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
dataset: string
base_dir: string
format: string
records: int64
unique_ids: int64
unique_candidate_code_hashes: int64
unique_gt_images: int64
hard_error_count: int64
warning_count: int64
anchor_checks: struct<total: int64, not_found_in_code: int64, line_range_or_alignment_warnings: int64>
child 0, total: int64
child 1, not_found_in_code: int64
child 2, line_range_or_alignment_warnings: int64
character_lengths: struct<input: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: d (... 213 chars omitted)
child 0, input: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
child 1, target: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
child 2, combined: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
target_error_count: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
category_counts: struct<attribute_content_error: int64, layout_relation_error: int64, entity_error: int64, style_rend (... 19 chars omitted)
child 0, attribute_content_error: int64
child 1, layout_relation_error: int64
child 2, entity_error: int64
child 3, style_rendering_error: int64
hard_errors: list<item: null>
child 0, item: null
warnings: list<item: string>
child 0, item: string
id: 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
metadata: struct<record_id: string, sample_id: string, source_model: string, code_type: string, figure_type: s (... 91 chars omitted)
child 0, record_id: string
child 1, sample_id: string
child 2, source_model: string
child 3, code_type: string
child 4, figure_type: string
child 5, candidate_code_file: string
child 6, gt_image_sha256: string
child 7, candidate_code_sha256: string
images: list<item: string>
child 0, item: string
to
{'id': Value('string'), 'messages': List({'role': Value('string'), 'content': Value('string')}), 'images': List(Value('string')), 'metadata': {'record_id': Value('string'), 'sample_id': Value('string'), 'source_model': Value('string'), 'code_type': Value('string'), 'figure_type': Value('string'), 'candidate_code_file': Value('string'), 'gt_image_sha256': Value('string'), 'candidate_code_sha256': 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
dataset: string
base_dir: string
format: string
records: int64
unique_ids: int64
unique_candidate_code_hashes: int64
unique_gt_images: int64
hard_error_count: int64
warning_count: int64
anchor_checks: struct<total: int64, not_found_in_code: int64, line_range_or_alignment_warnings: int64>
child 0, total: int64
child 1, not_found_in_code: int64
child 2, line_range_or_alignment_warnings: int64
character_lengths: struct<input: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: d (... 213 chars omitted)
child 0, input: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
child 1, target: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
child 2, combined: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
target_error_count: struct<min: int64, p50: int64, p90: int64, p95: int64, p99: int64, max: int64, mean: double>
child 0, min: int64
child 1, p50: int64
child 2, p90: int64
child 3, p95: int64
child 4, p99: int64
child 5, max: int64
child 6, mean: double
category_counts: struct<attribute_content_error: int64, layout_relation_error: int64, entity_error: int64, style_rend (... 19 chars omitted)
child 0, attribute_content_error: int64
child 1, layout_relation_error: int64
child 2, entity_error: int64
child 3, style_rendering_error: int64
hard_errors: list<item: null>
child 0, item: null
warnings: list<item: string>
child 0, item: string
id: 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
metadata: struct<record_id: string, sample_id: string, source_model: string, code_type: string, figure_type: s (... 91 chars omitted)
child 0, record_id: string
child 1, sample_id: string
child 2, source_model: string
child 3, code_type: string
child 4, figure_type: string
child 5, candidate_code_file: string
child 6, gt_image_sha256: string
child 7, candidate_code_sha256: string
images: list<item: string>
child 0, item: string
to
{'id': Value('string'), 'messages': List({'role': Value('string'), 'content': Value('string')}), 'images': List(Value('string')), 'metadata': {'record_id': Value('string'), 'sample_id': Value('string'), 'source_model': Value('string'), 'code_type': Value('string'), 'figure_type': Value('string'), 'candidate_code_file': Value('string'), 'gt_image_sha256': Value('string'), 'candidate_code_sha256': 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 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 | messages list | images list | metadata dict |
|---|---|---|---|
Claude_Sonnet_4_5::145996c5f8fa415fa7e78cf43146da23 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/39cfb749bf65f68a66b200b86e632737e79513f61d975b2860b92d61b9dd069c.png"
] | {
"record_id": "Claude_Sonnet_4_5::145996c5f8fa415fa7e78cf43146da23",
"sample_id": "145996c5f8fa415fa7e78cf43146da23",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "bar-chart",
"candidate_code_file": "candidate_code/ffeb83e72d5da933d8cfbb3f4238a36517fb02cb61b48618072984905e49... |
Claude_Sonnet_4_5::14a577c30da8439bb1aad142191ec27d | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/0ec93025054959a02a55e6cce07df44860ccde7fb14d55886cfa92155aad0ec0.png"
] | {
"record_id": "Claude_Sonnet_4_5::14a577c30da8439bb1aad142191ec27d",
"sample_id": "14a577c30da8439bb1aad142191ec27d",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "bar-chart",
"candidate_code_file": "candidate_code/08dc9534c8e0dda4dc7087b04079302161e6f53f0bf3f4ed8acbcde12735... |
Claude_Sonnet_4_5::19bc317e37c74f4c8ecb959749cadbc1 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/0fc0ca5e69ba22951f2df490f195b85693d968495e1bb1430dce29ea0afcc36a.png"
] | {
"record_id": "Claude_Sonnet_4_5::19bc317e37c74f4c8ecb959749cadbc1",
"sample_id": "19bc317e37c74f4c8ecb959749cadbc1",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "function-related",
"candidate_code_file": "candidate_code/b3805675fd1a4a904d9b3a30faccbc4238390c7f4eed58187deb0... |
Claude_Sonnet_4_5::1e178b1520014ae8b3308773960f4167 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/ff13d05cc62e19843295a19800eecd9e6f0421d0b05c80b6e9725d2318f486ff.png"
] | {
"record_id": "Claude_Sonnet_4_5::1e178b1520014ae8b3308773960f4167",
"sample_id": "1e178b1520014ae8b3308773960f4167",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "pie-chart",
"candidate_code_file": "candidate_code/207f89c0503033d29c35dbb85580f2d978bf40b45099251fd0f8d39bb2ca... |
Claude_Sonnet_4_5::2dacb2b606b744cc96402ce428ec037e | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/18b7315e2e7e7cb2cfc8d28b0f45ac45a61de4ba3ff20aa8a03a8751680d0d25.png"
] | {
"record_id": "Claude_Sonnet_4_5::2dacb2b606b744cc96402ce428ec037e",
"sample_id": "2dacb2b606b744cc96402ce428ec037e",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "line-graph",
"candidate_code_file": "candidate_code/ffe1022083ced20cc272fc2ef75e7d9dcb259e642190268e900b42ed483... |
Claude_Sonnet_4_5::2df40e34104e488096b4d41fa38ecf97 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/87bc3fa5db77a03a86f2fbce2005081ef42d6b75c27b071a7727af42bd07022d.png"
] | {
"record_id": "Claude_Sonnet_4_5::2df40e34104e488096b4d41fa38ecf97",
"sample_id": "2df40e34104e488096b4d41fa38ecf97",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "line-graph",
"candidate_code_file": "candidate_code/a03a79896fbc9cacf9bfe1ca814699c1e584a4d00228e84ed5fbc64497b... |
Claude_Sonnet_4_5::2f26492d5250418cb946f52e6213cc6a | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/0e5b85f19bcd18a76b427a952f2ff413854af3af2f1de790abfbeacb405bc9fe.png"
] | {
"record_id": "Claude_Sonnet_4_5::2f26492d5250418cb946f52e6213cc6a",
"sample_id": "2f26492d5250418cb946f52e6213cc6a",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "radar-chart",
"candidate_code_file": "candidate_code/5eeac1c8437164e6f38b384c88d9a9f78c3773b6eaca30ce78724a0af8... |
Claude_Sonnet_4_5::2f742517504c426cbd2eeb76a8262cf6 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/b7105b3fa7bfd80cf56204cdc3ce7731759f0a1e53d54a0fb3d27d7098de4e9b.png"
] | {
"record_id": "Claude_Sonnet_4_5::2f742517504c426cbd2eeb76a8262cf6",
"sample_id": "2f742517504c426cbd2eeb76a8262cf6",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "line-graph",
"candidate_code_file": "candidate_code/7f20c2f527fd6d6d62d61b9f60b3fd663dc7154825d7e2545dd21e06ed5... |
Claude_Sonnet_4_5::361bae861cb640fa8f5532d04f605b92 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/39e04599cd751d6fdf4f6507d6b406b5c03f583d64b0595a39c6c0f5c2c37025.png"
] | {
"record_id": "Claude_Sonnet_4_5::361bae861cb640fa8f5532d04f605b92",
"sample_id": "361bae861cb640fa8f5532d04f605b92",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "bar-chart",
"candidate_code_file": "candidate_code/e0550ece340866ec2e8eaf529755f29081d554a654bd1710fc6b47e47ac7... |
Claude_Sonnet_4_5::3839dec2738246b998e05e7c79eafaf9 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/63ce78ba2a650f6f5a94130a44748c21a3c9f607aff0e652fa167a621c7262fe.png"
] | {
"record_id": "Claude_Sonnet_4_5::3839dec2738246b998e05e7c79eafaf9",
"sample_id": "3839dec2738246b998e05e7c79eafaf9",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "other-figures",
"candidate_code_file": "candidate_code/286d34bb71c07325d8edc3c09df1e74c2109222c1a49eaeab93dfbff... |
Claude_Sonnet_4_5::3bb98a7fe50c4715b0a941a0524191e2 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/55452c563aec745c95cf01382a9df779823d45867807dea0ff497408f3b03818.png"
] | {
"record_id": "Claude_Sonnet_4_5::3bb98a7fe50c4715b0a941a0524191e2",
"sample_id": "3bb98a7fe50c4715b0a941a0524191e2",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "analytical-geometry",
"candidate_code_file": "candidate_code/75201b27d86623a5c2fb685772019058ee774853c5a064061d... |
Claude_Sonnet_4_5::3c5606f2c84d48bd9573d843ffe68c95 | [
{
"role": "user",
"content": "<image>\nYou are an auditor for Image-to-Code visual factuality.\n\nYou are given a ground-truth image, figure_type, code_type, and\ncandidate_code_lines. Each code line contains an authoritative 1-based line\nnumber `l_n` and exact source text `t`.\n\nCompare the visual facts ... | [
"gt_images/d7120151338462cf280fe6742c0b8fdfa617efae1bbf929ac8106cd9c218d520.png"
] | {
"record_id": "Claude_Sonnet_4_5::3c5606f2c84d48bd9573d843ffe68c95",
"sample_id": "3c5606f2c84d48bd9573d843ffe68c95",
"source_model": "Claude_Sonnet_4_5",
"code_type": "html-css",
"figure_type": "pie-chart",
"candidate_code_file": "candidate_code/b93640381f362f7ed6cf62c1d04993038c7085f486eb958324a45683e861... |
End of preview.
i2c_hallu_7k
This dataset supports supervised fine-tuning for image-to-code visual factuality auditing. Each example provides a ground-truth figure image and numbered candidate source-code lines. The target is a strict JSON list of visual inconsistencies, including the responsible code span when applicable.
Splits
splits_gt/train.jsonl: 7,014 examplessplits_gt/val.jsonl: 373 examplesgt_images/: 804 ground-truth PNG images referenced by the splitscandidate_code/: 7,387 source files retained for provenance and dataset validation
Each JSONL record uses the ms-swift multimodal schema: messages, images, and metadata. Image paths are relative to the repository root. The candidate code needed by the model is embedded in the user prompt; the files in candidate_code/ provide the corresponding source artifact referenced by metadata.candidate_code_file.
metadata.json contains dataset statistics. The validation and split reports document dataset integrity and the group-disjoint train/validation split.
Example usage
Run from the dataset root so the relative image paths resolve:
swift sft \
--model <MODEL_PATH_OR_ID> \
--dataset splits_gt/train.jsonl \
--val_dataset splits_gt/val.jsonl \
--template qwen3_vl
- Downloads last month
- -