datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
fineweb_synth_dense_ocr_gradients_with_grounding_Bfineweb_synth_dense_ocr_gradients_with_grounding_Agradient-shapes-svgThis is a data set of random shapes with gradients to use for evaluating or training vision language models!
uniformat-datasetsdc-gradient-chart-xhard-v1
sdc-gradient-chart-xhard-v1
Gradient chart and heatmap — Extra Hard tier
Dataset Info
Rows: 2
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart PNG
caption
Value('string')
Description
name
Value('string')
Chart ID
Generation Parameters
{
"script_name": "run_extra_hard_zero_shot.py",
"model": "gpt-5-2",
"description": "Gradient chart and heatmap \u2014 Extra Hard tier",
"experiment_id":… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-gradient-chart-xhard-v1.sdc-avg-gradient-chart-v1
sdc-avg-gradient-chart-v1
TIOBE popularity vs average pass@1 across all 4 EsoLang-Bench tiers, mainstream + esoteric baselines
Dataset Info
Rows: 1
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart PNG
caption
Value('string')
Description
name
Value('string')
Chart ID
Generation Parameters
{
"script_name": "manual (averaged results)",
"model": "gpt-5-2",
"description": "TIOBE popularity vs… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-avg-gradient-chart-v1.sdc-gradient-chart-v1
sdc-gradient-chart-v1
Data for TIOBE rank vs pass@1 gradient chart.
Dataset Info
Rows: 13
Columns: 6
Columns
Column
Type
Description
language
Value('string')
Programming language name
tiobe_pct
Value('float64')
TIOBE percentage (log-scale x-axis)
pass_at_1
Value('float64')
Pass@1 percentage (y-axis)
pass_at_1_std
Value('float64')
Error bar (std across 3 runs)
condition
Value('string')
zero-shot or self-scaffolding
source
Value('string')… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-gradient-chart-v1.sdc-gradient-chart-medium-v1
sdc-gradient-chart-medium-v1
Gradient chart and heatmap images — Medium tier
Dataset Info
Rows: 2
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart image (PNG)
caption
Value('string')
Chart description
name
Value('string')
Chart identifier
Generation Parameters
{
"script_name": "run_medium_zero_shot.py",
"model": "gpt-5-2",
"description": "Gradient chart and heatmap images \u2014 Medium… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-gradient-chart-medium-v1.sdc-gradient-chart-hard-v1
sdc-gradient-chart-hard-v1
Gradient chart and heatmap images — Hard tier
Dataset Info
Rows: 2
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart image (PNG)
caption
Value('string')
Chart description
name
Value('string')
Chart identifier
Generation Parameters
{
"script_name": "run_hard_zero_shot.py",
"model": "gpt-5-2",
"description": "Gradient chart and heatmap images \u2014 Hard tier"… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-gradient-chart-hard-v1.sdc-final-gradient-v1
sdc-final-gradient-v1
Final gradient chart: 15 languages + 5 esoteric baselines, all 4 tiers averaged
Dataset Info
Rows: 1
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart PNG (2x resolution)
caption
Value('string')
Description
name
Value('string')
Chart ID
Generation Parameters
{
"script_name": "combined analysis",
"model": "gpt-5-2",
"description": "Final gradient chart: 15 languages +… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-final-gradient-v1.sdc-combined-gradient-v1
sdc-combined-gradient-v1
Final gradient chart — 15 mainstream + 5 esoteric, all 4 tiers
Dataset Info
Rows: 1
Columns: 3
Columns
Column
Type
Description
image
Image(mode=None, decode=True)
Chart PNG
caption
Value('string')
Description
name
Value('string')
Chart ID
Generation Parameters
{
"script_name": "final combined",
"model": "gpt-5-2",
"description": "Final gradient chart \u2014 15 mainstream + 5 esoteric, all 4 tiers"… See the full description on the dataset page: https://huggingface.co/datasets/reasoning-degeneration-dev/sdc-combined-gradient-v1.