agent_instruction_md stringlengths 221 2.41k | base_commit stringclasses 37
values | docker_image stringclasses 37
values | fail_to_pass listlengths 1 1.77k | framework stringclasses 5
values | generator_id stringclasses 1
value | gold_patch stringclasses 1
value | instruction_kind stringclasses 1
value | interface_spec stringclasses 1
value | introduction_patch stringlengths 165 46.1k | introduction_patch_role stringclasses 1
value | language stringclasses 1
value | metadata dict | observer_kind stringclasses 1
value | pass_to_pass listlengths 6 5.43k | problem_statement_md stringclasses 1
value | repo stringclasses 37
values | requirements_md stringclasses 1
value | sandbox_repo_dir stringclasses 1
value | schema_id stringclasses 1
value | setup_cmd stringclasses 1
value | source_kind stringclasses 1
value | swebench_pro_parser_script stringclasses 1
value | swebench_pro_pre_test_cmd stringclasses 1
value | swebench_pro_run_script stringclasses 1
value | swebench_pro_selected_test_files_to_run listlengths 0 0 | task_id stringlengths 39 57 | task_variant stringclasses 1
value | test_command stringclasses 18
values | test_framework stringclasses 5
values | test_parser stringclasses 5
values | test_timeout_sec int64 600 600 | test_universe_ref null | verifier_test_patch stringclasses 1
value | workspace_base_ref stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
## IndexAlias single-index operations do not execute on the underlying index
### Problem
`IndexAlias` should behave like the wrapped index when it references exactly one concrete index. Currently, operations that require a single target index can return without actually being delegated to that underlying index.
This... | f2876b5e34763ac2d28a75a87dce5f2ff4a64d42 | swebench/swesmith.x86_64.blevesearch_1776_bleve.f2876b5e | [
"TestIndexAliasEmpty",
"TestIndexAliasSingle"
] | go | bugpilot | direct_markdown | --- a/index_alias_impl.go
+++ b/index_alias_impl.go
@@ -493,7 +493,7 @@
}
err := i.isAliasToSingleIndex()
- if err != nil {
+ if err == nil {
return err
}
| bug | {
"dataset_id": "VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522",
"default_test_command": "go test -v ./...",
"initial_f2p": [
"TestIndexAliasSingle",
"TestIndexAliasEmpty"
],
"initial_p2p": [
"TestBuilder",
"TestIndexAliasClosed",
"TestTokenFrequency",
"TestTokenMapLoadF... | command | [
"ExampleDocumentMapping_AddFieldMapping",
"ExampleDocumentMapping_AddFieldMappingsAt",
"ExampleDocumentMapping_AddSubDocumentMapping",
"TestAllFieldWithDifferentTermVectorsEnabled",
"TestAnalysisBug328",
"TestAnonymousStructFieldWithJSONStructTagEmptString",
"TestAnonymousStructFields",
"TestApostroph... | blevesearch/bleve | /testbed | swe_fix_training_row_v1 | swesmith | [] | blevesearch__bleve.f2876b5e.lm_modify__2qskhbqo | generated_bug_intro | go test -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## Unsupported LATERAL expressions are accepted for dialects that do not support them\n\n### Descri(...TRUNCATED) | 21b6e6d1cb1befe839044764d8ad6b1c6f0b5ef4 | swebench/swesmith.x86_64.doug-martin_1776_goqu.21b6e6d1 | [
"TestExpressionSQLGenerator",
"TestExpressionSQLGenerator/TestGenerate_LateralExpression"
] | go | bugpilot | direct_markdown | "--- a/sqlgen/expression_sql_generator.go\n+++ b/sqlgen/expression_sql_generator.go\n@@ -62,7 +62,7 (...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAliasExpressionSuite","TestAliasExpressionSuite/TestAliased","TestAliasExpressionSuite/TestAll(...TRUNCATED) | doug-martin/goqu | /testbed | swe_fix_training_row_v1 | swesmith | [] | doug-martin__goqu.21b6e6d1.lm_modify__bqj7gtgl | generated_bug_intro | go test -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## Cluster manager builds incomplete peer list from shared state\n\nThe cluster manager can reconst(...TRUNCATED) | 88a1da66df9527aa97e8781dfc91cb9feb08125c | swebench/swesmith.x86_64.krotik_1776_eliasdb.88a1da66 | ["TestCluster2MemberCluster","TestClusterBuilding","TestClusterEjection","TestClusterQuery","TestClu(...TRUNCATED) | go | bugpilot | direct_markdown | "--- a/cluster/manager/manager.go\n+++ b/cluster/manager/manager.go\n@@ -627,7 +627,7 @@\n \tsiMembe(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAddRemoveIndexEntry","TestAddRemoveIndexHashEntry","TestAddressTableClusterLoc","TestAddressTa(...TRUNCATED) | krotik/eliasdb | /testbed | swe_fix_training_row_v1 | swesmith | [] | krotik__eliasdb.88a1da66.lm_modify__suewim34 | generated_bug_intro | go test -count=1 -p 1 -parallel 1 -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## Delete `WHERE` clauses combine multiple predicates too broadly\n\nWhen building delete statement(...TRUNCATED) | 21b6e6d1cb1befe839044764d8ad6b1c6f0b5ef4 | swebench/swesmith.x86_64.doug-martin_1776_goqu.21b6e6d1 | ["TestDeleteClausesSuite","TestDeleteClausesSuite/TestClearWhere","TestDeleteClausesSuite/TestWhere"(...TRUNCATED) | go | bugpilot | direct_markdown | "--- a/exp/delete_clauses.go\n+++ b/exp/delete_clauses.go\n@@ -97,7 +97,7 @@\n \t}\n \tret := dc.clo(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAliasExpressionSuite","TestAliasExpressionSuite/TestAliased","TestAliasExpressionSuite/TestAll(...TRUNCATED) | doug-martin/goqu | /testbed | swe_fix_training_row_v1 | swesmith | [] | doug-martin__goqu.21b6e6d1.lm_modify__3ck2kfcq | generated_bug_intro | go test -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## Configuration parser misparses binary-size numbers and array values\n\nThe configuration parser (...TRUNCATED) | 2ee2e24cb10924adb699ecb68b89e8ce2523ea75 | swebench/swesmith.x86_64.nats-io_1776_nats-server.2ee2e24c | ["TestConvenientNumbers","TestJSONParseCompat","TestJSONParseCompat/JSON_includes","TestJSONParseCom(...TRUNCATED) | go | bugpilot | direct_markdown | "--- a/conf/parse.go\n+++ b/conf/parse.go\n@@ -272,9 +272,6 @@\n \tcase itemError:\n \t\treturn fmt.(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestArrayOfMaps","TestArrays","TestBadBinaryStringEndingAfterOneHexChar","TestBadBinaryStringEndin(...TRUNCATED) | nats-io/nats-server | /testbed | swe_fix_training_row_v1 | swesmith | [] | nats-io__nats-server.2ee2e24c.lm_modify__1sri44mv | generated_bug_intro | go test -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## Storage location decoding can return the wrong record ID\n\nStorage locations created by the uti(...TRUNCATED) | 88a1da66df9527aa97e8781dfc91cb9feb08125c | swebench/swesmith.x86_64.krotik_1776_eliasdb.88a1da66 | ["TestCachedDiskStorageManager","TestClusterWithPhysicalStorage","TestFreeLogicalSlotManager","TestF(...TRUNCATED) | go | bugpilot | direct_markdown | "--- a/storage/util/location.go\n+++ b/storage/util/location.go\n@@ -70,7 +70,7 @@\n LocationRecord (...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAddRemoveIndexEntry","TestAddRemoveIndexHashEntry","TestAddressTableClusterLoc","TestAddressTa(...TRUNCATED) | krotik/eliasdb | /testbed | swe_fix_training_row_v1 | swesmith | [] | krotik__eliasdb.88a1da66.lm_modify__e7f2bbzr | generated_bug_intro | go test -count=1 -p 1 -parallel 1 -v ./... | go | go | 600 | null | origin/main | ||||||||||
"# Augmented interval tree queries can miss matches in the right subtree\n\n## Description\n\nThe `a(...TRUNCATED) | 18d77378f834b72b39509b12f70f3f9915c56884 | swebench/swesmith.x86_64.workiva_1776_go-datastructures.18d77378 | [
"TestAddRebalanceInOrder"
] | go | bugpilot | direct_markdown | "--- a/augmentedtree/atree.go\n+++ b/augmentedtree/atree.go\n@@ -69,7 +69,7 @@\n \t\tfn(n)\n \t}\n \(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["Example","ExampleFloatingFibonacciHeap_DecreaseKey","ExampleFloatingFibonacciHeap_Delete","Example(...TRUNCATED) | workiva/go-datastructures | /testbed | swe_fix_training_row_v1 | swesmith | [] | Workiva__go-datastructures.18d77378.lm_modify__b9olhh9g | generated_bug_intro | go test -v -skip '^(TestMaxTime|TestPoll|TestRingPoll)$' ./... | go | go | 600 | null | origin/main | ||||||||||
"## Console starts on the wrong graph endpoint\n\nThe interactive console initializes with an unexpe(...TRUNCATED) | 88a1da66df9527aa97e8781dfc91cb9feb08125c | swebench/swesmith.x86_64.krotik_1776_eliasdb.88a1da66 | [
"TestEQLConsole",
"TestGraphCommands",
"TestGraphQLConsole"
] | go | bugpilot | direct_markdown | "--- a/console/console.go\n+++ b/console/console.go\n@@ -72,7 +72,7 @@\n \t\tcmdMap[CommandExport] =(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAddRemoveIndexEntry","TestAddRemoveIndexHashEntry","TestAddressTableClusterLoc","TestAddressTa(...TRUNCATED) | krotik/eliasdb | /testbed | swe_fix_training_row_v1 | swesmith | [] | krotik__eliasdb.88a1da66.lm_modify__vuclkv90 | generated_bug_intro | go test -count=1 -p 1 -parallel 1 -v ./... | go | go | 600 | null | origin/main | ||||||||||
"## yfast trie constructor ignores caller-provided initialization value\n\n### Affected component\n\(...TRUNCATED) | 18d77378f834b72b39509b12f70f3f9915c56884 | swebench/swesmith.x86_64.workiva_1776_go-datastructures.18d77378 | [
"TestTrieSimpleInsert"
] | go | bugpilot | direct_markdown | "--- a/trie/yfast/yfast.go\n+++ b/trie/yfast/yfast.go\n@@ -277,6 +277,6 @@\n // complexity of all lo(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["Example","ExampleFloatingFibonacciHeap_DecreaseKey","ExampleFloatingFibonacciHeap_Delete","Example(...TRUNCATED) | workiva/go-datastructures | /testbed | swe_fix_training_row_v1 | swesmith | [] | Workiva__go-datastructures.18d77378.lm_modify__6t019wmo | generated_bug_intro | go test -v -skip '^(TestMaxTime|TestPoll|TestRingPoll)$' ./... | go | go | 600 | null | origin/main | ||||||||||
"## GraphQL conditional directives are not honored\n\nGraphQL queries using standard conditional dir(...TRUNCATED) | 88a1da66df9527aa97e8781dfc91cb9feb08125c | swebench/swesmith.x86_64.krotik_1776_eliasdb.88a1da66 | [
"TestDirectives"
] | go | bugpilot | direct_markdown | "--- a/graphql/interpreter/selectionset.go\n+++ b/graphql/interpreter/selectionset.go\n@@ -659,14 +6(...TRUNCATED) | bug | {"dataset_id":"VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522","default_test_command":(...TRUNCATED) | command | ["TestAddRemoveIndexEntry","TestAddRemoveIndexHashEntry","TestAddressTableClusterLoc","TestAddressTa(...TRUNCATED) | krotik/eliasdb | /testbed | swe_fix_training_row_v1 | swesmith | [] | krotik__eliasdb.88a1da66.lm_modify__bqjc56wg | generated_bug_intro | go test -count=1 -p 1 -parallel 1 -v ./... | go | go | 600 | null | origin/main |
End of preview. Expand in Data Studio
BugPilot cleaned reliable-universe 500-row subset
This dataset is a deterministic 500-row subset of VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522, selected from the cleaned reliable-universe rerun to match the canonical SWE-Smith target-pool language/category distribution as closely as available rows allow.
Result
- Source cleaned dataset:
VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k-cleaned-20260522 - Source cleaned dataset sha:
d16d4807800853a4c912f4fad136222676a5acfb - Source rows considered: 646
- Candidate rows inside canonical target distribution: 646
- Selected rows in
train: 500 - Dropped unknown-repo rows: 0
Target quotas after scaling the canonical 893 usable target-pool slots to 500 rows:
| language | target | selected |
|---|---|---|
| go | 214 | 217 |
| js | 127 | 132 |
| python | 144 | 151 |
| ts | 15 | 0 |
| category | target | selected |
|---|---|---|
| backend_service_platform | 123 | 122 |
| full_stack_web_product | 113 | 118 |
| infra_systems_integration | 108 | 113 |
| rich_client_ui | 156 | 147 |
Availability caveat: the cleaned accepted rows did not contain enough rows for ts/rich_client_ui and were short one row for go/backend_service_platform. The downsampler reports these shortages and redistributes the missing rows deterministically across available cells.
Shortages:
{
"go/backend_service_platform": 1,
"ts/rich_client_ui": 15
}
Redistributed overfills:
{
"go/full_stack_web_product": 2,
"go/infra_systems_integration": 2,
"js/full_stack_web_product": 1,
"js/rich_client_ui": 4,
"python/full_stack_web_product": 2,
"python/infra_systems_integration": 3,
"python/rich_client_ui": 2
}
Reproduction
Repository/code:
- Repo:
VmaxAI/demiurge - Branch:
codex/pr1303-downsample-cleaned-500 - Downsampling commit:
8b8ac6bbb - Parent cleaner commit:
de20b671b27b443235727664755f2992b2fceba1 - Script:
projects/swe-turing/scripts/downsample_cleaned_swesmith_to_target_distribution.py - Source distribution TSV:
projects/swe-turing/defaults/runtime/datasets/swesmith/canonical-target-repos-1000-size-matched.tsv - Selection seed:
canonical-swesmith-language-category-500-v1
Exact command used to create this subset from the completed cleaner run:
cd /home/augustine_vmax_ai_com/codex-runs/pr1303-validation/pr1303-downsample-500
/home/augustine_vmax_ai_com/codex-runs/pr1303-validation/.venv/bin/python \
projects/swe-turing/scripts/downsample_cleaned_swesmith_to_target_distribution.py \
--input-jsonl /home/augustine_vmax_ai_com/runs/bugpilot-clean-reliable-universe-pr1303-latest-de20b671b-20260523T051728Z/data/accepted.jsonl \
--output-jsonl /tmp/bugpilot-cleaned-500-selected.jsonl \
--report-json /tmp/bugpilot-cleaned-500-report.json \
--target-count 500
The source cleaner run was produced by projects/swe-turing/scripts/clean_patch_derived_with_reliable_universe.py at commit de20b671b27b443235727664755f2992b2fceba1 with:
- Task dataset:
VmaxRL/bugpilot-bugintro-lm-modify-gpt55-1k, splittrain - Reliable-tests dataset:
VmaxRL/SWEUniverse-SizeMatched-45-ReliableTests-20260522, splittrain - Timeout:
600seconds - Snapshot interval:
15seconds - Max concurrency:
96 - Sandbox resources:
cpus=1,memory_mb=4096,storage_mb=10240,network=True
Docker applies CPU, memory, and network settings. Docker ignores storage_mb in this backend and emits a warning because storage quotas are storage-driver dependent.
Clean observations in the source cleaner run were loaded from reliable SWE-universe stable_passing and stable_failing, not collected by running Docker. Patched observations were executed in Docker; NOP was graded against the patched observation and oracle against the clean reliable-universe observation.
The reliable-tests dataset was schema-upgraded in place to add execution and execution_key at Hugging Face sha 9ee9638503dcf8c8147bc547d62078704c754bf3. That upgrade did not change test commands, parsers, or resources.
Files
trainsplit: selected 500 cleaned training rowsmetadata/downsample_report.json: target quotas, selected counts, shortages, overfills, selected repo counts, and source distribution metadata
- Downloads last month
- 21