Z=0.01 (thin)\n    # Temple: X=0.01, Y=0.01, Z=0.17432\n    \n    def make_box(mesh_prim, size):\n        # Very basic box generation\n        points = [\n            Gf.Vec3f(-size[0]/2, -size[1]/2, -size[2]/2),\n            Gf.Vec3f(size[0]/2, -size[1]/2, -size[2]/2),\n            Gf.Vec3f(size[0]/2, size[1]/2, -size[2]/2),\n            Gf.Vec3f(-size[0]/2, size[1]/2, -size[2]/2),\n            Gf.Vec3f(-size[0]/2, -size[1]/2, size[2]/2),\n            Gf.Vec3f(size[0]/2, -size[1]/2, size[2]/2),\n            Gf.Vec3f(size[0]/2, size[1]/2, size[2]/2),\n            Gf.Vec3f(-size[0]/2, size[1]/2, size[2]/2),\n        ]\n        mesh_prim.CreatePointsAttr().Set(points)\n        face_counts = [4] * 6\n        face_indices = [\n            [0, 1, 2, 3], [4, 5, 6, 7], [0, 1, 5, 4],\n            [2, 3, 7, 6], [0, 3, 7, 4], [1, 2, 6, 5]\n        ]\n        # Flat list for indices\n        flat_indices = [idx for face in face_indices for idx in face]\n        mesh_prim.CreateFaceVertexCountsAttr().Set(face_counts)\n        mesh_prim.CreateFaceVertexIndicesAttr().Set(flat_indices)\n\n    make_box(frame, [0.14458, 0.05692, 0.01])\n    make_box(left_temple, [0.005, 0.005, 0.17432])\n    make_box(right_temple, [0.005, 0.005, 0.17432])\n    \n    # Position temples relative to hinges\n    left_temple.AddTranslateOp().Set(Gf.Vec3f(-0.07229, 0.02, 0))\n    right_temple.AddTranslateOp().Set(Gf.Vec3f(0.07229, 0.02, 0))\n    \n    stage.GetRootLayer().Save()\n\ncreate_asset()\n",
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a2daaa8870e91e31",
-            "content": "Successfully wrote to /tmp/author_asset.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 4020,
-        "completion_tokens": 1992,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6012
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "d6d05e22",
-        "pi_parent_id": "f26655b9",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "b6de84af1185c11aee5d2ced5c0f1746",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:43:31.688Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-bb40859836cefda1",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-bb40859836cefda1",
-            "content": "Traceback (most recent call last):\n  File \"/tmp/author_asset.py\", line 113, in <module>\n    create_asset()\n  File \"/tmp/author_asset.py\", line 8, in create_asset\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n                                  ^^^^^^^^^^^^^^\nAttributeError: module 'pxr.UsdGeom' has no attribute 'UpAxis'\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6029,
-        "completion_tokens": 22,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6051
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "c675f2b1",
-        "pi_parent_id": "8606b90a",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "22c5c66ca19c1eebe5ae8cd858270b25",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:43:35.344Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-92b5803d5e14a673",
-          "function_name": "edit",
-          "arguments": {
-            "edits": [
-              {
-                "newText": "    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)",
-                "oldText": "    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)"
-              }
-            ],
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-92b5803d5e14a673",
-            "content": "No changes made to /tmp/author_asset.py. The replacement produced identical content. This might indicate an issue with special characters or the text not existing as expected.",
-            "extra": {
-              "tool_name": "edit",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6176,
-        "completion_tokens": 86,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6262
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "66b59214",
-        "pi_parent_id": "89c3cfb5",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "8da04b5043dd6d20ce11c260690f9114",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 21995,
-    "total_completion_tokens": 2179,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 2,
-      "pi_total_tokens": 24174
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/box_cardboard_open_a01/reconstruction-1/harbor-trajectory-4.json

@@ -1,320 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a2d-64c4-77b5-9da8-90b0df736747",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:38:08.711Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.6357795000076294, 0.5118491649627686, 0.30237919092178345]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "9490f41b",
-        "pi_parent_id": "2c7a6ee6"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:38:19.635Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-ae554b9d2ab79489",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-a3573dc8b94b63f0",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-82f393b077ef4511",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-ae554b9d2ab79489",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-a3573dc8b94b63f0",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-82f393b077ef4511",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 1920,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 1976
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "a834ce22",
-        "pi_parent_id": "9490f41b",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "88d615ee285955b111044b6a7ae0ff20",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:38:26.043Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-bc45a81e1f26642a",
-          "function_name": "bash",
-          "arguments": {
-            "command": "ls /task/thumbnail.png"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-bc45a81e1f26642a",
-            "content": "/task/thumbnail.png",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3848,
-        "completion_tokens": 23,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 3871
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "f0b93a84",
-        "pi_parent_id": "eebdc8d9",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "e4ce46097e880eb28ee1e66c9c155c5e",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:40:47.487Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a57265e885482200",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.usd as usd\nfrom pxr import UsdGeom, UsdShade, Sdf, Gf, UsdPhysics\n\ndef create_asset():\n    stage = usd.Stage.CreateNew('/output/asset.usd')\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n\n    # Dimensions: [0.6357795000076294, 0.5118491649627686, 0.30237919092178345]\n    # X: length, Y: height, Z: width (based on thumbnail image orientation)\n    # Let's re-examine: the image shows a rectangular box.\n    # Nominal dimensions given as [0.6358, 0.5118, 0.3024]\n    # In the image, the longest dimension is the length (along the box axis).\n    # Let's assume: \n    # Length (X) = 0.6358\n    # Width (Z) = 0.5118\n    # Height (Y) = 0.3024\n    # Wait, looking at the thumbnail, the box is open. \n    # Dimensions are usually given as [L, W, H] or [X, Y, Z].\n    # Given the thumbnail, the box's long axis is along what I'd call length.\n    # X = 0.6357795, Y = 0.3023792 (height), Z = 0.5118492 (width).\n    \n    L = 0.6357795\n    W = 0.5118492\n    H = 0.3023792\n    t = 0.002 # thickness\n\n    # The box is a single piece of cardboard folded.\n    # It consists of a bottom and four walls. \n    # The side walls are folded up.\n    \n    # Root Xform\n    root = UsdGeom.Xform.Define(stage, '/root')\n    \n    # Cardboard Material\n    # Outer: white, Inner: brown/corrugated\n    # Since we can't easily do double-sided materials on a single mesh without complex shaders,\n    # we'll use two shells or separate faces.\n    \n    def create_material(name, color):\n        mat = UsdShade.Material.Define(stage, f'/materials/{name}')\n        pbr = UsdShade.Shader.Define(stage, f'/materials/{name}/PBR')\n        pbr.CreateIdAttr('UsdPreviewSurface')\n        pbr.CreateInput('diffuseColor', Sdf.ValueTypeNames.Color3f).Set(color)\n        pbr.CreateInput('roughness', Sdf.ValueTypeNames.Float).Set(0.8)\n        mat.CreateSurfaceOutput().ConnectToSource(pbr.ConnectableAPI(), 'surface')\n        return mat\n\n    white_mat = create_material('white', Gf.Vec3f(0.9, 0.9, 0.9))\n    brown_mat = create_material('brown', Gf.Vec3f(0.7, 0.5, 0.3))\n\n    # Box Geometry\n    # We can represent the box as a set of thin boxes or a custom mesh.\n    # Bottom:\n    bottom = UsdGeom.Mesh.Define(stage, '/root/bottom')\n    # Bottom rectangle: from -L/2 to L/2 and -W/2 to W/2\n    # Points for a thin box (6 faces, 12 vertices)\n    # Bottom face (Z-up for the mesh locally, but we'll orient it)\n    # Actually, let's just use a few Mesh prims for walls and bottom.\n    \n    def create_box_mesh(path, size, pos, mat_outer, mat_inner):\n        mesh = UsdGeom.Mesh.Define(stage, path)\n        sx, sy, sz = size\n        px, py, pz = pos\n        \n        # 8 vertices for a box\n        verts = [\n            Gf.Vec3f(px - sx/2, py - sy/2, pz - sz/2),\n            Gf.Vec3f(px + sx/2, py - sy/2, pz - sz/2),\n            Gf.Vec3f(px + sx/2, py + sy/2, pz - sz/2),\n            Gf.Vec3f(px - sx/2, py + sy/2, pz - sz/2),\n            Gf.Vec3f(px - sx/2, py - sy/2, pz + sz/2),\n            Gf.Vec3f(px + sx/2, py - sy/2, pz + sz/2),\n            Gf.Vec3f(px + sx/2, py + sy/2, pz + sz/2),\n            Gf.Vec3f(px - sx/2, py + sy/2, pz + sz/2),\n        ]\n        mesh.CreatePointsAttr().Set(verts)\n        \n        # Faces\n        faces = [\n            [0, 1, 2, 3], # Bottom\n            [4, 5, 6, 7], # Top\n            [0, 1, 5, 4], # Front\n            [2, 3, 7, 6], # Back\n            [1, 2, 6, 5], # Right\n            [0, 3, 7, 4], # Left\n        ]\n        mesh.CreateFaceVertexCountsAttr().Set([4]*6)\n        mesh.CreateFaceVertexIndicesAttr().Set([idx for face in faces for idx in face])\n        \n        # Material binding (simplification: one material per mesh)\n        UsdShade.MaterialBindingAPI.Apply(mesh).Bind(mat_outer)\n        \n        # Physics\n        UsdPhysics.RigidBodyAPI.Apply(mesh)\n        UsdPhysics.CollisionAPI.Apply(mesh)\n        \n        return mesh\n\n    # Bottom Plate\n    create_box_mesh('/root/bottom', (L, t, W), (0, 0, 0), brown_mat, brown_mat)\n    \n    # Long Walls (X-axis)\n    create_box_mesh('/root/wall_left', (L, H, t), (0, H/2, -W/2), white_mat, brown_mat)\n    create_box_mesh('/root/wall_right', (L, H, t), (0, H/2, W/2), white_mat, brown_mat)\n    \n    # Short Walls (Z-axis)\n    create_box_mesh('/root/wall_front', (t, H, W), (L/2, H/2, 0), white_mat, brown_mat)\n    create_box_mesh('/root/wall_back', (t, H, W), (-L/2, H/2, 0), white_mat, brown_mat)\n\n    # To make it one rigid body, we can use fixed joints or just group them.\n    # But since it's a cardboard box, it's one piece.\n    # We should probably use a single mesh for the whole thing if possible, or just \n    # weld them with FixedJoints.\n    \n    for part in ['bottom', 'wall_left', 'wall_right', 'wall_front', 'wall_back']:\n        path = f'/root/{part}'\n        # Only the bottom is the root of the physics assembly\n        if part != 'bottom':\n            joint = UsdPhysics.FixedJoint.Define(stage, f'/root/joint_{part}')\n            joint.CreateBody0Rel().SetTargets([Sdf.Path('/root/bottom')])\n            joint.CreateBody1Rel().SetTargets([Sdf.Path(path)])\n            # Joint position is local to the body? \n            # Let's just set local poses to match the mesh positions.\n            # Actually, for FixedJoint, if the prims are already positioned in world space,\n            # the joint should be at the connection point.\n    \n    # The prompt asks for \"rigid parts, joints and their motion\".\n    # A cardboard box like this is usually rigid.\n    \n    stage.GetRootLayer().Save()\n\ncreate_asset()\n",
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a57265e885482200",
-            "content": "Successfully wrote to /tmp/author_asset.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3884,
-        "completion_tokens": 2015,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5899
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "8f3f021c",
-        "pi_parent_id": "d3979778",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "3f8694fe2c865a6e06a1ce2c5d7bcbd4",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:40:58.547Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-85a4b046d6f3b295",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-85a4b046d6f3b295",
-            "content": "Traceback (most recent call last):\n  File \"/tmp/author_asset.py\", line 1, in <module>\n    import pxr.usd as usd\nModuleNotFoundError: No module named 'pxr.usd'\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5916,
-        "completion_tokens": 22,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 5938
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "8137bf71",
-        "pi_parent_id": "49a59920",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "82eee52c051b4e53bad9a5bacae8666d",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:41:16.174Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a118e62791b4ceae",
-          "function_name": "bash",
-          "arguments": {
-            "command": "pip list | grep pxr"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a118e62791b4ceae",
-            "content": "/bin/bash: line 1: pip: command not found\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6005,
-        "completion_tokens": 22,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 6027
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "054a11a6",
-        "pi_parent_id": "5c1cc686",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "8a1ba2dd0f88089421fe8054a4ce7450",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 21573,
-    "total_completion_tokens": 2138,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 6,
-      "pi_total_tokens": 23711
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/box_cardboard_open_a01/reconstruction-2/harbor-trajectory-4.json

@@ -1,320 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a30-9720-723d-9040-dda6958b1ccf",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:41:38.298Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.6357795000076294, 0.5118491649627686, 0.30237919092178345]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "a6d8b66a",
-        "pi_parent_id": "a23094e4"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:41:46.242Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-bf5e5891125c2097",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-946e4c9bcada6eb4",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-b39a1417af9f9428",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-bf5e5891125c2097",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-946e4c9bcada6eb4",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-b39a1417af9f9428",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 2055,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 2111
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "378a79c3",
-        "pi_parent_id": "a6d8b66a",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "27178edd03ea1882b80f6ce2f60e01e4",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:42:26.779Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-a28cc274407a700f",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.usd as Usd\nimport pxr.usd.geom as UsdGeom\nimport pxr.usd.physics as UsdPhysics\nimport pxr.gfx as Gfx\nimport pxr.Sdf as Sdf\nfrom pxr import UsdGeom, Gf, UsdShade\n\ndef create_box(stage, path, size, color):\n    mesh = UsdGeom.Mesh.Define(stage, path)\n    \n    # For a simple box, we define the points and face vertex counts\n    # size: (w, h, d)\n    w, h, d = size\n    points = [\n        (-w/2, 0, -d/2), (w/2, 0, -d/2), (w/2, 0, d/2), (-w/2, 0, d/2), # bottom\n        (-w/2, h, -d/2), (w/2, h, -d/2), (w/2, h, d/2), (-w/2, h, d/2), # top\n    ]\n    \n    # Since this is a cardboard box, it's basically an open cube (no top face)\n    # Bottom face: 0, 1, 2, 3\n    # Front face: 0, 1, 5, 4\n    # Back face: 3, 2, 6, 7\n    # Left face: 0, 3, 7, 4\n    # Right face: 1, 2, 6, 5\n    \n    face_counts = [4, 4, 4, 4, 4]\n    face_indices = [\n        (0, 1, 2, 3), # Bottom\n        (0, 1, 5, 4), # Front\n        (3, 2, 6, 7), # Back\n        (0, 3, 7, 4), # Left\n        (1, 2, 6, 5), # Right\n    ]\n    \n    # Flatten face_indices\n    flat_indices = []\n    for face in face_indices:\n        flat_indices.extend(face)\n        \n    mesh.GetPointsAttr().Set(points)\n    mesh.GetFaceVertexCountsAttr().Set(face_counts)\n    mesh.GetFaceVertexIndicesAttr().Set(flat_indices)\n    \n    # Material\n    material = UsdShade.Material.Define(stage, f\"{path}_mat\")\n    pbr = UsdShade.PbrSurface.Define(material)\n    pbr.CreateInput(\"diffuseColor\", Sdf.ValueTypeNames.Color3f).Set(color)\n    \n    # Apply material\n    geom_bind = UsdGeom.Mesh.GetMaterialBindingAPI(mesh)\n    geom_bind.Bind(material)\n    \n    return mesh\n\ndef main():\n    stage = Usd.Stage.CreateInMemory()\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n    \n    # Nominal dimensions: [0.6357795000076294, 0.5118491649627686, 0.30237919092178345]\n    # Based on the image, it looks like a rectangular open box.\n    # Looking at the dimensions: \n    # Width (X) = 0.6357795\n    # Height (Y) = 0.3023792 (Looking at typical box proportions and the image, the smallest dimension is probably height)\n    # Depth (Z) = 0.5118492\n    # Wait, the provided dimensions are [X, Y, Z]. Let's assume they are [width, depth, height]?\n    # Or maybe [Max X, Max Y, Max Z]?\n    # Image shows it's wider than it is deep, and taller than its wall thickness.\n    # Let's check: 0.635 (longest), 0.511 (mid), 0.302 (shortest).\n    # In the thumbnail, the box is open.\n    # Longest dimension is the length of the box.\n    # Medium is the width.\n    # Shortest is the height.\n    # So length = 0.635, width = 0.511, height = 0.302.\n    # Looking at the orientation in the image: it's angled.\n    \n    # Let's assign:\n    # x = 0.6357795 (length)\n    # z = 0.5118492 (width)\n    # y = 0.3023792 (height)\n    \n    # Let's use a single rigid body since it's just a cardboard box.\n    box_path = \"/asset/box\"\n    box_mesh = create_box(stage, box_path, (0.6357795, 0.3023792, 0.5118492), Gf.Vec3f(0.9, 0.9, 0.8))\n    \n    # Physics\n    physics_api = UsdPhysics.RigidBodyAPI.Apply(box_mesh)\n    collision_api = UsdPhysics.CollisionAPI.Apply(box_mesh)\n    \n    # Center of mass and inertia\n    # For a simple box, centroid is the average of points.\n    # Points were defined from -w/2 to w/2 and 0 to h.\n    # So center is (0, h/2, 0)\n    physics_api.CreateCenterOfMassAttr().Set(Gf.Vec3f(0, 0.3023792/2, 0))\n    \n    # To avoid the box falling through the floor in simulation, \n    # we usually don't define the floor here, but make sure the asset is self-contained.\n    \n    # Save\n    stage.GetRootLayer().Save(\"/output/asset.usd\")\n\nif __name__ == \"__main__\":\n    main()\n",
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-a28cc274407a700f",
-            "content": "Successfully wrote to /tmp/author_asset.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3983,
-        "completion_tokens": 1485,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 5468
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "89749300",
-        "pi_parent_id": "f3354697",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "7a36ca2b56e9704462dcf2544a477cba",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:42:32.154Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-ba73b8f977683187",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-ba73b8f977683187",
-            "content": "Traceback (most recent call last):\n  File \"/tmp/author_asset.py\", line 1, in <module>\n    import pxr.usd as Usd\nModuleNotFoundError: No module named 'pxr.usd'\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5347,
-        "completion_tokens": 26,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5373
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "fdecdd4e",
-        "pi_parent_id": "d9859d8c",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "60f06f528e4fc1998cd07651d78b7e92",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:42:38.088Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-90c21a802ef3e05d",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 -c \"import pxr; print(pxr.__version__)\" || python3 -c \"import usd; print('usd imported')\" || echo \"No USD\""
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-90c21a802ef3e05d",
-            "content": "Traceback (most recent call last):\n  File \"<string>\", line 1, in <module>\nAttributeError: module 'pxr' has no attribute '__version__'\nTraceback (most recent call last):\n  File \"<string>\", line 1, in <module>\nModuleNotFoundError: No module named 'usd'\nNo USD",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5437,
-        "completion_tokens": 53,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5490
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "6917ad80",
-        "pi_parent_id": "bb57ca4a",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "ca97b824c4ee4e304235a84000499622",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:42:42.159Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-bb092c2351320fa6",
-          "function_name": "bash",
-          "arguments": {
-            "command": "pip list | grep -i usd"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-bb092c2351320fa6",
-            "content": "/bin/bash: line 1: pip: command not found\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 5570,
-        "completion_tokens": 23,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 5593
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "0c22da72",
-        "pi_parent_id": "437e526a",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "5f25a4f105ee0c026f3c3acfc2347b9f",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 22392,
-    "total_completion_tokens": 1643,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 6,
-      "pi_total_tokens": 24035
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/chair_office_amf_black_a01/reconstruction-1/harbor-trajectory-4.json

@@ -1,320 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a2e-0e9a-71f0-9439-d36e42aa9c23",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:38:53.167Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.5397186279296875, 0.5722833871841431, 0.8096046261489391]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "4db8d2a7",
-        "pi_parent_id": "af79d356"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:38:59.227Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-948eb6e07c1a5877",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-817ba9c2f166bf16",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-8046ee4fd31928a2",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-948eb6e07c1a5877",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-817ba9c2f166bf16",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-8046ee4fd31928a2",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 1919,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 1975
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "d699346a",
-        "pi_parent_id": "4db8d2a7",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "5b6d9c33e2e143b106168c79b7be3081",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:39:04.292Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-8b732190a3c7b6b9",
-          "function_name": "bash",
-          "arguments": {
-            "command": "ls /task/thumbnail.png"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-8b732190a3c7b6b9",
-            "content": "/task/thumbnail.png",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3847,
-        "completion_tokens": 23,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 3870
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "16f5c36a",
-        "pi_parent_id": "39253b4b",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "ca792a7d67222848ce803bdc4d4c32c6",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:42:02.747Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-b6d54936f514fd2b",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.usd as Usd\nimport pxr.UsdGeom as UsdGeom\nimport pxr.UsdPhysics as UsdPhysics\nimport pxr.UsdShade as UsdShade\nimport pxr.Gf as Gf\nimport numpy as np\n\ndef create_chair():\n    stage = Usd.Stage.CreateNew('/output/asset.usd')\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n\n    # Dimensions\n    # [0.5397186279296875, 0.5722833871841431, 0.8096046261489391]\n    # Width: 0.54, Depth: 0.57, Height: 0.81\n    width = 0.5397186279296875\n    depth = 0.5722833871841431\n    height = 0.8096046261489391\n\n    # Root Xform\n    root = UsdGeom.Xform.Define(stage, '/asset')\n    \n    # Materials\n    # Black plastic/metal for frame, dark grey fabric for seat\n    frame_mat = UsdShade.Material.Define(stage, '/asset/materials/frame_mat')\n    frame_pbr = UsdShade.PbrSurface.Define(stage, '/asset/materials/frame_pbr')\n    frame_pbr.CreateInput('diffuseColor', Gf.Vec3f(0.05, 0.05, 0.05))\n    frame_pbr.CreateInput('roughness', 0.4)\n    frame_pbr.CreateInput('metallic', 0.1)\n    frame_mat.CreateSurfaceOutput().ConnectToSource(frame_pbr.ConnectableAPI(), 'surface')\n\n    seat_mat = UsdShade.Material.Define(stage, '/asset/materials/seat_mat')\n    seat_pbr = UsdShade.PbrSurface.Define(stage, '/asset/materials/seat_pbr')\n    seat_pbr.CreateInput('diffuseColor', Gf.Vec3f(0.1, 0.1, 0.1))\n    seat_pbr.CreateInput('roughness', 0.9)\n    seat_pbr.CreateInput('metallic', 0.0)\n    seat_mat.CreateSurfaceOutput().ConnectToSource(seat_pbr.ConnectableAPI(), 'surface')\n\n    # Rigid Bodies\n    # The chair looks like a single piece for this simple reconstruction, \n    # but following \"Identify rigid parts\", it consists of a seat shell and a leg frame.\n    # In many such chairs, they are bolted together. I'll treat them as two rigid bodies joined by fixed joints.\n\n    # 1. Frame\n    frame_path = '/asset/frame'\n    frame_xform = UsdGeom.Xform.Define(stage, frame_path)\n    UsdPhysics.RigidBodyAPI.Apply(frame_xform)\n    \n    # Legs (simplified as 4 cylinders/boxes)\n    leg_radius = 0.02\n    leg_height = height * 0.7\n    \n    # Leg positions (approx)\n    leg_pos = [\n        Gf.Vec3f(-width*0.4, 0, -depth*0.4),\n        Gf.Vec3f(width*0.4, 0, -depth*0.4),\n        Gf.Vec3f(-width*0.4, 0, depth*0.4),\n        Gf.Vec3f(width*0.4, 0, depth*0.4),\n    ]\n    \n    for i, pos in enumerate(leg_pos):\n        leg_path = f'{frame_path}/leg_{i}'\n        leg = UsdGeom.Cylinder.Define(stage, leg_path)\n        leg.GetRadiusAttr().Set(leg_radius)\n        leg.GetHeightAttr().Set(leg_height)\n        leg.AddTranslateOp().Set(Gf.Vec3f(pos[0], leg_height/2, pos[2]))\n        UsdShade.MaterialBindingAPI.Apply(leg).Bind(UsdShade.MaterialBindingAPI.AllPurpose(), frame_mat)\n        # Collision\n        UsdPhysics.CollisionAPI.Apply(leg)\n\n    # 2. Seat Shell\n    seat_path = '/asset/seat'\n    seat_xform = UsdGeom.Xform.Define(stage, seat_path)\n    UsdPhysics.RigidBodyAPI.Apply(seat_xform)\n    \n    # Seat mesh (approximate as a box with rounded edges)\n    seat_geo = UsdGeom.Mesh.Define(stage, f'{seat_path}/mesh')\n    # Simplified box mesh for seat\n    sw = width * 0.9\n    sd = depth * 0.9\n    sh = 0.05\n    \n    # Box vertices\n    verts = [\n        (-sw/2, 0, -sd/2), (sw/2, 0, -sd/2), (sw/2, 0, sd/2), (-sw/2, 0, sd/2),\n        (-sw/2, sh, -sd/2), (sw/2, sh, -sd/2), (sw/2, sh, sd/2), (-sw/2, sh, sd/2),\n    ]\n    faces = [\n        (0, 1, 2, 3), (4, 5, 6, 7), # bottom, top\n        (0, 1, 5, 4), (1, 2, 6, 5), (2, 3, 7, 6), (3, 0, 4, 7) # sides\n    ]\n    \n    seat_geo.GetPointsAttr().Set(verts)\n    face_counts = [4] * 6\n    face_indices = [idx for f in faces for idx in f]\n    seat_geo.GetFaceVertexCountsAttr().Set(face_counts)\n    seat_geo.GetFaceVertexIndicesAttr().Set(face_indices)\n    \n    seat_xform.AddTranslateOp().Set(Gf.Vec3f(0, leg_height, 0))\n    UsdShade.MaterialBindingAPI.Apply(seat_geo).Bind(UsdShade.MaterialBindingAPI.AllPurpose(), seat_mat)\n    UsdPhysics.CollisionAPI.Apply(seat_geo)\n\n    # Backrest (approx)\n    back_geo = UsdGeom.Mesh.Define(stage, f'{seat_path}/backrest')\n    bw = width * 0.9\n    bh = height * 0.3\n    bd = 0.04\n    \n    # Backrest vertices\n    b_verts = [\n        (-bw/2, 0, -bd/2), (bw/2, 0, -bd/2), (bw/2, 0, bd/2), (-bw/2, 0, bd/2),\n        (-bw/2, bh, -bd/2), (bw/2, bh, -bd/2), (bw/2, bh, bd/2), (-bw/2, bh, bd/2),\n    ]\n    b_faces = [\n        (0, 1, 2, 3), (4, 5, 6, 7),\n        (0, 1, 5, 4), (1, 2, 6, 5), (2, 3, 7, 6), (3, 0, 4, 7)\n    ]\n    back_geo.GetPointsAttr().Set(b_verts)\n    back_geo.GetFaceVertexCountsAttr().Set([4]*6)\n    back_geo.GetFaceVertexIndicesAttr().Set([idx for f in b_faces for idx in f])\n    back_geo.AddTranslateOp().Set(Gf.Vec3f(0, 0, -sd/2)) # Put it at the back\n    UsdShade.MaterialBindingAPI.Apply(back_geo).Bind(UsdShade.MaterialBindingAPI.AllPurpose(), seat_mat)\n    UsdPhysics.CollisionAPI.Apply(back_geo)\n\n    # Joint: Seat to Frame\n    joint = UsdPhysics.FixedJoint.Define(stage, '/asset/seat_to_frame_joint')\n    joint.GetBody0Rel().SetTargets([frame_path])\n    joint.GetBody1Rel().SetTargets([seat_path])\n    joint.GetLocalPos0Attr().Set(Gf.Vec3f(0, leg_height, 0))\n    joint.GetLocalPos1Attr().Set(Gf.Vec3f(0, 0, 0))\n\n    stage.GetRootLayer().Save()\n\nif __name__ == '__main__':\n    create_chair()\n",
-            "path": "/tmp/author_chair.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-b6d54936f514fd2b",
-            "content": "Successfully wrote to /tmp/author_chair.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3883,
-        "completion_tokens": 2164,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 6047
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "0362beb5",
-        "pi_parent_id": "0853f784",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "53acf89c1ee80437c05e3e33598a4913",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:42:08.515Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-9fda1841b4aa9707",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_chair.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-9fda1841b4aa9707",
-            "content": "Traceback (most recent call last):\n  File \"/tmp/author_chair.py\", line 1, in <module>\n    import pxr.usd as Usd\nModuleNotFoundError: No module named 'pxr.usd'\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6063,
-        "completion_tokens": 22,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6085
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "1f45d604",
-        "pi_parent_id": "d9534045",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "a7b1cde5be60d08df000d0cbad005664",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:42:13.464Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-b81c3de32540a0f8",
-          "function_name": "bash",
-          "arguments": {
-            "command": "pip list | grep pxr"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-b81c3de32540a0f8",
-            "content": "/bin/bash: line 1: pip: command not found\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6153,
-        "completion_tokens": 22,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 6175
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "07b95ca7",
-        "pi_parent_id": "b1d0ef55",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "7e5931df505b87c6727a1d48a8c2187b",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 21865,
-    "total_completion_tokens": 2287,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 6,
-      "pi_total_tokens": 24152
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/evidence/chair_office_amf_black_a01/reconstruction-2/harbor-trajectory-4.json

@@ -1,326 +0,0 @@
-{
-  "schema_version": "ATIF-v1.7",
-  "session_id": "01a15a31-7fcc-75e6-acf7-29b49beb029c",
-  "agent": {
-    "name": "pi",
-    "version": "0.85.1",
-    "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita"
-  },
-  "steps": [
-    {
-      "step_id": 1,
-      "timestamp": "2026-10-20T16:42:38.152Z",
-      "source": "user",
-      "message": [
-        {
-          "type": "text",
-          "text": "<file name=\"/task/thumbnail.png\"></file>\nReconstruct the supplied object as /output/asset.usd.\nNominal dimensions in canonical metres XYZ: [0.5397186279296875, 0.5722833871841431, 0.8096046261489391]\nRead /task/contract.md, /task/tool-docs.md and /task/scoring-policy.json. Identify the rigid parts, joints and their motion before authoring, then reconstruct appearance and mechanical structure. Use Python/PXR. Do not use Blender. Validate and inspect the NVIDIA findings before submitting.\n"
-        },
-        {
-          "type": "image",
-          "source": {
-            "media_type": "image/png",
-            "path": "images/pi_image_1.png"
-          }
-        }
-      ],
-      "extra": {
-        "pi_event_id": "51e55e29",
-        "pi_parent_id": "5398a1e7"
-      }
-    },
-    {
-      "step_id": 2,
-      "timestamp": "2026-10-20T16:42:50.023Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-83ed7452dcd30198",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/contract.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-985107f36be794ff",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/tool-docs.md"
-          }
-        },
-        {
-          "tool_call_id": "chatcmpl-tool-904c649ac218e6e4",
-          "function_name": "read",
-          "arguments": {
-            "path": "/task/scoring-policy.json"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-83ed7452dcd30198",
-            "content": "# Reconstruction contract\n\nReconstruct the supplied object at /output/asset.usd using Python/PXR. Include\nits local USD and texture dependencies and place authoring scripts in /tmp.\nRead the supplied dimensions, image and scoring-policy.json.\n\nThe evaluation reports three independent scores:\n\n- Formal fidelity: NVIDIA SimReady-Foundation Robotics-Prop 4.0.0. Each of its\n  46 mandatory requirements counts once, with 95 * passed / 46 points and five\n  further points when all pass. The 51 optional requirements are reported separately.\n- Visual fidelity: reference-scale shape, colours, textures and surface appearance.\n- Kinematic fidelity: the object's connected rigid parts, joint types, hinge or\n  sliding axes, attachment locations and allowed ranges of motion.\n\nIdentify the object's parts and their mechanical relationships before authoring.\nKeep parts that move relative to one another as distinct rigid bodies and author\nthe corresponding UsdPhysics joints. Specify each joint's connected bodies,\nlocal positions, local orientations, axis and appropriate motion limits. For\nexample, folding eyewear needs hinged temples rather than one rigid mesh.\nUse fixed joints for genuine fixed attachments and keep genuinely rigid objects\nrigid. Base this structure on the supplied observations and the object's function.\n\nCheck that joint frames meet at each attachment, that movable parts can travel\nthrough their intended range and that the bodies and joints have the appropriate\nenabled and kinematic states. Preserve the observed shape and materials while\nconstructing this mechanical structure.\n\nUse validate_candidate to inspect the NVIDIA findings and repair the candidate.\nThe final submission freezes the authored files for evaluation.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-985107f36be794ff",
-            "content": "# Authoring tools\n\nget_task_thumbnail returns the reference image. validate_candidate accepts {\"request_id\":\"check-1\"} and returns the current score, remaining validation calls and nvidia_validator_report.file_path. Search that JSON file with rg for a requirement ID or candidate prim path. get_validation_report reads a selected range of the same report.\n\nUse a new request ID after each repair. submit_asset accepts {\"request_id\":\"final\"} and stops authoring. The model has 32 inference requests, with provider-native output limits. Captures appear in the final report.",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          },
-          {
-            "source_call_id": "chatcmpl-tool-904c649ac218e6e4",
-            "content": "{\n  \"benchmark_output_token_cap\": null,\n  \"composite_scores\": {\n    \"formal\": \"SimReady Robotics-Prop mandatory requirements\",\n    \"kinematic\": \"joint recovery and allowed relative motion correspondence\",\n    \"visual\": \"geometric mean of shape and appearance material correspondence\"\n  },\n  \"limits\": {\n    \"agent_seconds\": 1800,\n    \"final_seconds\": 600,\n    \"inference_requests\": 32,\n    \"preview_requests\": 12,\n    \"preview_seconds\": 90,\n    \"validation_calls\": 5,\n    \"validation_seconds\": 120,\n    \"views_per_preview\": 4\n  },\n  \"model_call_budget\": 32,\n  \"reward_reporting\": {\n    \"formula\": \"95 * passed / fixed_applicable + 5_if_all_pass\",\n    \"harbor_key\": \"simready_reward\",\n    \"scale\": [\n      0,\n      100\n    ]\n  },\n  \"semantic_labels\": {\n    \"criterion\": \"mandatory NVIDIA SL IDs only\",\n    \"required\": true,\n    \"scoring\": true\n  },\n  \"validator_scoring\": {\n    \"all_pass_bonus_percent\": 5,\n    \"applicability\": \"fixed_profile_and_task_before_candidate\",\n    \"applicable_requirement_ids\": [\n      \"com.nvidia.simready.HI.010\",\n      \"com.nvidia.simready.NP.002\",\n      \"com.nvidia.simready.NP.003\",\n      \"com.nvidia.simready.NP.004\",\n      \"com.nvidia.simready.NP.005\",\n      \"com.nvidia.simready.NP.007\",\n      \"com.nvidia.simready.NP.008\",\n      \"com.nvidia.simready.RB.001\",\n      \"com.nvidia.simready.RB.007\",\n      \"com.nvidia.simready.RB.010\",\n      \"com.nvidia.simready.RB.COL.001\",\n      \"com.nvidia.simready.RB.COL.002\",\n      \"com.nvidia.simready.RB.COL.003\",\n      \"com.nvidia.simready.SL.001\",\n      \"com.nvidia.simready.SL.002\",\n      \"com.nvidia.simready.SL.003\",\n      \"com.nvidia.simready.SL.MAT.001\",\n      \"com.nvidia.simready.SL.QCODE.001\",\n      \"com.nvidia.simready.SL.TIME.001\",\n      \"com.nvidia.simready.SR.002\",\n      \"com.nvidia.simready.SR.004\",\n      \"com.nvidia.simready.UN.006\",\n      \"com.nvidia.simready.UN.007\",\n      \"com.nvidia.simready.VG.008\",\n      \"com.nvidia.simready.VG.025\",\n      \"com.nvidia.simready.VG.MESH.001\",\n      \"com.nvidia.simready.VM.MAT.001\",\n      \"com.nvidia.simready.VM.PS.002\",\n      \"com.nvidia.simready.VM.TEX.001\",\n      \"com.nvidia.simready.VM.TEX.003\",\n      \"com.nvidia.usd.AA.001\",\n      \"com.nvidia.usd.AA.002\",\n      \"com.nvidia.usd.HI.001\",\n      \"com.nvidia.usd.HI.003\",\n      \"com.nvidia.usd.HI.004\",\n      \"com.nvidia.usd.RB.003\",\n      \"com.nvidia.usd.RB.005\",\n      \"com.nvidia.usd.RB.009\",\n      \"com.nvidia.usd.RB.COL.004\",\n      \"com.nvidia.usd.VG.002\",\n      \"com.nvidia.usd.VG.014\",\n      \"com.nvidia.usd.VG.027\",\n      \"com.nvidia.usd.VG.028\",\n      \"com.nvidia.usd.VG.029\",\n      \"com.nvidia.usd.VM.BIND.001\",\n      \"com.nvidia.usd.VM.PS.001\"\n    ],\n    \"base_credit_percent\": 95,\n    \"early_stop_penalty_percent\": 0,\n    \"optional_requirements\": \"report_only\",\n    \"policy_id\": \"NVIDIA-Robotics-Prop-4.0.0-mandatory-95-plus-5-v1\",\n    \"process_shaping_percent\": 0,\n    \"profile\": {\n      \"id\": \"Robotics-Prop\",\n      \"version\": \"4.0.0\"\n    },\n    \"profile_inventory_sha256\": \"5a8dccd559cd3625025cedd08f7e2c174df32b8bbe0b17da9c94a588c86da13c\",\n    \"report_only_mandatory_ids\": [],\n    \"schema\": \"simready-validator-scoring-policy-v1\",\n    \"scope\": \"all_mandatory_profile\",\n    \"scoring_unit\": \"one_mandatory_requirement_ID\",\n    \"stop_bonus_percent\": 0,\n    \"unknown\": \"null_score_fixed_denominator\"\n  }\n}",
-            "extra": {
-              "tool_name": "read",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 1919,
-        "completion_tokens": 56,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 1975
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "c6614b8e",
-        "pi_parent_id": "51e55e29",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "9fda0ac4673901a2ae09b47140568753",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 3,
-      "timestamp": "2026-10-20T16:43:01.271Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-87894982e95859f0",
-          "function_name": "bash",
-          "arguments": {
-            "command": "ls -R /task"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-87894982e95859f0",
-            "content": "/task:\ncontract.md\ninstruction.md\nscoring-policy.json\nsimready-bench-native-output-policy.ts\nthumbnail.png\ntool-docs.md",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 3847,
-        "completion_tokens": 21,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 2,
-          "pi_total_tokens": 3868
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "8a0dfb9d",
-        "pi_parent_id": "df959fb0",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "24ce51535e7d3721a79ece5c094b8cb2",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 4,
-      "timestamp": "2026-10-20T16:43:57.885Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-9518345d315e2cd1",
-          "function_name": "write",
-          "arguments": {
-            "content": "import pxr.Usd as Usd\nimport pxr.UsdGeom as UsdGeom\nimport pxr.UsdPhysics as UsdPhysics\nimport pxr.UsdShade as UsdShade\nimport pxr.Gf as Gf\nimport numpy as np\n\ndef create_material(stage, name, color, roughness=0.5, metallic=0.0):\n    material = UsdShade.Material.Define(stage, f\"/World/Materials/{name}\")\n    pbr_shader = UsdShade.Shader.Define(stage, f\"/World/Materials/{name}/Shader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(roughness))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(metallic))\n    \n    material.Create prepaid_output(\"surface\", pbr_shader.ConnectToPrimitive().GetOutput()) # This is wrong, need correct link\n    # Correct linking\n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material\n\n# Re-do create_material with correct API\ndef create_material_fixed(stage, name, color, roughness=0.5, metallic=0.0):\n    material = UsdShade.Material.Define(stage, f\"/World/Materials/{name}\")\n    pbr_shader = UsdShade.Shader.Define(stage, f\"/World/Materials/{name}/Shader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(roughness))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(metallic))\n    \n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material\n\ndef create_box(stage, path, size, pos, rot, material):\n    box = UsdGeom.Cube.Define(stage, path)\n    box.CreateSizeAttr(size)\n    box.AddTranslateOp().Set(pos)\n    box.AddRotateXYZOp().Set(rot)\n    \n    # Assign material\n    UsdShade.MaterialBindingAPI.Apply(box.GetPrim()).Bind(material)\n    return box\n\ndef create_cylinder(stage, path, radius, height, pos, rot, material):\n    cyl = UsdGeom.Cylinder.Define(stage, path)\n    cyl.CreateRadiusAttr(radius)\n    cyl.CreateHeightAttr(height)\n    cyl.AddTranslateOp().Set(pos)\n    cyl.AddRotateXYZOp().Set(rot)\n    \n    UsdShade.MaterialBindingAPI.Apply(cyl.GetPrim()).Bind(material)\n    return cyl\n\ndef main():\n    stage = Usd.Stage.CreateNew(\"/output/asset.usd\")\n    UsdGeom.SetStageUpAxis(stage, UsdGeom.UpAxis.Y)\n    \n    world = UsdGeom.Xform.Define(stage, \"/World\")\n    \n    # Dimensions: [0.5397186279296875, 0.5722833871841431, 0.8096046261489391]\n    # X: 0.54, Y: 0.57, Z: 0.81\n    # Looking at the thumbnail: it's a chair.\n    # X is likely width (across the seat), Z is depth (front to back), Y is height.\n    # Wait, the provided dimensions are [X, Y, Z]. \n    # X = 0.54 (width), Y = 0.57 (maybe depth or height?), Z = 0.81 (likely height)\n    # Let's assume: Width=0.54, Depth=0.57, Height=0.81\n    \n    width = 0.54\n    depth = 0.57\n    height = 0.81\n    \n    mat_black = create_material_fixed(stage, \"black_plastic\", (0.1, 0.1, 0.1), roughness=0.6)\n    mat_fabric = create_material_fixed(stage, \"black_fabric\", (0.05, 0.05, 0.05), roughness=0.9)\n\n    # Rigid bodies\n    # 1. Seat and Backrest (likely one piece or joined)\n    # 2. Legs/Frame\n    \n    # Frame (Legs)\n    # Front legs\n    # Rear legs (connected to backrest)\n    \n    # In this chair, the legs look like a single continuous tube structure.\n    # Let's simplify: a \"Frame\" rig and a \"Seat\" rig.\n    \n    frame = UsdGeom.Xform.Define(stage, \"/World/Frame\")\n    UsdPhysics.RigidBodyAPI.Apply(frame.GetPrim())\n    UsdPhysics.CollisionAPI.Apply(frame.GetPrim())\n    \n    # Simplified Frame as 4 cylinders\n    leg_radius = 0.015\n    leg_height = 0.4\n    \n    # Front Left\n    create_cylinder(stage, \"/World/Frame/Leg_FL\", leg_radius, leg_height, \n                    Gf.Vec3f(-width/2 + 0.05, leg_height/2, depth/2 - 0.05), \n                    Gf.Vec3f(0, 0, 0), mat_black)\n    # Front Right\n    create_cylinder(stage, \"/World/Frame/Leg_FR\", leg_radius, leg_height, \n                    Gf.Vec3f(width/2 - 0.05, leg_height/2, depth/2 - 0.05), \n                    Gf.Vec3f(0, 0, 0), mat_black)\n    # Back Left\n    create_cylinder(stage, \"/World/Frame/Leg_BL\", leg_radius, height, \n                    Gf.Vec3f(-width/2 + 0.05, height/2, -depth/2 + 0.05), \n                    Gf.Vec3f(0, 0, 0), mat_black)\n    # Back Right\n    create_cylinder(stage, \"/World/Frame/Leg_BR\", leg_radius, height, \n                    Gf.Vec3f(width/2 - 0.05, height/2, -depth/2 + 0.05), \n                    Gf.Vec3f(0, 0, 0), mat_black)\n\n    # Seat\n    seat = UsdGeom.Xform.Define(stage, \"/World/Seat\")\n    UsdPhysics.RigidBodyAPI.Apply(seat.GetPrim())\n    UsdPhysics.CollisionAPI.Apply(seat.GetPrim())\n    \n    # Seat cushion\n    create_box(stage, \"/World/Seat/Cushion\", Gf.Vec3f(width, 0.05, depth), \n               Gf.Vec3f(0, leg_height, 0), Gf.Vec3f(0, 0, 0), mat_fabric)\n    \n    # Backrest\n    create_box(stage, \"/World/Seat/Backrest\", Gf.Vec3f(width, height - leg_height, 0.05), \n               Gf.Vec3f(0, (height + leg_height)/2, -depth/2 + 0.025), Gf.Vec3f(0, 0, 0), mat_black)\n    \n    # Joint between Frame and Seat (Fixed)\n    joint = UsdPhysics.FixedJoint.Define(stage, \"/World/Seat_Frame_Joint\")\n    joint.CreateBody0Rel().SetTargets([\"/World/Frame\"])\n    joint.CreateBody1Rel().SetTargets([\"/World/Seat\"])\n    joint.CreateLocalPose0Rel().Set(Gf.XformMatrix(1)) # Identity\n    joint.CreateLocalPose1Rel().Set(Gf.XformMatrix(1)) # Identity\n    \n    stage.GetRootLayer().Save()\n\nif __name__ == \"__main__\":\n    main()\n",
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-9518345d315e2cd1",
-            "content": "Successfully wrote to /tmp/author_asset.py",
-            "extra": {
-              "tool_name": "write",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 4050,
-        "completion_tokens": 2029,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6079
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "64ae0944",
-        "pi_parent_id": "536fa6e3",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "7e7915f862da16e2413e042832e1094d",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 5,
-      "timestamp": "2026-10-20T16:44:00.415Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-84aa07e9d6126a60",
-          "function_name": "bash",
-          "arguments": {
-            "command": "python3 /tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-84aa07e9d6126a60",
-            "content": "File \"/tmp/author_asset.py\", line 16\n    material.Create prepaid_output(\"surface\", pbr_shader.ConnectToPrimitive().GetOutput()) # This is wrong, need correct link\n                    ^^^^^^^^^^^^^^\nSyntaxError: invalid syntax\n\n\nCommand exited with code 1",
-            "extra": {
-              "tool_name": "bash",
-              "is_error": true
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6096,
-        "completion_tokens": 26,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6122
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "6d0e3fdc",
-        "pi_parent_id": "3dc74f87",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "1ebe4a69b3fcbdb3f57c19b6258ecc68",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    },
-    {
-      "step_id": 6,
-      "timestamp": "2026-10-20T16:44:13.641Z",
-      "source": "agent",
-      "model_name": "harbor-endpoint/google/gemma-4-31B-it:novita",
-      "message": "",
-      "tool_calls": [
-        {
-          "tool_call_id": "chatcmpl-tool-9a0625b6713bba76",
-          "function_name": "edit",
-          "arguments": {
-            "edits": [
-              {
-                "newText": "def create_material_fixed(stage, name, color, roughness=0.5, metallic=0.0):\n    material = UsdShade.Material.Define(stage, f\"/World/Materials/{name}\")\n    pbr_shader = UsdShade.Shader.Define(stage, f\"/World/Materials/{name}/Shader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(roughness))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(metallic))\n    \n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material",
-                "oldText": "def create_material(stage, name, color, roughness=0.5, metallic=0.0):\n    material = UsdShade.Material.Define(stage, f\"/World/Materials/{name}\")\n    pbr_shader = UsdShade.Shader.Define(stage, f\"/World/Materials/{name}/Shader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(roughness))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(metallic))\n    \n    material.Create prepaid_output(\"surface\", pbr_shader.ConnectToPrimitive().GetOutput()) # This is wrong, need correct link\n    # Correct linking\n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material\n\n# Re-do create_material with correct API\ndef create_material_fixed(stage, name, color, roughness=0.5, metallic=0.0):\n    material = UsdShade.Material.Define(stage, f\"/World/Materials/{name}\")\n    pbr_shader = UsdShade.Shader.Define(stage, f\"/World/Materials/{name}/Shader\")\n    pbr_shader.CreateIdAttr(\"UsdPreviewSurface\")\n    pbr_shader.CreateInput(\"diffuseColor\", Gf.Vec3f(color))\n    pbr_shader.CreateInput(\"roughness\", Gf.Float(roughness))\n    pbr_shader.CreateInput(\"metallic\", Gf.Float(metallic))\n    \n    material.CreateSurfaceOutput().ConnectToSource(pbr_shader.ConnectableAPI(), \"surface\")\n    return material"
-              }
-            ],
-            "path": "/tmp/author_asset.py"
-          }
-        }
-      ],
-      "observation": {
-        "results": [
-          {
-            "source_call_id": "chatcmpl-tool-9a0625b6713bba76",
-            "content": "Successfully replaced 1 block(s) in /tmp/author_asset.py.",
-            "extra": {
-              "tool_name": "edit",
-              "is_error": false
-            }
-          }
-        ]
-      },
-      "metrics": {
-        "prompt_tokens": 6193,
-        "completion_tokens": 641,
-        "cached_tokens": 0,
-        "extra": {
-          "cache_write_tokens": 0,
-          "reasoning_tokens": 0,
-          "pi_total_tokens": 6834
-        }
-      },
-      "llm_call_count": 1,
-      "extra": {
-        "pi_event_id": "86ac29ed",
-        "pi_parent_id": "cdec7daa",
-        "provider": "harbor-endpoint",
-        "api": "openai-completions",
-        "response_id": "90bcb8379fabc011428c03b0a572e857",
-        "response_model": "google/gemma-4-31b-it",
-        "stop_reason": "toolUse",
-        "raw_stop_reason": "tool_calls"
-      }
-    }
-  ],
-  "final_metrics": {
-    "total_prompt_tokens": 22105,
-    "total_completion_tokens": 2773,
-    "total_cached_tokens": 0,
-    "total_steps": 6,
-    "extra": {
-      "cache_write_tokens": 0,
-      "reasoning_tokens": 2,
-      "pi_total_tokens": 24878
-    }
-  }
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/execution-interventions.json

@@ -1,73 +0,0 @@
-{
-  "schema": "simreadybench-execution-interventions/v1",
-  "run_id": "fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020",
-  "model": "google/gemma-4-31B-it:novita",
-  "released_version": "1.0.0",
-  "source_commit": "be18d4e4c6fd0bb181a11cfc85c256ea3545bb5b",
-  "wheel_sha256": "d00b77f3ee9fe8572f8c417de42e7c4d33c06159c2dc259f8c8a4fe969152059",
-  "accounting_interruption": {
-    "cause": "The Daytona loopback adapter omitted the verified HF route pin from usage settlement evidence",
-    "original_settlement_rejection": "usage_hf_route_pin_required",
-    "observed_outcome": "Six conditions exhausted conservative request holds before a valid submission",
-    "affected_conditions": [
-      {
-        "condition_id": "condition-2a49dc3249ebc5a481bbd3eeda5e12de15a8ea5abfb19c1e6544936b46197cf8",
-        "asset_id": "accessory_eyeglasses_a01",
-        "repetition": 0,
-        "termination_reason": "project_spend_budget_exhausted"
-      },
-      {
-        "condition_id": "condition-45aec585a3af5e25a792b8db34a7c1d9a04a2b9432f465a7dfa8624bdfb04fba",
-        "asset_id": "accessory_eyeglasses_a01",
-        "repetition": 1,
-        "termination_reason": "project_spend_budget_exhausted"
-      },
-      {
-        "condition_id": "condition-1080513aeffe831210ac7aeb2958e518183cd857e8ca9391bb434e39c3d70f29",
-        "asset_id": "box_cardboard_open_a01",
-        "repetition": 0,
-        "termination_reason": "project_spend_budget_exhausted"
-      },
-      {
-        "condition_id": "condition-a8be3130d1a409b0ba4d4cfe87536383ae9bd3f7d4e6012c99addcfccc87dfce",
-        "asset_id": "box_cardboard_open_a01",
-        "repetition": 1,
-        "termination_reason": "project_spend_budget_exhausted"
-      },
-      {
-        "condition_id": "condition-8d236f324c42282c4917c28eda029d5fa82378eb77b346ed3480550b0b9008f1",
-        "asset_id": "chair_office_amf_black_a01",
-        "repetition": 0,
-        "termination_reason": "project_spend_budget_exhausted"
-      },
-      {
-        "condition_id": "condition-c5c63362f2f9827294e482e4d613133c580afff62c1caed83a3c01cf48ade4d1",
-        "asset_id": "chair_office_amf_black_a01",
-        "repetition": 1,
-        "termination_reason": "project_spend_budget_exhausted"
-      }
-    ],
-    "original_results_commit": "086d3fc6c7a09be53513b76e371c63ebf1e9601e",
-    "original_scores_preserved": true,
-    "closed_conditions_retried": false
-  },
-  "controller_hold": {
-    "held_at": "2026-10-20T16:45:04.037455+00:00",
-    "resumed_at": "2026-10-20T16:59:45.805528+00:00",
-    "same_sandboxes_resumed": true
-  },
-  "active_episode_repair": {
-    "helper_sha256": "26e70e66cd89843d4e8ccdfd1b3c943385261896834420206c5cfdd8475b24fb",
-    "route_metadata_sha256": "8db6609a62912776c0ac017cdae69a20a4f528c291c641d7466964e8e3067ede",
-    "method": "Released usage-priced settlement with separately verified prelaunch route metadata",
-    "original_gateway_records_unchanged": true,
-    "original_reservations_unchanged": true,
-    "terminated_episodes_excluded": true
-  },
-  "maximum_model_spend_usd": 75,
-  "maximum_generation_requests": 3200,
-  "maximum_requests_per_condition": 32,
-  "budget_increase_usd": 0,
-  "scorer_changes": false,
-  "scientific_identity_changes": false
-}

runs/fresh-v1-nv25-gemma-4-31b-it-novita-pi-none-20261020/report.html

@@ -1 +0,0 @@
-<!doctype html><html lang="en"><meta charset="utf-8"><meta name="viewport" content="width=device-width"><meta http-equiv="Content-Security-Policy" content="default-src 'none'; img-src 'self'; style-src 'unsafe-inline'"><title>SimReadyBench results</title><style>body{font:16px Georgia,serif;line-height:1.6;max-width:1200px;margin:3rem auto;padding:0 1.5rem;background:#fff;color:#202020}a{color:#254e75}h1,h2,h3{line-height:1.25}summary{cursor:pointer}details{margin:1rem 0}.images{display:grid;grid-template-columns:repeat(auto-fit,minmax(260px,1fr));gap:1.5rem}figure{margin:0}img{width:100%;height:auto}figcaption{font-size:14px}table{border-collapse:collapse;width:100%;font-size:14px}th,td{text-align:left;vertical-align:top;border-bottom:1px solid #ddd;padding:8px}.table{overflow-x:auto;margin:1rem 0}section{border-top:1px solid #ddd;padding:1.5rem 0}.artefacts{font-size:14px}</style><h1>SimReadyBench results</h1><p>Formal fidelity measures SimReady Robotics-Prop requirements. Visual fidelity combines shape and appearance materials. Kinematic fidelity measures joint recovery and motion correspondence. Each score is reported independently.</p><p>6 of 100 reconstruction attempts completed.</p><p><a href="scores.csv">Scores CSV</a> | <a href="results.json">Detailed results JSON</a></p><h2>Results by asset</h2><div class="table"><table><thead><tr><th>Asset</th><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>accessory_eyeglasses_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>0% (0/2)</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/4)</td><td>0% (0/29)</td><td>0% (0/15341)</td></tr><tr><td>accessory_eyeglasses_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>0% (0/2)</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/4)</td><td>0% (0/29)</td><td>0% (0/15341)</td></tr><tr><td>box_cardboard_open_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/14336)</td></tr><tr><td>box_cardboard_open_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/14336)</td></tr><tr><td>chair_office_amf_black_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/16384)</td></tr><tr><td>chair_office_amf_black_a01</td><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/16384)</td></tr></tbody></table></div><section id="attempt-1"><h2>accessory_eyeglasses_a01 / google/gemma-4-31B-it:novita / repetition 1</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>0% (0/2)</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/4)</td><td>0% (0/29)</td><td>0% (0/15341)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/4</td><td>4</td><td>4</td><td>100</td></tr><tr><td>Joints</td><td>0</td><td>0/29</td><td>29</td><td>29</td><td>100</td></tr><tr><td>Materials</td><td>0</td><td>0/15341</td><td>15341</td><td>15341</td><td>62.42</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>0</td><td>0/1</td><td>1/1</td><td>100</td></tr><tr><td>Moving joints</td><td>0</td><td>0/2</td><td>2/2</td><td>100</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody><tr><td>/RootNode/Joints/joint_arm_02_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr><tr><td>/RootNode/Joints/joint_arm_01_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr><tr><td>/RootNode/Joints/joint_glass_01_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>3.34</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/arm_01_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/arm_02_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/base_01_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/glass_01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>3</td><td>0</td><td>0/29</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/4</td><td>4/4</td></tr><tr><td>body_kinematic</td><td>0/4</td><td>4/4</td></tr></tbody></table></div><h3>Joint checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>existence_type_connectivity</td><td>0/3</td><td>3/3</td></tr><tr><td>joint_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body0_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body0_kinematic</td><td>0/3</td><td>3/3</td></tr><tr><td>body1_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body1_kinematic</td><td>0/3</td><td>3/3</td></tr><tr><td>anchors</td><td>0/3</td><td>3/3</td></tr><tr><td>axes</td><td>0/2</td><td>2/2</td></tr><tr><td>relative_orientation</td><td>0/3</td><td>3/3</td></tr><tr><td>limits</td><td>0/2</td><td>2/2</td></tr><tr><td>joint_count</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:metallic</td><td>0/1005</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="assets/accessory_eyeglasses_a01/reconstruction-1/candidate/asset.usd">Generated USD</a> | <a href="assets/accessory_eyeglasses_a01/reconstruction-1/candidate/manifest.json">Asset package manifest</a> | <a href="evidence/accessory_eyeglasses_a01/reconstruction-1/nvidia-validator.json">NVIDIA validation findings</a> | <a href="evidence/accessory_eyeglasses_a01/reconstruction-1/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section><section id="attempt-2"><h2>accessory_eyeglasses_a01 / google/gemma-4-31B-it:novita / repetition 2</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>0% (0/2)</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/4)</td><td>0% (0/29)</td><td>0% (0/15341)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/4</td><td>4</td><td>4</td><td>100</td></tr><tr><td>Joints</td><td>0</td><td>0/29</td><td>29</td><td>29</td><td>100</td></tr><tr><td>Materials</td><td>0</td><td>0/15341</td><td>15341</td><td>15341</td><td>62.42</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>0</td><td>0/1</td><td>1/1</td><td>100</td></tr><tr><td>Moving joints</td><td>0</td><td>0/2</td><td>2/2</td><td>100</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody><tr><td>/RootNode/Joints/joint_arm_02_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr><tr><td>/RootNode/Joints/joint_arm_01_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr><tr><td>/RootNode/Joints/joint_glass_01_joint_00</td><td>Not measured</td><td>0</td><td>Not measured</td><td>missing_joint</td></tr></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>3.34</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/arm_01_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/arm_02_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/base_01_obj_00</td><td>Not measured</td><td>0</td></tr><tr><td>/RootNode/Geometry/glass_01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>3</td><td>0</td><td>0/29</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/4</td><td>4/4</td></tr><tr><td>body_kinematic</td><td>0/4</td><td>4/4</td></tr></tbody></table></div><h3>Joint checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>existence_type_connectivity</td><td>0/3</td><td>3/3</td></tr><tr><td>joint_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body0_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body0_kinematic</td><td>0/3</td><td>3/3</td></tr><tr><td>body1_enabled</td><td>0/3</td><td>3/3</td></tr><tr><td>body1_kinematic</td><td>0/3</td><td>3/3</td></tr><tr><td>anchors</td><td>0/3</td><td>3/3</td></tr><tr><td>axes</td><td>0/2</td><td>2/2</td></tr><tr><td>relative_orientation</td><td>0/3</td><td>3/3</td></tr><tr><td>limits</td><td>0/2</td><td>2/2</td></tr><tr><td>joint_count</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:metallic</td><td>0/1005</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="assets/accessory_eyeglasses_a01/reconstruction-2/candidate/asset.usd">Generated USD</a> | <a href="assets/accessory_eyeglasses_a01/reconstruction-2/candidate/manifest.json">Asset package manifest</a> | <a href="evidence/accessory_eyeglasses_a01/reconstruction-2/nvidia-validator.json">NVIDIA validation findings</a> | <a href="evidence/accessory_eyeglasses_a01/reconstruction-2/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section><section id="attempt-3"><h2>box_cardboard_open_a01 / google/gemma-4-31B-it:novita / repetition 1</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/14336)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Joints</td><td>Not applicable</td><td>0/0</td><td>0</td><td>0</td><td>0</td></tr><tr><td>Materials</td><td>0</td><td>0/14336</td><td>14336</td><td>14336</td><td>58.33</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr><tr><td>Moving joints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>9.7</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/sm_box_a01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>0/0</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/1</td><td>1/1</td></tr><tr><td>body_kinematic</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="evidence/box_cardboard_open_a01/reconstruction-1/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section><section id="attempt-4"><h2>box_cardboard_open_a01 / google/gemma-4-31B-it:novita / repetition 2</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/14336)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Joints</td><td>Not applicable</td><td>0/0</td><td>0</td><td>0</td><td>0</td></tr><tr><td>Materials</td><td>0</td><td>0/14336</td><td>14336</td><td>14336</td><td>58.33</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr><tr><td>Moving joints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>9.7</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/sm_box_a01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>0/0</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/1</td><td>1/1</td></tr><tr><td>body_kinematic</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="evidence/box_cardboard_open_a01/reconstruction-2/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section><section id="attempt-5"><h2>chair_office_amf_black_a01 / google/gemma-4-31B-it:novita / repetition 1</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>1</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/16384)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Joints</td><td>Not applicable</td><td>0/0</td><td>0</td><td>0</td><td>0</td></tr><tr><td>Materials</td><td>0</td><td>0/16384</td><td>16384</td><td>16384</td><td>66.67</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr><tr><td>Moving joints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>12.29</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/sm_chair_office_amf_a01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>0/0</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/1</td><td>1/1</td></tr><tr><td>body_kinematic</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:diffuseColor</td><td>0/2048</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="evidence/chair_office_amf_black_a01/reconstruction-1/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section><section id="attempt-6"><h2>chair_office_amf_black_a01 / google/gemma-4-31B-it:novita / repetition 2</h2><div class="table"><table><thead><tr><th>Model</th><th>Harness</th><th>Skills</th><th>Repetition</th><th>Status</th></tr></thead><tbody><tr><td>google/gemma-4-31B-it:novita</td><td>pi</td><td>none</td><td>2</td><td>completed</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Formal fidelity (%)</th><th>Visual fidelity (%)</th><th>Kinematic fidelity (%)</th><th>SimReady Robotics-Prop optional</th><th>Shape (%)</th><th>Rigid parts (%)</th><th>Joints (%)</th><th>Materials (%)</th></tr></thead><tbody><tr><td>0% (0/46)</td><td>0% (0/2)</td><td>N/A</td><td>Not measured (0/51), 51 unmeasured, bounds 0%-100%</td><td>0% (0/1)</td><td>0% (0/1)</td><td>N/A</td><td>0% (0/16384)</td></tr></tbody></table></div><p>0 of 46 mandatory requirements passed.</p><h3>Direct USD reconstruction fidelity</h3><div class="table"><table><thead><tr><th>Component</th><th>Fidelity (%)</th><th>Credit</th><th>Measured</th><th>Eligible</th><th>Reference coverage (%)</th></tr></thead><tbody><tr><td>Shape</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Rigid parts</td><td>0</td><td>0/1</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Joints</td><td>Not applicable</td><td>0/0</td><td>0</td><td>0</td><td>0</td></tr><tr><td>Materials</td><td>0</td><td>0/16384</td><td>16384</td><td>16384</td><td>66.67</td></tr></tbody></table></div><h3>Appearance materials</h3><div class="table"><table><thead><tr><th>Score (%)</th><th>Earned/eligible</th><th>Status</th></tr></thead><tbody><tr><td>0</td><td>0/2048</td><td>measured</td></tr></tbody></table></div><h3>Kinematic correspondence</h3><div class="table"><table><thead><tr><th>Factor</th><th>Score (%)</th><th>Earned/eligible</th><th>Measured/eligible</th><th>Coverage (%)</th></tr></thead><tbody><tr><td>Fixed constraints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr><tr><td>Moving joints</td><td>Not measured</td><td>0/0</td><td>0/0</td><td>0</td></tr></tbody></table></div><div class="table"><table><thead><tr><th>Reference joint</th><th>Candidate joint</th><th>Score (%)</th><th>Motion RMS (mm)</th><th>Status</th></tr></thead><tbody></tbody></table></div><h3>Surface correspondence</h3><div class="table"><table><thead><tr><th>Precision (%)</th><th>Recall (%)</th><th>Samples per direction</th><th>Distance band (mm)</th><th>RMS distance (mm)</th><th>95th-percentile distance (mm)</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>2048</td><td>12.29</td><td>Not measured</td><td>Not measured</td></tr></tbody></table></div><h3>Rigid-part correspondence</h3><div class="table"><table><thead><tr><th>Reference body</th><th>Candidate body</th><th>Surface correspondence (%)</th></tr></thead><tbody><tr><td>/RootNode/Geometry/sm_chair_office_amf_a01_obj_00</td><td>Not measured</td><td>0</td></tr></tbody></table></div><h3>Joint correspondence</h3><div class="table"><table><thead><tr><th>Reference joints</th><th>Candidate joints</th><th>Passed/eligible checks</th></tr></thead><tbody><tr><td>0</td><td>0</td><td>0/0</td></tr></tbody></table></div><h3>Rigid-structure checks</h3><div class="table"><table><thead><tr><th>Feature</th><th>Passed/eligible</th><th>Measured/eligible</th></tr></thead><tbody><tr><td>body_count</td><td>0/1</td><td>1/1</td></tr><tr><td>joint_inventory_count</td><td>0/1</td><td>1/1</td></tr><tr><td>body_enabled</td><td>0/1</td><td>1/1</td></tr><tr><td>body_kinematic</td><td>0/1</td><td>1/1</td></tr></tbody></table></div><h3>Material correspondence</h3><div class="table"><table><thead><tr><th>Property</th><th>Credit/eligible</th><th>Correspondence (%)</th></tr></thead><tbody><tr><td>physics:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>physics:density</td><td>0/2048</td><td>0</td></tr><tr><td>physics:dynamicFriction</td><td>0/2048</td><td>0</td></tr><tr><td>physics:restitution</td><td>0/2048</td><td>0</td></tr><tr><td>physics:staticFriction</td><td>0/2048</td><td>0</td></tr><tr><td>visual:assignment</td><td>0/2048</td><td>0</td></tr><tr><td>visual:diffuseColor</td><td>0/2048</td><td>0</td></tr><tr><td>visual:opacity</td><td>0/2048</td><td>0</td></tr></tbody></table></div><p class="artefacts"><a href="evidence/chair_office_amf_black_a01/reconstruction-2/harbor-trajectory-4.json">Harbor trajectory 1</a></p></section></html>

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-1/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:0c74645ae9267c952681d8124637f2b3965b8b11f40252c4facbd3ee166bf34d
-size 192046

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-1/candidate/manifest.json

@@ -1,29 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "0c74645ae9267c952681d8124637f2b3965b8b11f40252c4facbd3ee166bf34d",
-      "size_bytes": 192046
-    },
-    {
-      "relative_path": "textures/acetate.png",
-      "sha256": "006960e42e326fd374b2affc2411e4e89076eedecaa7c0c36db6eb7bbe26a718",
-      "size_bytes": 724776
-    },
-    {
-      "relative_path": "textures/lens.png",
-      "sha256": "5d26ee9248f71c72ac8a5f057b4d66dc6506b8743a43abb517ca16cd6bfd4503",
-      "size_bytes": 811
-    },
-    {
-      "relative_path": "textures/lettering.png",
-      "sha256": "8ae8828c46f5e1260fe2c661c35453527d8dd36680be1c90501bcaf83652a0c4",
-      "size_bytes": 4629
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 922262,
-  "snapshot_sha256": "885a50ba0bf16676b4318eda0a8589d4868a1376f075e8689343ece1fc58bcff"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-2/candidate/manifest.json

@@ -1,24 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "402eff31f89f535216709a98ae2464d016eb5d0791da26bd149413695f567f5f",
-      "size_bytes": 492
-    },
-    {
-      "relative_path": "textures/acetate.png",
-      "sha256": "9b01b6359e5c7e7dc961326e4f918617070f74175429c69d09816963c67832d0",
-      "size_bytes": 707521
-    },
-    {
-      "relative_path": "textures/temple.png",
-      "sha256": "140797f8d8184427966fd676d6d08bc2979c5ec4778265d2cbb107f1ed2b48b9",
-      "size_bytes": 711984
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 1419997,
-  "snapshot_sha256": "cfa743b2a9f7e7dee7ad2c503f8a7e3d67553edd2ea4734e69fccb3c46d658e8"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-3/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:48419f9a73a8c4cbb159ef24f086a453eaf737b1233934f79f523c9476fde011
-size 324823

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-3/candidate/manifest.json

@@ -1,19 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "48419f9a73a8c4cbb159ef24f086a453eaf737b1233934f79f523c9476fde011",
-      "size_bytes": 324823
-    },
-    {
-      "relative_path": "textures/acetate.png",
-      "sha256": "07fbd0eb84d1c1c9bd39213da9fb3c6101d7b265cefb3333ac070fab34efbb3a",
-      "size_bytes": 673807
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 998630,
-  "snapshot_sha256": "9632953833bdd58ae9233544cebc9027722309f437aaa765c73d2b3565e85b96"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-4/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:fe38cf997738603b4ff9871e874334eab62548149ad03a44a3aef1f69afc7308
-size 280411

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/accessory_eyeglasses_a01/reconstruction-4/candidate/manifest.json

@@ -1,19 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "fe38cf997738603b4ff9871e874334eab62548149ad03a44a3aef1f69afc7308",
-      "size_bytes": 280411
-    },
-    {
-      "relative_path": "textures/acetate.png",
-      "sha256": "3874009b0ae12338b900a2a819e748c12ddaa6799a7d454755086c40af21db89",
-      "size_bytes": 605406
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 885817,
-  "snapshot_sha256": "3abffa036bd61bf16841bfcf0b96709bc9f1f83b2aaef6136f4fd93167ba8945"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-2/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a0f465b1a62876a164380e90c6e9ec55551e6c008a19a87baddbf07450132ec5
-size 216768

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-2/candidate/manifest.json

@@ -1,39 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": ".thumbs/256x256/asset.usd.png",
-      "sha256": "09fa1d7a5abd55434fc3c4fe612364e565a8ca7507f82a40f11c6d2e62f89b38",
-      "size_bytes": 44883
-    },
-    {
-      "relative_path": "asset.usd",
-      "sha256": "a0f465b1a62876a164380e90c6e9ec55551e6c008a19a87baddbf07450132ec5",
-      "size_bytes": 216768
-    },
-    {
-      "relative_path": "textures/KraftPaper.png",
-      "sha256": "21bbd9f755c77dc87b0584ab441dbb003bc7d2d2ae45711753e7681a6ff364fb",
-      "size_bytes": 942769
-    },
-    {
-      "relative_path": "textures/KraftPaper_rough.png",
-      "sha256": "63110544e6e86e6b6801653f4c2455ee41ab510320e681a98d50af4ceeb179ba",
-      "size_bytes": 537561
-    },
-    {
-      "relative_path": "textures/WhitePaper.png",
-      "sha256": "4c29dafb8c4cef70cea6bdc367d12d339b7d3df7b85326272904893e87ed3c11",
-      "size_bytes": 756626
-    },
-    {
-      "relative_path": "textures/WhitePaper_rough.png",
-      "sha256": "27f30f5fcbbd06e43b81fbf0c927c49a31efa1e99eafc08744847549d3e4f7f2",
-      "size_bytes": 419024
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 2917631,
-  "snapshot_sha256": "919ef3f8f5205c817980c6e237fceb2966e9a2082baf4fbda132fa066616ef0f"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-3/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b79c6f1339369ba98df7aa43e16abe1cdaea87f32772e785b809cc4ac3f6b6c6
-size 144088

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-3/candidate/manifest.json

@@ -1,44 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": ".thumbs/256x256/asset.usd.png",
-      "sha256": "573cfa4cccf7534e0080ff363a2efd631bffb7d9bb3dbc7e040901aaf66a0a76",
-      "size_bytes": 35023
-    },
-    {
-      "relative_path": "asset.usd",
-      "sha256": "b79c6f1339369ba98df7aa43e16abe1cdaea87f32772e785b809cc4ac3f6b6c6",
-      "size_bytes": 144088
-    },
-    {
-      "relative_path": "preview.png",
-      "sha256": "3095e57a644dcfa592da0568c2b6b1314f81e9cfe11faae2134ae24bf87c620e",
-      "size_bytes": 232470
-    },
-    {
-      "relative_path": "textures/edge.png",
-      "sha256": "35d494a41d24dec8a84752d89cfaddcd6dd5671e9e292d6c8c24e99ff1678dab",
-      "size_bytes": 788462
-    },
-    {
-      "relative_path": "textures/insideWhite.png",
-      "sha256": "2b84dd0b50cb7e5d81c7898a8c31d8198cded2d955d86dac728ec6c57fb1d6c3",
-      "size_bytes": 463353
-    },
-    {
-      "relative_path": "textures/kraft.png",
-      "sha256": "b6ba66f16b2318cc80f08b9f4edebdabfb76c8e9a642b101c5254584c01675a7",
-      "size_bytes": 744871
-    },
-    {
-      "relative_path": "textures/white.png",
-      "sha256": "1c2fa0d8bdb76384e77873a87745081651e152976881536369c9c98744079a49",
-      "size_bytes": 399152
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 2807419,
-  "snapshot_sha256": "46c09d365b7747051598468b370ca9ce78a80d2fabd5ce1c30401f51192c75c4"
-}

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-4/candidate/asset.usd

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:07e3ae1c761dac30e26d917cb86205a398bfff5e80f23790e5fd517f09c08925
-size 125414

runs/fresh-v1-nv25-gpt-6.1-sol-pi-none-medium-20261020/assets/box_cardboard_open_a01/reconstruction-4/candidate/manifest.json

@@ -1,24 +0,0 @@
-{
-  "copied_independently": true,
-  "entrypoint": "asset.usd",
-  "files": [
-    {
-      "relative_path": "asset.usd",
-      "sha256": "07e3ae1c761dac30e26d917cb86205a398bfff5e80f23790e5fd517f09c08925",
-      "size_bytes": 125414
-    },
-    {
-      "relative_path": "textures/kraft.png",
-      "sha256": "b1ba21b2121a60e1859728fa5a54ef9d6d38fcfc644324aaefb63dec5117a0e9",
-      "size_bytes": 1043865
-    },
-    {
-      "relative_path": "textures/white_paper.png",
-      "sha256": "830c6c892913bd3ab0c72498ed88ac3943c8f4b1a36f57821bfbd2e14335f2db",
-      "size_bytes": 841177
-    }
-  ],
-  "schema": "simready-bench-submission-v1",
-  "size_bytes": 2010456,
-  "snapshot_sha256": "f678b662e8c735b23e73a16d649054b88f2cf49526fb6f545db21a49a9feb6e7"
-}