Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: CastError
Message: Couldn't cast
task: string
dataset_dir: string
output_dir: string
seed: int64
max_pairs_per_trace: int64
max_user_chars: int64
balance_positions: bool
shuffle_output: bool
splits: struct<train: struct<split: string, input_path: string, output_path: string, source_pointwise_rows: (... 2432 chars omitted)
child 0, train: struct<split: string, input_path: string, output_path: string, source_pointwise_rows: int64, num_tra (... 734 chars omitted)
child 0, split: string
child 1, input_path: string
child 2, output_path: string
child 3, source_pointwise_rows: int64
child 4, num_traces: int64
child 5, num_pairwise_rows: int64
child 6, label_counts: struct<B: int64, A: int64>
child 0, B: int64
child 1, A: int64
child 7, repair_success_pattern: struct<A=False|B=True: int64, A=True|B=False: int64>
child 0, A=False|B=True: int64
child 1, A=True|B=False: int64
child 8, skipped_traces: struct<all_positive: int64, all_negative_or_unknown: int64>
child 0, all_positive: int64
child 1, all_negative_or_unknown: int64
child 9, traces_with_pairs: int64
child 10, traces_without_pairs: int64
child 11, total_possible_pairs: int64
child 12, total_written_pairs: int64
child 13, max_pairs_per_trace: int64
child 14, truncated_user_prompts: int64
child 15, forbidden_prompt_hits: struct<>
child 16, extraction_sources_in_pairs: struct<input: int64>
...
child 0, repair_a: struct<row_id: string, trace_id: string, repair_id: string, candidate_text_hash: string, candidate_r (... 32 chars omitted)
child 0, row_id: string
child 1, trace_id: string
child 2, repair_id: string
child 3, candidate_text_hash: string
child 4, candidate_rank: int64
child 5, repair_type: string
child 1, repair_b: struct<row_id: string, trace_id: string, repair_id: string, candidate_text_hash: string, candidate_r (... 32 chars omitted)
child 0, row_id: string
child 1, trace_id: string
child 2, repair_id: string
child 3, candidate_text_hash: string
child 4, candidate_rank: int64
child 5, repair_type: string
child 2, positive_repair_id: string
child 3, negative_repair_id: string
child 4, source_positive_row_id: string
child 5, source_negative_row_id: string
child 6, truncated_user_prompt: bool
child 7, forbidden_prompt_hits: list<item: null>
child 0, item: null
child 8, extraction_source_positive: string
child 9, extraction_source_negative: string
messages: list<item: struct<role: string, content: string>>
child 0, item: struct<role: string, content: string>
child 0, role: string
child 1, content: string
id: string
trace_id: string
label: struct<preferred: string, repair_a_is_successful: bool, repair_b_is_successful: bool>
child 0, preferred: string
child 1, repair_a_is_successful: bool
child 2, repair_b_is_successful: bool
task_type: string
to
{'id': Value('string'), 'trace_id': Value('string'), 'task_type': Value('string'), 'messages': List({'role': Value('string'), 'content': Value('string')}), 'label': {'preferred': Value('string'), 'repair_a_is_successful': Value('bool'), 'repair_b_is_successful': Value('bool')}, 'pair_metadata': {'repair_a': {'row_id': Value('string'), 'trace_id': Value('string'), 'repair_id': Value('string'), 'candidate_text_hash': Value('string'), 'candidate_rank': Value('int64'), 'repair_type': Value('string')}, 'repair_b': {'row_id': Value('string'), 'trace_id': Value('string'), 'repair_id': Value('string'), 'candidate_text_hash': Value('string'), 'candidate_rank': Value('int64'), 'repair_type': Value('string')}, 'positive_repair_id': Value('string'), 'negative_repair_id': Value('string'), 'source_positive_row_id': Value('string'), 'source_negative_row_id': Value('string'), 'truncated_user_prompt': Value('bool'), 'forbidden_prompt_hits': List(Value('null')), 'extraction_source_positive': Value('string'), 'extraction_source_negative': Value('string')}}
because column names don't match
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1816, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
task: string
dataset_dir: string
output_dir: string
seed: int64
max_pairs_per_trace: int64
max_user_chars: int64
balance_positions: bool
shuffle_output: bool
splits: struct<train: struct<split: string, input_path: string, output_path: string, source_pointwise_rows: (... 2432 chars omitted)
child 0, train: struct<split: string, input_path: string, output_path: string, source_pointwise_rows: int64, num_tra (... 734 chars omitted)
child 0, split: string
child 1, input_path: string
child 2, output_path: string
child 3, source_pointwise_rows: int64
child 4, num_traces: int64
child 5, num_pairwise_rows: int64
child 6, label_counts: struct<B: int64, A: int64>
child 0, B: int64
child 1, A: int64
child 7, repair_success_pattern: struct<A=False|B=True: int64, A=True|B=False: int64>
child 0, A=False|B=True: int64
child 1, A=True|B=False: int64
child 8, skipped_traces: struct<all_positive: int64, all_negative_or_unknown: int64>
child 0, all_positive: int64
child 1, all_negative_or_unknown: int64
child 9, traces_with_pairs: int64
child 10, traces_without_pairs: int64
child 11, total_possible_pairs: int64
child 12, total_written_pairs: int64
child 13, max_pairs_per_trace: int64
child 14, truncated_user_prompts: int64
child 15, forbidden_prompt_hits: struct<>
child 16, extraction_sources_in_pairs: struct<input: int64>
...
child 0, repair_a: struct<row_id: string, trace_id: string, repair_id: string, candidate_text_hash: string, candidate_r (... 32 chars omitted)
child 0, row_id: string
child 1, trace_id: string
child 2, repair_id: string
child 3, candidate_text_hash: string
child 4, candidate_rank: int64
child 5, repair_type: string
child 1, repair_b: struct<row_id: string, trace_id: string, repair_id: string, candidate_text_hash: string, candidate_r (... 32 chars omitted)
child 0, row_id: string
child 1, trace_id: string
child 2, repair_id: string
child 3, candidate_text_hash: string
child 4, candidate_rank: int64
child 5, repair_type: string
child 2, positive_repair_id: string
child 3, negative_repair_id: string
child 4, source_positive_row_id: string
child 5, source_negative_row_id: string
child 6, truncated_user_prompt: bool
child 7, forbidden_prompt_hits: list<item: null>
child 0, item: null
child 8, extraction_source_positive: string
child 9, extraction_source_negative: string
messages: list<item: struct<role: string, content: string>>
child 0, item: struct<role: string, content: string>
child 0, role: string
child 1, content: string
id: string
trace_id: string
label: struct<preferred: string, repair_a_is_successful: bool, repair_b_is_successful: bool>
child 0, preferred: string
child 1, repair_a_is_successful: bool
child 2, repair_b_is_successful: bool
task_type: string
to
{'id': Value('string'), 'trace_id': Value('string'), 'task_type': Value('string'), 'messages': List({'role': Value('string'), 'content': Value('string')}), 'label': {'preferred': Value('string'), 'repair_a_is_successful': Value('bool'), 'repair_b_is_successful': Value('bool')}, 'pair_metadata': {'repair_a': {'row_id': Value('string'), 'trace_id': Value('string'), 'repair_id': Value('string'), 'candidate_text_hash': Value('string'), 'candidate_rank': Value('int64'), 'repair_type': Value('string')}, 'repair_b': {'row_id': Value('string'), 'trace_id': Value('string'), 'repair_id': Value('string'), 'candidate_text_hash': Value('string'), 'candidate_rank': Value('int64'), 'repair_type': Value('string')}, 'positive_repair_id': Value('string'), 'negative_repair_id': Value('string'), 'source_positive_row_id': Value('string'), 'source_negative_row_id': Value('string'), 'truncated_user_prompt': Value('bool'), 'forbidden_prompt_hits': List(Value('null')), 'extraction_source_positive': Value('string'), 'extraction_source_negative': Value('string')}}
because column names don't match
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1683, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1869, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
id string | trace_id string | task_type string | messages list | label dict | pair_metadata dict |
|---|---|---|---|---|---|
5a7e801b5542997cc2c47557::PAIR::455aea487e964a3b | 5a7e801b5542997cc2c47557 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7e801b5542997cc2c47557::R1",
"trace_id": "5a7e801b5542997cc2c47557",
"repair_id": "R1",
"candidate_text_hash": "a0cdb310f09e",
"candidate_rank": 1,
"repair_type": "add_answer_completeness_check"
},
"repair_b": {
"row_id": "5a7e801b5542997cc2c47557::R2",
... |
5ab545d3554299494045ef5c::PAIR::f3dc0594fc44537f | 5ab545d3554299494045ef5c | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ab545d3554299494045ef5c::R4",
"trace_id": "5ab545d3554299494045ef5c",
"repair_id": "R4",
"candidate_text_hash": "e882e0fc3604",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5ab545d3554299494045ef5c::R2",
"tr... |
5ae527085542993aec5ec165::PAIR::7a6aa643ce74fb72 | 5ae527085542993aec5ec165 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae527085542993aec5ec165::R3",
"trace_id": "5ae527085542993aec5ec165",
"repair_id": "R3",
"candidate_text_hash": "710b969b9023",
"candidate_rank": 3,
"repair_type": "add_multihop_decomposition"
},
"repair_b": {
"row_id": "5ae527085542993aec5ec165::R1",
"t... |
5ab808b85542991d3222380d::PAIR::417e39b7ca38a8c2 | 5ab808b85542991d3222380d | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ab808b85542991d3222380d::R2",
"trace_id": "5ab808b85542991d3222380d",
"repair_id": "R2",
"candidate_text_hash": "7b5bbff9b923",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5ab808b85542991d32223... |
5ac275f155429921a00aaf82::PAIR::fd91ca8d4d079b79 | 5ac275f155429921a00aaf82 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ac275f155429921a00aaf82::R5",
"trace_id": "5ac275f155429921a00aaf82",
"repair_id": "R5",
"candidate_text_hash": "66d79dad8c57",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5ac275f155429921a00aaf82::R3",
"trac... |
5ab446875542990594ba9bea::PAIR::9641a3cc522b20e9 | 5ab446875542990594ba9bea | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ab446875542990594ba9bea::R5",
"trace_id": "5ab446875542990594ba9bea",
"repair_id": "R5",
"candidate_text_hash": "b027848c6009",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5ab446875542990594ba9bea::R4",
"trac... |
5a7cff5755429907fabef09d::PAIR::b19929b752b0e4e5 | 5a7cff5755429907fabef09d | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7cff5755429907fabef09d::R4",
"trace_id": "5a7cff5755429907fabef09d",
"repair_id": "R4",
"candidate_text_hash": "53ff84a18175",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a7cff5755429907fabef09d::R5",
"tr... |
5a7b85ac5542997c3ec971ef::PAIR::9b592150837aef20 | 5a7b85ac5542997c3ec971ef | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a7b85ac5542997c3ec971ef::R4",
"trace_id": "5a7b85ac5542997c3ec971ef",
"repair_id": "R4",
"candidate_text_hash": "661cf29fab2f",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a7b85ac5542997c3ec971ef::R5",
"tr... |
5a8bc0485542995d1e6f143d::PAIR::babbbad32023bd0c | 5a8bc0485542995d1e6f143d | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a8bc0485542995d1e6f143d::R1",
"trace_id": "5a8bc0485542995d1e6f143d",
"repair_id": "R1",
"candidate_text_hash": "07dc87ef0c82",
"candidate_rank": 1,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5a8bc0485542995d1e6f143d::R2",
"trace_id... |
5ae6c0f35542991bbc97616f::PAIR::3f42c02fe7201475 | 5ae6c0f35542991bbc97616f | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ae6c0f35542991bbc97616f::R4",
"trace_id": "5ae6c0f35542991bbc97616f",
"repair_id": "R4",
"candidate_text_hash": "fc7a459038cd",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5ae6c0f35542991bbc97616f::R5",
"tr... |
5a81e04455429903bc27b9f9::PAIR::232f4c422b3b0e2a | 5a81e04455429903bc27b9f9 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a81e04455429903bc27b9f9::R4",
"trace_id": "5a81e04455429903bc27b9f9",
"repair_id": "R4",
"candidate_text_hash": "7588fd29e322",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a81e04455429903bc27b9f9::R3",
"tr... |
5a7147fd5542994082a3e758::PAIR::440b08f31c65c25c | 5a7147fd5542994082a3e758 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7147fd5542994082a3e758::R5",
"trace_id": "5a7147fd5542994082a3e758",
"repair_id": "R5",
"candidate_text_hash": "2283632c5583",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5a7147fd5542994082a3e758::R2",
"trac... |
5a7331705542991f9a20c67a::PAIR::47f515aaa6c54853 | 5a7331705542991f9a20c67a | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7331705542991f9a20c67a::R5",
"trace_id": "5a7331705542991f9a20c67a",
"repair_id": "R5",
"candidate_text_hash": "9c27a846ee94",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5a7331705542991f9a20c67a::R4",
"trac... |
5ae6050f55429929b0807a5e::PAIR::fa01c683d54d114b | 5ae6050f55429929b0807a5e | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae6050f55429929b0807a5e::R5",
"trace_id": "5ae6050f55429929b0807a5e",
"repair_id": "R5",
"candidate_text_hash": "dcfa88f3f741",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5ae6050f55429929b0807a5e::R2",
"trac... |
5a8febb555429916514e73e4::PAIR::f4fdbf6e4d994fde | 5a8febb555429916514e73e4 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a8febb555429916514e73e4::R4",
"trace_id": "5a8febb555429916514e73e4",
"repair_id": "R4",
"candidate_text_hash": "f6bef0f3894c",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a8febb555429916514e73e4::R3",
"tr... |
5ae141475542997b2ef7d185::PAIR::1381a9d79237b1df | 5ae141475542997b2ef7d185 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ae141475542997b2ef7d185::R3",
"trace_id": "5ae141475542997b2ef7d185",
"repair_id": "R3",
"candidate_text_hash": "bc2796a252b1",
"candidate_rank": 3,
"repair_type": "add_multihop_decomposition"
},
"repair_b": {
"row_id": "5ae141475542997b2ef7d185::R5",
"t... |
5ae3790f5542992e3233c415::PAIR::62f050ce672993ef | 5ae3790f5542992e3233c415 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ae3790f5542992e3233c415::R1",
"trace_id": "5ae3790f5542992e3233c415",
"repair_id": "R1",
"candidate_text_hash": "c05be0ef5030",
"candidate_rank": 1,
"repair_type": "add_reasoning_verification"
},
"repair_b": {
"row_id": "5ae3790f5542992e3233c415::R2",
"t... |
5ae33cf45542990afbd1e0ea::PAIR::59ded011128e9f82 | 5ae33cf45542990afbd1e0ea | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae33cf45542990afbd1e0ea::R1",
"trace_id": "5ae33cf45542990afbd1e0ea",
"repair_id": "R1",
"candidate_text_hash": "1759984b6e3a",
"candidate_rank": 1,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5ae33cf45542990afbd1e0ea::R3",
"trace_id... |
5a77846f5542995d83181187::PAIR::1565e16347d90e7b | 5a77846f5542995d83181187 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a77846f5542995d83181187::R1",
"trace_id": "5a77846f5542995d83181187",
"repair_id": "R1",
"candidate_text_hash": "85f22589756c",
"candidate_rank": 2,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5a77846f5542995d83181187::R4",
"trace_id... |
5a8d12ca5542994ba4e3dbe2::PAIR::20bb4341fa661cfc | 5a8d12ca5542994ba4e3dbe2 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a8d12ca5542994ba4e3dbe2::R4",
"trace_id": "5a8d12ca5542994ba4e3dbe2",
"repair_id": "R4",
"candidate_text_hash": "ec97c46fac06",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a8d12ca5542994ba4e3dbe2::R3",
"tr... |
5a8d838955429941ae14dfe5::PAIR::3de2d6ddbbda9627 | 5a8d838955429941ae14dfe5 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a8d838955429941ae14dfe5::R1",
"trace_id": "5a8d838955429941ae14dfe5",
"repair_id": "R1",
"candidate_text_hash": "5263dcf8ce80",
"candidate_rank": 1,
"repair_type": "add_reasoning_verification"
},
"repair_b": {
"row_id": "5a8d838955429941ae14dfe5::R3",
"t... |
5a7140585542994082a3e6fa::PAIR::bf46220a8414b7bc | 5a7140585542994082a3e6fa | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7140585542994082a3e6fa::R4",
"trace_id": "5a7140585542994082a3e6fa",
"repair_id": "R4",
"candidate_text_hash": "da52fd13f1a8",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a7140585542994082a3e6fa::R1",
"tr... |
5a762d53554299109176e6a7::PAIR::c9e52531ab838cac | 5a762d53554299109176e6a7 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a762d53554299109176e6a7::R3",
"trace_id": "5a762d53554299109176e6a7",
"repair_id": "R3",
"candidate_text_hash": "bf842d437d37",
"candidate_rank": 3,
"repair_type": "add_multihop_decomposition"
},
"repair_b": {
"row_id": "5a762d53554299109176e6a7::R4",
"t... |
5a81e8a0554299676cceb157::PAIR::8443a2e79159a239 | 5a81e8a0554299676cceb157 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a81e8a0554299676cceb157::R4",
"trace_id": "5a81e8a0554299676cceb157",
"repair_id": "R4",
"candidate_text_hash": "911a29ed256f",
"candidate_rank": 4,
"repair_type": "increase_retrieval_budget"
},
"repair_b": {
"row_id": "5a81e8a0554299676cceb157::R3",
"tr... |
5ab8cd3d5542991b5579f009::PAIR::18e5bc98b46cc141 | 5ab8cd3d5542991b5579f009 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ab8cd3d5542991b5579f009::R2",
"trace_id": "5ab8cd3d5542991b5579f009",
"repair_id": "R2",
"candidate_text_hash": "61155a6b9146",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5ab8cd3d5542991b5579f... |
5ae4c2b95542995dadf243ef::PAIR::e8750a75de2913b8 | 5ae4c2b95542995dadf243ef | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae4c2b95542995dadf243ef::R1",
"trace_id": "5ae4c2b95542995dadf243ef",
"repair_id": "R1",
"candidate_text_hash": "3807fee157da",
"candidate_rank": 1,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5ae4c2b95542995dadf243ef::R3",
"trace_id... |
5ae6368a5542992ae0d16284::PAIR::1fe43dd229acddd8 | 5ae6368a5542992ae0d16284 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ae6368a5542992ae0d16284::R1",
"trace_id": "5ae6368a5542992ae0d16284",
"repair_id": "R1",
"candidate_text_hash": "81dcb80b2aed",
"candidate_rank": 2,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5ae6368a5542992ae0d16284::R2",
"trace_id... |
5ac38bf355429939154137ba::PAIR::ac561140b83c9218 | 5ac38bf355429939154137ba | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5ac38bf355429939154137ba::R2",
"trace_id": "5ac38bf355429939154137ba",
"repair_id": "R2",
"candidate_text_hash": "a2fc82abf4e5",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5ac38bf35542993915413... |
5ab808b85542991d3222380d::PAIR::382c43b59a302bee | 5ab808b85542991d3222380d | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ab808b85542991d3222380d::R3",
"trace_id": "5ab808b85542991d3222380d",
"repair_id": "R3",
"candidate_text_hash": "5672e4fca4f8",
"candidate_rank": 3,
"repair_type": "add_multihop_decomposition"
},
"repair_b": {
"row_id": "5ab808b85542991d3222380d::R1",
"t... |
5ae7954b554299540e5a5620::PAIR::1cac2d41ca1f2ee4 | 5ae7954b554299540e5a5620 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae7954b554299540e5a5620::R1",
"trace_id": "5ae7954b554299540e5a5620",
"repair_id": "R1",
"candidate_text_hash": "e1294cba9d8c",
"candidate_rank": 1,
"repair_type": "add_missing_context"
},
"repair_b": {
"row_id": "5ae7954b554299540e5a5620::R5",
"trace_id... |
5a84807d554299123d8c2278::PAIR::2763e0afc586ed76 | 5a84807d554299123d8c2278 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a84807d554299123d8c2278::R2",
"trace_id": "5a84807d554299123d8c2278",
"repair_id": "R2",
"candidate_text_hash": "62710a61d50a",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5a84807d554299123d8c2... |
5a8cdcb0554299441c6b9f33::PAIR::d24ebf4880da42d5 | 5a8cdcb0554299441c6b9f33 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a8cdcb0554299441c6b9f33::R3",
"trace_id": "5a8cdcb0554299441c6b9f33",
"repair_id": "R3",
"candidate_text_hash": "461586412080",
"candidate_rank": 3,
"repair_type": "add_final_answer_check"
},
"repair_b": {
"row_id": "5a8cdcb0554299441c6b9f33::R1",
"trace... |
5a88bd1d554299206df2b357::PAIR::ec900290fc3b09db | 5a88bd1d554299206df2b357 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a88bd1d554299206df2b357::R1",
"trace_id": "5a88bd1d554299206df2b357",
"repair_id": "R1",
"candidate_text_hash": "da91ff2e5232",
"candidate_rank": 1,
"repair_type": "add_reasoning_verification"
},
"repair_b": {
"row_id": "5a88bd1d554299206df2b357::R3",
"t... |
5ac4e9ab5542996feb3fe974::PAIR::b97748413e2e2a60 | 5ac4e9ab5542996feb3fe974 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ac4e9ab5542996feb3fe974::R5",
"trace_id": "5ac4e9ab5542996feb3fe974",
"repair_id": "R5",
"candidate_text_hash": "df604fad19c7",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5ac4e9ab5542996feb3fe974::R2",
"trac... |
5a7b85ac5542997c3ec971ef::PAIR::f7edcbb9fb868987 | 5a7b85ac5542997c3ec971ef | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "B",
"repair_a_is_successful": false,
"repair_b_is_successful": true
} | {
"repair_a": {
"row_id": "5a7b85ac5542997c3ec971ef::R5",
"trace_id": "5a7b85ac5542997c3ec971ef",
"repair_id": "R5",
"candidate_text_hash": "2a3968da7a38",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5a7b85ac5542997c3ec971ef::R2",
"trac... |
5ae76d185542997ec2727641::PAIR::94045227c4ec9e4a | 5ae76d185542997ec2727641 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae76d185542997ec2727641::R5",
"trace_id": "5ae76d185542997ec2727641",
"repair_id": "R5",
"candidate_text_hash": "acbac2a8a54d",
"candidate_rank": 5,
"repair_type": "rewrite_retrieval_query"
},
"repair_b": {
"row_id": "5ae76d185542997ec2727641::R1",
"trac... |
5a8af1e255429950cd6afc28::PAIR::192c11e940e2e631 | 5a8af1e255429950cd6afc28 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a8af1e255429950cd6afc28::R2",
"trace_id": "5a8af1e255429950cd6afc28",
"repair_id": "R2",
"candidate_text_hash": "32ae9fd67dea",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5a8af1e255429950cd6af... |
5a84bbb35542991dd0999dc0::PAIR::a728d35a1ba946ce | 5a84bbb35542991dd0999dc0 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5a84bbb35542991dd0999dc0::R2",
"trace_id": "5a84bbb35542991dd0999dc0",
"repair_id": "R2",
"candidate_text_hash": "b1f07a63602a",
"candidate_rank": 2,
"repair_type": "add_missing_context_with_answer_extraction"
},
"repair_b": {
"row_id": "5a84bbb35542991dd0999... |
5ae4f96755429913cc204551::PAIR::b24c26868b7a9ce1 | 5ae4f96755429913cc204551 | pairwise_repair_ranking | [
{
"role": "system",
"content": "You are AgentBlackBox's learned pairwise counterfactual repair ranker. Given a failed agent trace and two executable candidate repairs, decide which repair is more likely to resolve the failure if applied. Output only A or B."
},
{
"role": "user",
"content": "{\n ... | {
"preferred": "A",
"repair_a_is_successful": true,
"repair_b_is_successful": false
} | {
"repair_a": {
"row_id": "5ae4f96755429913cc204551::R3",
"trace_id": "5ae4f96755429913cc204551",
"repair_id": "R3",
"candidate_text_hash": "8ef3f289f51a",
"candidate_rank": 3,
"repair_type": "add_multihop_decomposition"
},
"repair_b": {
"row_id": "5ae4f96755429913cc204551::R1",
"t... |
End of preview.
AgentBlackBox RAG Repair Outcome Dataset
This dataset contains replay-labeled repair outcome data for AgentBlackBox, a counterfactual debugging framework for language agents.
The data is built around failed RAG/document-recall agent traces, candidate repairs, counterfactual replay labels, and repair-ranking evaluation outputs.
Contents
datasets/ world_model_ranker_dataset_v2_train10k/ pointwise/ listwise/ stats.json
qwen_pairwise_sft_v2_train10k_fixed/ train.jsonl dev.jsonl test.jsonl stats.json sample_preview.json
raw_runs/ hotpotqa_validation_20260610_210103/ summary.jsonl repair_outcomes.jsonl summary.csv repair_outcomes.csv
hotpotqa_train_combined_10k_v1/ summary.jsonl repair_outcomes.jsonl
evals/ world_model_ranker_baseline_v2_train10k/ v2_package_eval_qwen_pairwise_r32_e3_logodds/ qwen3_8b_pairwise_lora_eval_v2_train10k_r32_e3_logodds/
results/ final_tables/
Dataset Summary
The dataset supports research on:
- language-agent failure analysis
- RAG repair generation
- counterfactual replay evaluation
- repair outcome modeling
- pairwise repair ranking
- world models for agent debugging
The main learning problem is:
failed trace + candidate repair(s) -> replay-verified repair outcome / repair ranking
Main Splits
The main ranker dataset is located at:
datasets/world_model_ranker_dataset_v2_train10k/
It includes pointwise and listwise formats for repair outcome modeling and repair ranking.
The Qwen pairwise SFT dataset is located at:
datasets/qwen_pairwise_sft_v2_train10k_fixed/
It includes train/dev/test JSONL files for pairwise repair ranking.
Evaluation Results
Final result tables are included in:
results/final_tables/
The main ranking table is:
results/final_tables/table_ranker_performance_wide.csv
The mixed-trace analysis table is:
results/final_tables/table_mixed_trace_analysis.csv
Related Model
The Qwen3-8B LoRA pairwise repair ranker is available at:
Eyerf/agentblackbox-qwen3-8b-rag-repair-ranker-lora
Related Code
The AgentBlackBox code repository is available at:
https://github.com/eyerf/agentblackbox
Source and License Notes
This dataset is derived from RAG experiments based on HotpotQA-style question answering and AgentBlackBox-generated repair/replay outputs.
The released files contain derived failed-trace metadata, repair candidates, replay labels, pairwise ranking data, and evaluation outputs.
Please cite the original QA benchmark sources and AgentBlackBox when using this dataset.
Limitations
- The dataset is focused on RAG/document-recall agent failures.
- Replay labels depend on the AgentBlackBox replay/evaluation setup.
- The dataset should not be interpreted as a general benchmark for all agent failures.
- The pairwise ranking data reflects the candidate repair generation and replay policy used in the experiments.
- Downloads last month
- 39