Replace release task 001 with audited source task 120

data/statistics.md

@@ -42,12 +42,13 @@ Science-knowledge ablation rows: **91**.
 | `computational-mechanics-and-soft-filament-dynamics` | 1 |
 | `computational-oceanography-and-spherical-geometry` | 1 |
 | `computational-oceanography-finite-volume-grids` | 1 |
+| `computational-reaction-chemistry` | 1 |
 | `computational-thermodynamics-calphad` | 1 |
 | `computer-algebra` | 1 |
 | `condensed-matter-magnetism-and-magnetic-symmetry` | 1 |
 | `convex-optimization-parametric-workflows` | 1 |
 | `crystallography-file-format-semantics` | 1 |
-| `density-functional-theory` | 2 |
+| `density-functional-theory` | 1 |
 | `diffusion-mri-denoising` | 1 |
 | `drug-discovery-cheminformatics` | 1 |
 | `earth-science-data-semantics` | 1 |
@@ -141,14 +142,14 @@ Science-knowledge ablation rows: **91**.
 | `Academic-NonCommercial` | 1 |
 | `Apache-2.0` | 13 |
 | `BSD-2-Clause` | 2 |
-| `BSD-3-Clause` | 37 |
+| `BSD-3-Clause` | 36 |
 | `GPL-3.0` | 1 |
 | `GPL-3.0-or-later` | 1 |
 | `LGPL-2.1-family` | 1 |
 | `LGPL-2.1-family and MIT (upstream conflict)` | 1 |
 | `LGPL-3.0-family` | 8 |
 | `LGPL-3.0-or-later` | 4 |
-| `MIT` | 45 |
+| `MIT` | 46 |
 | `MIT AND BSD-3-Clause` | 1 |
 | `MIT AND LGPL-family AND BSD-3-Clause` | 1 |
 | `MIT/X11 (BSD-like; see upstream copyright manifest)` | 1 |

data/tasks.csv

@@ -1,5 +1,5 @@
 task_id,science_knowledge_ablation,title,domain,language,repository_url,base_commit,source_license,gpl_family,restricted_license,license_gate,material_license,material_license_source,material_restricted,materials_gate,materials_manifest_sha256,material_licenses,materials_provenance,restricted_reason,environment_image,verifier_image,image_platform,task_path,status
-001,False,Repair omega-dependent range-separated PBE workflow in a real PySCF STRSF repository,density-functional-theory,python,https://github.com/aditisingh4812/simplified_tuned_range_separated,f794ca5265d38906782506f1ac4e197b4e273fc4,BSD-3-Clause,False,False,none,,,,,,[],,,docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:ba7befeae4625761cb107ea17ff7962f0d7fd45a71eb25cf75369dc93da8f9d3,docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64,linux/amd64,tasks/task_001,dockerhub-published
+001,False,Repair an inconsistent transition-state rotor workflow,computational-reaction-chemistry,python,https://github.com/Auto-Mech/autochem,cbe1a0c1102b8125608841eb8ac536cfe3b09671,MIT,False,False,none,,,False,False,,[],,,docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:0b6b41e556cd862f10e9e8c489e9b539eda27d277718e0504de440048c62d3c7,docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0,linux/amd64,tasks/task_001,dockerhub-published
 002,True,Repair an inconsistent reduced validation workflow,density-functional-theory,python,https://github.com/aditisingh4812/simplified_tuned_range_separated,f794ca5265d38906782506f1ac4e197b4e273fc4,BSD-3-Clause,False,False,none,,,,,,[],,,docker.io/kevinxulearning/swe-bench-science-environment-python-task-002:v0.1.0@sha256:13e8c428f486c685ec92b031e46f3520bac3d4e26b6bdd83fde9826a488b7761,docker.io/kevinxulearning/swe-bench-science-verifier-python-task-002:v0.1.0@sha256:a863b98f920e1e9b1aede8bd0460d039628aab19a413cce3dad7417028e1bd1a,linux/amd64,tasks/task_002,dockerhub-published
 003,True,Repair extxyz stress and virial tensor-label preservation,atomistic-machine-learning-data,python,https://github.com/deepmodeling/dpdata,4d8048b51ee0cc00caa4854844c28b7f892fd908,LGPL-3.0-family,True,True,gpl-family,,,,,,[],,,docker.io/kevinxulearning/swe-bench-science-environment-python-task-003:v0.1.0@sha256:f161a1e658abd71c5c3abda59b61814a2c5bf140a4826b5598a917d62a258f4e,docker.io/kevinxulearning/swe-bench-science-verifier-python-task-003:v0.1.0@sha256:4d2e13b726b99a0c9d302ba913cdaa727c49fdc403b65b68918ec70c1713a790,linux/amd64,tasks/task_003,dockerhub-published
 004,True,Repair split-read exon-overlap filtering,rna-seq-genomic-interval-semantics,c++,https://github.com/arq5x/bedtools2,0f45761e8ef6ffc093f649b0145788577a62ed33,MIT,False,False,none,,,,,,[],,,docker.io/kevinxulearning/swe-bench-science-environment-c-task-004:v0.1.0@sha256:290695ddb1d2a160142311b604ccb61ff5d51d9500b6d218c56b43cbb94d9620,docker.io/kevinxulearning/swe-bench-science-verifier-c-task-004:v0.1.0@sha256:2553def59dc88f1886ad281da1d561600560a6f8391f177d33bb9f4975e7cc4f,linux/amd64,tasks/task_004,dockerhub-published

manifests/release_provenance.json

@@ -16,7 +16,7 @@
     "verifier_digest_count": 119,
     "pair_definition": "For each release_id, the environment_image and verifier_image fields in manifests/tasks.jsonl form the immutable pair used by the runner.",
     "verifier_base_rule": "A verifier is built FROM the corresponding environment reference pinned by its digest.",
-    "manifest_sha256": "d078e9b47f112af8c05e36571ff112566ed91dac007a32338e50c16032722554"
+    "manifest_sha256": "c635721fb0685ffa186dbf2eacb0a8052882581bd4558feddb63b8a6d34966b2"
   },
   "clean_evaluation": {
     "authoritative_image_source": "The immutable Docker Hub references in manifests/tasks.jsonl; no alternate or historical image set is selected at runtime.",

manifests/tasks.jsonl

@@ -1,4 +1,4 @@
-{"base_commit": "f794ca5265d38906782506f1ac4e197b4e273fc4", "domain": "density-functional-theory", "environment_image": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:ba7befeae4625761cb107ea17ff7962f0d7fd45a71eb25cf75369dc93da8f9d3", "environment_image_digest": "sha256:ba7befeae4625761cb107ea17ff7962f0d7fd45a71eb25cf75369dc93da8f9d3", "environment_image_id": "sha256:ba7befeae4625761cb107ea17ff7962f0d7fd45a71eb25cf75369dc93da8f9d3", "environment_image_tag": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0", "gpl_family": false, "image_platform": "linux/amd64", "language": "python", "license_gate": "none", "license_source": "LICENSE", "public_payload_sha256": "9ed895761e11ebcd5d3d383429d1b50b2039969eab6be3369298f32077d581dd", "release_id": "001", "repository_url": "https://github.com/aditisingh4812/simplified_tuned_range_separated", "restricted_license": false, "schema_version": 1, "science_knowledge_ablation": false, "source_license": "BSD-3-Clause", "status": "dockerhub-published", "task_path": "tasks/task_001", "title": "Repair omega-dependent range-separated PBE workflow in a real PySCF STRSF repository", "verifier_image": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64", "verifier_image_digest": "sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64", "verifier_image_id": "sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64", "verifier_image_tag": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0"}
+{"base_commit": "cbe1a0c1102b8125608841eb8ac536cfe3b09671", "domain": "computational-reaction-chemistry", "environment_image": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:0b6b41e556cd862f10e9e8c489e9b539eda27d277718e0504de440048c62d3c7", "environment_image_digest": "sha256:0b6b41e556cd862f10e9e8c489e9b539eda27d277718e0504de440048c62d3c7", "environment_image_id": "sha256:0b6b41e556cd862f10e9e8c489e9b539eda27d277718e0504de440048c62d3c7", "environment_image_tag": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0", "gpl_family": false, "image_platform": "linux/amd64", "language": "python", "license_gate": "none", "license_source": "upstream LICENSE at pinned source commit (author audit verified)", "material_license": "", "material_license_source": "", "material_restricted": false, "materials_gate": false, "materials_manifest_sha256": "", "public_payload_sha256": "624261111bb1223ce6af304c999ee84e484496f1c9b2ac90898d2e387d901fcd", "release_id": "001", "repository_url": "https://github.com/Auto-Mech/autochem", "restricted_license": false, "restricted_reason": "", "schema_version": 1, "source_license": "MIT", "status": "dockerhub-published", "task_path": "tasks/task_001", "title": "Repair an inconsistent transition-state rotor workflow", "verifier_image": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0", "verifier_image_digest": "sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0", "verifier_image_id": "sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0", "verifier_image_tag": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0"}
 {"base_commit": "f794ca5265d38906782506f1ac4e197b4e273fc4", "domain": "density-functional-theory", "environment_image": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-002:v0.1.0@sha256:13e8c428f486c685ec92b031e46f3520bac3d4e26b6bdd83fde9826a488b7761", "environment_image_digest": "sha256:13e8c428f486c685ec92b031e46f3520bac3d4e26b6bdd83fde9826a488b7761", "environment_image_id": "sha256:13e8c428f486c685ec92b031e46f3520bac3d4e26b6bdd83fde9826a488b7761", "environment_image_tag": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-002:v0.1.0", "gpl_family": false, "image_platform": "linux/amd64", "language": "python", "license_gate": "none", "license_source": "LICENSE", "public_payload_sha256": "413aae9a4fa7428016b80984ed9c66d8b5e9e3af97d9588571f64fcac7a1dc46", "release_id": "002", "repository_url": "https://github.com/aditisingh4812/simplified_tuned_range_separated", "restricted_license": false, "schema_version": 1, "science_knowledge_ablation": true, "source_license": "BSD-3-Clause", "status": "dockerhub-published", "task_path": "tasks/task_002", "title": "Repair an inconsistent reduced validation workflow", "verifier_image": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-002:v0.1.0@sha256:a863b98f920e1e9b1aede8bd0460d039628aab19a413cce3dad7417028e1bd1a", "verifier_image_digest": "sha256:a863b98f920e1e9b1aede8bd0460d039628aab19a413cce3dad7417028e1bd1a", "verifier_image_id": "sha256:a863b98f920e1e9b1aede8bd0460d039628aab19a413cce3dad7417028e1bd1a", "verifier_image_tag": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-002:v0.1.0"}
 {"base_commit": "4d8048b51ee0cc00caa4854844c28b7f892fd908", "domain": "atomistic-machine-learning-data", "environment_image": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-003:v0.1.0@sha256:f161a1e658abd71c5c3abda59b61814a2c5bf140a4826b5598a917d62a258f4e", "environment_image_digest": "sha256:f161a1e658abd71c5c3abda59b61814a2c5bf140a4826b5598a917d62a258f4e", "environment_image_id": "sha256:f161a1e658abd71c5c3abda59b61814a2c5bf140a4826b5598a917d62a258f4e", "environment_image_tag": "docker.io/kevinxulearning/swe-bench-science-environment-python-task-003:v0.1.0", "gpl_family": true, "image_platform": "linux/amd64", "language": "python", "license_gate": "gpl-family", "license_source": "LICENSE", "public_payload_sha256": "1f00289f07104c48c4894348dc606860079c7e5647e213570bb613d52603927e", "release_id": "003", "repository_url": "https://github.com/deepmodeling/dpdata", "restricted_license": true, "schema_version": 1, "science_knowledge_ablation": true, "source_license": "LGPL-3.0-family", "status": "dockerhub-published", "task_path": "tasks/task_003", "title": "Repair extxyz stress and virial tensor-label preservation", "verifier_image": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-003:v0.1.0@sha256:4d2e13b726b99a0c9d302ba913cdaa727c49fdc403b65b68918ec70c1713a790", "verifier_image_digest": "sha256:4d2e13b726b99a0c9d302ba913cdaa727c49fdc403b65b68918ec70c1713a790", "verifier_image_id": "sha256:4d2e13b726b99a0c9d302ba913cdaa727c49fdc403b65b68918ec70c1713a790", "verifier_image_tag": "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-003:v0.1.0"}
 {"base_commit": "0f45761e8ef6ffc093f649b0145788577a62ed33", "domain": "rna-seq-genomic-interval-semantics", "environment_image": "docker.io/kevinxulearning/swe-bench-science-environment-c-task-004:v0.1.0@sha256:290695ddb1d2a160142311b604ccb61ff5d51d9500b6d218c56b43cbb94d9620", "environment_image_digest": "sha256:290695ddb1d2a160142311b604ccb61ff5d51d9500b6d218c56b43cbb94d9620", "environment_image_id": "sha256:290695ddb1d2a160142311b604ccb61ff5d51d9500b6d218c56b43cbb94d9620", "environment_image_tag": "docker.io/kevinxulearning/swe-bench-science-environment-c-task-004:v0.1.0", "gpl_family": false, "image_platform": "linux/amd64", "language": "c++", "license_gate": "none", "license_source": "LICENSE", "public_payload_sha256": "8cc8d65610e3d79b5906805b086ce8163a7f03a586bfcbc62badbe9c62d8c555", "release_id": "004", "repository_url": "https://github.com/arq5x/bedtools2", "restricted_license": false, "schema_version": 1, "science_knowledge_ablation": true, "source_license": "MIT", "status": "dockerhub-published", "task_path": "tasks/task_004", "title": "Repair split-read exon-overlap filtering", "verifier_image": "docker.io/kevinxulearning/swe-bench-science-verifier-c-task-004:v0.1.0@sha256:2553def59dc88f1886ad281da1d561600560a6f8391f177d33bb9f4975e7cc4f", "verifier_image_digest": "sha256:2553def59dc88f1886ad281da1d561600560a6f8391f177d33bb9f4975e7cc4f", "verifier_image_id": "sha256:2553def59dc88f1886ad281da1d561600560a6f8391f177d33bb9f4975e7cc4f", "verifier_image_tag": "docker.io/kevinxulearning/swe-bench-science-verifier-c-task-004:v0.1.0"}

tasks/task_001/environment/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=python:3.11-slim
+ARG BASE_IMAGE=python:3.12-slim
 FROM ${BASE_IMAGE}
 
 ARG DEBIAN_MIRROR=https://mirrors.tuna.tsinghua.edu.cn/debian
@@ -24,6 +24,8 @@ RUN apt-get -o Acquire::Retries=10 update \
         curl \
         git \
         ripgrep \
+        libxext6 \
+        libxrender1 \
     && rm -rf /var/lib/apt/lists/*
 
 RUN if [ "${INSTALL_FORTRAN}" = "1" ]; then \

tasks/task_001/instruction.md

@@ -1,13 +1,17 @@
-The included repository implements a tuned range-separated DFT workflow. The
-baseline calculation runs, and the
-workflow reports a tuned range-separation parameter. However, the final
-calculation does not show the kind of change one would expect after a successful
-tuning step.
+# Repair an inconsistent transition-state rotor workflow
 
-Inspect the source code and public reproduction workflow, and repair
-the implementation so that the reported tuning step has a real effect on the
-final scientific calculation. Do not hard-code a single molecule, a single
-parameter value, or a fixed expected output.
+A reaction-chemistry workflow is using the supplied historical source to
+prepare transition-state graph data for a high-energy reaction campaign. The
+inputs are valid graph representations and the workflow reaches the graph
+analysis stage, but equivalent presentations of one transition state do not
+give a consistent rotor observation.
 
-The task is about diagnosing why the reproduced tuned workflow still behaves too
-much like its untuned baseline.
+Run the offline reproduction, read the supplied scientific method material, and
+inspect the complete source snapshot. Repair the implementation so that the
+documented transition-state rotor behavior is representation-invariant for all
+supported valid molecular graphs and reaction-edge arrangements. Do not
+hard-code the supplied graph, atom keys, a fixed output, or the public report.
+
+Work only under this task directory.  The evaluation environment is offline;
+all required source files and dependencies are provided.  Do not modify the
+scientific fixtures, the public reproduction, or generated reports.

tasks/task_001/metadata.json

@@ -1,13 +1,13 @@
 {
   "task_id": "001",
   "science_knowledge_ablation": false,
-  "title": "Repair omega-dependent range-separated PBE workflow in a real PySCF STRSF repository",
-  "domain": "density-functional-theory",
+  "title": "Repair an inconsistent transition-state rotor workflow",
+  "domain": "computational-reaction-chemistry",
   "language": "python",
-  "source_repository": "https://github.com/aditisingh4812/simplified_tuned_range_separated",
-  "source_commit": "f794ca5265d38906782506f1ac4e197b4e273fc4",
-  "source_license": "BSD-3-Clause",
-  "license_source": "LICENSE",
+  "source_repository": "https://github.com/Auto-Mech/autochem",
+  "source_commit": "cbe1a0c1102b8125608841eb8ac536cfe3b09671",
+  "source_license": "MIT",
+  "license_source": "upstream LICENSE at pinned source commit (author audit verified)",
   "gpl_family": false,
   "restricted_license": false,
   "license_gate": "none",
@@ -19,5 +19,5 @@
   "material_licenses": [],
   "materials_provenance": "",
   "restricted_reason": "",
-  "public_payload_sha256": "9ed895761e11ebcd5d3d383429d1b50b2039969eab6be3369298f32077d581dd"
+  "public_payload_sha256": "624261111bb1223ce6af304c999ee84e484496f1c9b2ac90898d2e387d901fcd"
 }

tasks/task_001/task.toml

@@ -3,21 +3,21 @@ artifacts = ["/logs/artifacts/model.patch"]
 
 [task]
 name = "openmoss/swe-bench-science-001"
-description = "Repair omega-dependent range-separated PBE workflow in a real PySCF STRSF repository"
+description = "Repair an inconsistent transition-state rotor workflow"
 authors = []
 keywords = ["scientific-computing", "software-engineering"]
 
 [metadata]
 task_id = "001"
 language = "python"
-base_commit_hash = "f794ca5265d38906782506f1ac4e197b4e273fc4"
+base_commit_hash = "cbe1a0c1102b8125608841eb8ac536cfe3b09671"
 
 [agent]
 network_mode = "no-network"
 timeout_sec = 5400.0
 
 [environment]
-docker_image = "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:ba7befeae4625761cb107ea17ff7962f0d7fd45a71eb25cf75369dc93da8f9d3"
+docker_image = "docker.io/kevinxulearning/swe-bench-science-environment-python-task-001:v0.1.0@sha256:0b6b41e556cd862f10e9e8c489e9b539eda27d277718e0504de440048c62d3c7"
 os = "linux"
 allow_internet = false
 cpus = 2
@@ -31,7 +31,7 @@ environment_mode = "separate"
 timeout_sec = 1800.0
 
 [verifier.environment]
-docker_image = "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64"
+docker_image = "docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0"
 os = "linux"
 allow_internet = false
 cpus = 2

tasks/task_001/tests/Dockerfile

@@ -1,5 +1,7 @@
-ARG VERIFIER_IMAGE=docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:3cf671ebe637b067246b05330c0c95c82bd0e1445fafaa3de1b6c1c4d1d7cd64
+ARG VERIFIER_IMAGE=docker.io/kevinxulearning/swe-bench-science-verifier-python-task-001:v0.1.0@sha256:7af68a1c44876615b0f961abd18e7fb656dae5f117820a35ff8e1d10faa430b0
 FROM ${VERIFIER_IMAGE}
 
+# Keep the runtime entrypoint in sync with the task bundle. This is also
+# mounted by the task compose override when a prebuilt verifier image is used.
 COPY grader.py /tests/grader.py
 RUN chmod 755 /tests/grader.py