Datasets:

vinhable
/

sgp_vinhable

Error code:   DatasetGenerationError
Exception:    CastError
Message:      Couldn't cast
c3_provenance: struct<assistant_content_tokens: int64, component_root: string, content_sha256: string, context_sha2 (... 199 chars omitted)
  child 0, assistant_content_tokens: int64
  child 1, component_root: string
  child 2, content_sha256: string
  child 3, context_sha256_recomputed: string
  child 4, input_lane: string
  child 5, input_line: int64
  child 6, input_split: string
  child 7, rendered_tokens: int64
  child 8, risk_flags: list<item: string>
      child 0, item: string
  child 9, scanner_disposition: string
  child 10, source_sampling_weight: double
completion_start: int64
cut_point: int64
dataset_version: string
eval_run_id: string
instance_id: null
messages: list<item: struct<content: string, loss: bool, role: string>>
  child 0, item: struct<content: string, loss: bool, role: string>
      child 0, content: string
      child 1, loss: bool
      child 2, role: string
model_uri: string
observation_format: string
origin: string
protocol: string
sample_id: string
sample_phase: string
snapshot_hash: null
source: string
split: string
task_group: string
trajectory_id: string
alternatives_in_run_context: int64
context_sha256: string
generated_line: int64
generated_source: string
license: string
questions_sha256: string
reference_index: int64
reference_scoring_line: int64
reference_self_score: null
reference_sha256: string
scoring_source: string
selection_signals: struct<correctness: string, historical_self_score_only: bool, rank_without_self_score: list<item: in (... 127 chars omitted)
  child 0, correctness: string
  child 1, historical_self_score_only: bool
  child 2, rank_without_self_score: list<item: int64>
      child 0, item: int64
  child 3, risk_flags: struct<environment_error: int64>
      child 0, environment_error: int64
  child 4, scanner_disposition: string
  child 5, self_score_used: bool
  child 6, targeted_inspect: bool
status: string
task_coordinate: string
to
{'c3_provenance': {'assistant_content_tokens': Value('int64'), 'component_root': Value('string'), 'content_sha256': Value('string'), 'context_sha256_recomputed': Value('string'), 'input_lane': Value('string'), 'input_line': Value('int64'), 'input_split': Value('string'), 'rendered_tokens': Value('int64'), 'risk_flags': List(Value('string')), 'scanner_disposition': Value('string'), 'source_sampling_weight': Value('float64')}, 'completion_start': Value('int64'), 'cut_point': Value('int64'), 'dataset_version': Value('string'), 'eval_run_id': Value('string'), 'instance_id': Value('null'), 'messages': List({'content': Value('string'), 'loss': Value('bool'), 'role': Value('string')}), 'model_uri': Value('string'), 'observation_format': Value('string'), 'origin': Value('string'), 'protocol': Value('string'), 'sample_id': Value('string'), 'sample_phase': Value('string'), 'snapshot_hash': Value('null'), 'source': Value('string'), 'split': Value('string'), 'task_group': Value('string'), 'trajectory_id': 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 1827, 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 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              c3_provenance: struct<assistant_content_tokens: int64, component_root: string, content_sha256: string, context_sha2 (... 199 chars omitted)
                child 0, assistant_content_tokens: int64
                child 1, component_root: string
                child 2, content_sha256: string
                child 3, context_sha256_recomputed: string
                child 4, input_lane: string
                child 5, input_line: int64
                child 6, input_split: string
                child 7, rendered_tokens: int64
                child 8, risk_flags: list<item: string>
                    child 0, item: string
                child 9, scanner_disposition: string
                child 10, source_sampling_weight: double
              completion_start: int64
              cut_point: int64
              dataset_version: string
              eval_run_id: string
              instance_id: null
              messages: list<item: struct<content: string, loss: bool, role: string>>
                child 0, item: struct<content: string, loss: bool, role: string>
                    child 0, content: string
                    child 1, loss: bool
                    child 2, role: string
              model_uri: string
              observation_format: string
              origin: string
              protocol: string
              sample_id: string
              sample_phase: string
              snapshot_hash: null
              source: string
              split: string
              task_group: string
              trajectory_id: string
              alternatives_in_run_context: int64
              context_sha256: string
              generated_line: int64
              generated_source: string
              license: string
              questions_sha256: string
              reference_index: int64
              reference_scoring_line: int64
              reference_self_score: null
              reference_sha256: string
              scoring_source: string
              selection_signals: struct<correctness: string, historical_self_score_only: bool, rank_without_self_score: list<item: in (... 127 chars omitted)
                child 0, correctness: string
                child 1, historical_self_score_only: bool
                child 2, rank_without_self_score: list<item: int64>
                    child 0, item: int64
                child 3, risk_flags: struct<environment_error: int64>
                    child 0, environment_error: int64
                child 4, scanner_disposition: string
                child 5, self_score_used: bool
                child 6, targeted_inspect: bool
              status: string
              task_coordinate: string
              to
              {'c3_provenance': {'assistant_content_tokens': Value('int64'), 'component_root': Value('string'), 'content_sha256': Value('string'), 'context_sha256_recomputed': Value('string'), 'input_lane': Value('string'), 'input_line': Value('int64'), 'input_split': Value('string'), 'rendered_tokens': Value('int64'), 'risk_flags': List(Value('string')), 'scanner_disposition': Value('string'), 'source_sampling_weight': Value('float64')}, 'completion_start': Value('int64'), 'cut_point': Value('int64'), 'dataset_version': Value('string'), 'eval_run_id': Value('string'), 'instance_id': Value('null'), 'messages': List({'content': Value('string'), 'loss': Value('bool'), 'role': Value('string')}), 'model_uri': Value('string'), 'observation_format': Value('string'), 'origin': Value('string'), 'protocol': Value('string'), 'sample_id': Value('string'), 'sample_phase': Value('string'), 'snapshot_hash': Value('null'), 'source': Value('string'), 'split': Value('string'), 'task_group': Value('string'), 'trajectory_id': 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 1694, 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 1880, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

c3_provenance dict	completion_start int64	cut_point int64	dataset_version string	eval_run_id string	instance_id null	messages list	model_uri string	observation_format string	origin string	protocol string	sample_id string	sample_phase string	snapshot_hash null	source string	split string	task_group string	trajectory_id string
{ "assistant_content_tokens": 3392, "component_root": "context:4c130582861b718e56658e3afde23695c54120cf8da8040d7a6da2a1bfed49a7", "content_sha256": "df1e512538e3f0b1f9d70a1c678086fd499223698a0352b6a51a966b75efe59d", "context_sha256_recomputed": "4c130582861b718e56658e3afde23695c54120cf8da8040d7a6da2a1bfed49a7",...	6	2	sft-c2-teacher-first-phase-aware-v2	32652a57-a3be-4389-b8e3-f4380d971719	null	[ { "content": "You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.\nYour response must contain exactly ONE bash code block with ONE command (or commands connected with && or \|\|).\n\nInclude a THOUGHT section before your command where you explain your rea...	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00024-of-00060.parquet:4744:2	cold	null	mini-coder	train	coord:mini-coder/data/train-00024-of-00060.parquet:4744	0003699be57d223282cd114eda5ac2d77af54620ec3db894b44ea1b06f337278
{ "assistant_content_tokens": 1127, "component_root": "context:4e013c2f7569b867188ddf811fd8a25bb65cb9b6ba8aa683856225e5321b71a9", "content_sha256": "c705068dc546d8561143414cf796e5bfdcc4a59a78771d97f3cbebb7033bda00", "context_sha256_recomputed": "f1dab3b00025d90b952fd1f90f1462410e7d6642cd0c35656a7400d5a83eaf92",...	18	8	sft-c2-teacher-first-phase-aware-v2	f5f477f2-f755-449a-bd2d-97886b4c339d	null	[ { "content": "You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.\nYour response must contain exactly ONE bash code block with ONE command (or commands connected with && or \|\|).\n\nInclude a THOUGHT section before your command where you explain your rea...	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00009-of-00060.parquet:429:8	at_edit	null	mini-coder	train	coord:mini-coder/data/train-00009-of-00060.parquet:429	000574d110d39874b3dea11dd4b3364395c6f9c5f88732b463d8977a7bbcb41a
{ "assistant_content_tokens": 2189, "component_root": "context:cefb7d7ba4d3b4f9b2dd35867c425f5a0f3f5d0f548cc46ef92e7ec70676f18c", "content_sha256": "35f1b7ec3c33d932ea17850fb29f2e48ce72d3e582001733105ab0e4e11e340a", "context_sha256_recomputed": "cefb7d7ba4d3b4f9b2dd35867c425f5a0f3f5d0f548cc46ef92e7ec70676f18c",...	6	2	sft-c2-teacher-first-phase-aware-v2	3387cebb-940f-4c83-934e-ef943c8d94fd	null	[ { "content": "You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.\nYour response must contain exactly ONE bash code block with ONE command (or commands connected with && or \|\|).\n\nInclude a THOUGHT section before your command where you explain your rea...	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00045-of-00060.parquet:1597:2	cold	null	mini-coder	train	coord:mini-coder/data/train-00045-of-00060.parquet:1597	0010be03971dfc136e8f3a1accd4f977853a7e28f3cce0b21fd230d35a61bb4f
{ "assistant_content_tokens": 1459, "component_root": "context:af6b7f9e4903c45346063c577cecc9fc24cc3e4401e3975f5973debb842a11e7", "content_sha256": "9ed38729effb3379b92e6d4591c7d733c6dbaf1c347e1ee790a45810f7559247", "context_sha256_recomputed": "af6b7f9e4903c45346063c577cecc9fc24cc3e4401e3975f5973debb842a11e7",...	18	8	sft-c2-teacher-first-phase-aware-v2	cbe5dc78-aafd-49df-911c-2351b514c7f0	null	[ { "content": "You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.\nYour response must contain exactly ONE bash code block with ONE command (or commands connected with && or \|\|).\n\nInclude a THOUGHT section before your command where you explain your rea...	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00019-of-00060.parquet:996:8	at_edit	null	mini-coder	train	coord:mini-coder/data/train-00019-of-00060.parquet:996	0013209231e7b46750b847820d02b6266eb4c03e8c5fbe605a8e3c7384e1f082
{"assistant_content_tokens":3723,"component_root":"context:79433b193c58a990f8c55cd5952444d60c3b3545b(...TRUNCATED)	14	6	sft-c2-teacher-first-phase-aware-v2	a2b25a76-5cbd-488d-9ea8-0a27fdaa36d4	null	[{"content":"You are a helpful assistant that can interact multiple times with a computer shell to s(...TRUNCATED)	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00047-of-00060.parquet:1355:6	at_edit	null	mini-coder	train	coord:mini-coder/data/train-00047-of-00060.parquet:1355	00167b563f66c8167265798b0b8cd4ecd92918ec8911e5f303fac2ea84673ca0
{"assistant_content_tokens":4419,"component_root":"context:afc16e6a59e4b84fe34f41d9a4d27511be7c69a44(...TRUNCATED)	12	5	sft-c2-teacher-first-phase-aware-v2	f12d30b0-1227-41ed-a623-f7ef8817c2c3	null	[{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED)	z-ai/glm-5.2	openhands	teacher	legacy	open-swe-traces/data/train-00000.parquet:1912:5	pre_edit	null	open-swe-traces	train	coord:open-swe-traces/data/train-00000.parquet:1912	001a27c8e7998055c34fa942f80336616471932e90121566b1ebdf17d1d91aa2
{"assistant_content_tokens":991,"component_root":"context:32739953b23311ecd27a85f53bfd5ec62fd05c1731(...TRUNCATED)	4	1	sft-c2-teacher-first-phase-aware-v2	ab6a8988-0be4-4113-b26f-4ba54189c7a7	null	[{"content":"You are a helpful assistant that can interact multiple times with a computer shell to s(...TRUNCATED)	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00048-of-00060.parquet:1277:1	cold	null	mini-coder	train	coord:mini-coder/data/train-00048-of-00060.parquet:1277	001f45defa13701c4ef2815b356f488b0188f6ba23455c5b40f805574fd114ab
{"assistant_content_tokens":1736,"component_root":"context:3d1b3af7d47dabbd3ac74c20df6557b8b4685bc1d(...TRUNCATED)	6	2	sft-c2-teacher-first-phase-aware-v2	68b9d350-e87b-434e-a16c-be49b10b53a3	null	[{"content":"You are a helpful assistant that can interact multiple times with a computer shell to s(...TRUNCATED)	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00050-of-00060.parquet:2214:2	cold	null	mini-coder	train	coord:mini-coder/data/train-00050-of-00060.parquet:2214	0025d2d558b1e6be62147ea96a046fb08ef1dcbcc2cca7c6be1aca46e8b2b0aa
{"assistant_content_tokens":1610,"component_root":"context:ed733d110aab7441fb82de582479b8d6c99da9c06(...TRUNCATED)	8	3	sft-c2-teacher-first-phase-aware-v2	d1c23b68-d417-477a-8fb0-0031e276e9c1	null	[{"content":"You are a helpful assistant that can interact multiple times with a computer shell to s(...TRUNCATED)	z-ai/glm-5.2	returncode	teacher	legacy	mini-coder/data/train-00049-of-00060.parquet:1423:3	pre_edit	null	mini-coder	train	coord:mini-coder/data/train-00049-of-00060.parquet:1423	00286b61c3cb58bb6dc82e89b1bf1021b4fad727418642bf4b00391718ec069a
{"assistant_content_tokens":689,"component_root":"context:1881dde17aa1caf34df40ea4c4a431733229c8bf52(...TRUNCATED)	4	1	sft-c2-teacher-first-phase-aware-v2	dcea6e03-fdd8-4caf-960b-94660a1fdcd7	null	[{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED)	z-ai/glm-5.2	openhands	teacher	legacy	open-swe-traces/data/train-00006.parquet:2136:1	cold	null	open-swe-traces	train	coord:open-swe-traces/data/train-00006.parquet:2136	0031aab488d0920c06e2961628360adca6a041061f9b8f709671050dc4050db4

End of preview.

SGP Vinhable

Teacher-only supervised fine-tuning trajectories for multi-turn software-engineering tool use. This is a derived research dataset and is not an official Albedo release.

Files

messages.jsonl: 14,318 unique trajectories in messages format.
train-plan.jsonl: 15,748 ordered epoch slots; use line to select a 1-indexed row from messages.jsonl and preserve intentional repetition.
summary.json: counts, provenance distributions, hashes, and sanitization summary.
excluded-sensitive-patterns.jsonl: identifiers of conservatively omitted rows; it never includes the matched secret-like values.

Loss masking

Each message has a boolean loss field. Prefix/system/user messages use false. Assistant continuation messages at or after completion_start use true. A compliant trainer should calculate loss only where loss=true.

Dataset summary

Epoch slots: 15,748
Unique trajectories: 14,318
Unique task components: 12,586
Supervised assistant-content tokens per epoch: 25,820,308
Origins: teacher only
Maximum repetition: 2
Secret-like fixture rows excluded before publication: 8

The sampling plan is authoritative. Do not concatenate the plan and message pool or apply repetition a second time.

The default Hugging Face configuration exposes messages.jsonl. Load the ordered sampling plan explicitly with the sampling_plan configuration.

Limitations

The data contains simulated command observations and historical model trajectories. Offline quality flags are diagnostics, not correctness labels. Dataset-level scoring does not guarantee that a checkpoint trained on this corpus will outperform a King model; checkpoint rollouts still require held-out evaluation.

Downloads last month: 33