dataset stringclasses 1
value | domain stringclasses 1
value | example_id stringlengths 17 28 | group_id stringlengths 17 28 | input_text stringlengths 8.51k 461k | language stringclasses 1
value | measurement_descriptions dict | measurements dict | metadata dict | prompt_components dict | reference_outputs listlengths 0 0 | source_document_id stringlengths 1 6 | source_text stringclasses 1
value | split stringclasses 1
value | target_descriptions dict | targets dict | variant stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
GraphArch | architecture_regression | GraphArch:Hiaml_654 | GraphArch:Hiaml:654 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_343[FLOAT, 64x3x3x3]
%onnx::Conv_344[FLOAT, 64]
%onnx::Conv_346[FLOAT, 64x64x3x3]
%onnx::Conv_349[FLOAT, 64x64x1x3]
%onnx::Conv_352[FLOAT, 64x64x3x1]
%onnx::Conv_355[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 217724416,
"params": 2207050,
"val_accuracy": 92.97
} | {
"arch_str": "654",
"identifier": "Hiaml_654",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "654"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 654 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 217724416,
"val_accuracy": 92.97
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_2772 | GraphArch:Hiaml:2772 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_377[FLOAT, 64x3x3x3]
%onnx::Conv_378[FLOAT, 64]
%onnx::Conv_380[FLOAT, 64x64x3x3]
%onnx::Conv_383[FLOAT, 64x64x1x1]
%onnx::Conv_386[FLOAT, 64x64x1x1]
%onnx::Conv_389[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 215791104,
"params": 2167114,
"val_accuracy": 92.4
} | {
"arch_str": "2772",
"identifier": "Hiaml_2772",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "2772"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 2772 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 215791104,
"val_accuracy": 92.4
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_1380 | GraphArch:Hiaml:1380 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_436[FLOAT, 64x3x3x3]
%onnx::Conv_437[FLOAT, 64]
%onnx::Conv_439[FLOAT, 64x64x1x1]
%onnx::Conv_442[FLOAT, 64x64x1x3]
%onnx::Conv_445[FLOAT, 64x64x3x1]
%onnx::Conv_448[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 181515776,
"params": 1996362,
"val_accuracy": 91.89
} | {
"arch_str": "1380",
"identifier": "Hiaml_1380",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "1380"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 1380 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 181515776,
"val_accuracy": 91.89
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_1617 | GraphArch:Hiaml:1617 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_407[FLOAT, 64x3x3x3]
%onnx::Conv_408[FLOAT, 64]
%onnx::Conv_410[FLOAT, 64x64x1x1]
%onnx::Conv_413[FLOAT, 64x64x3x3]
%onnx::Conv_416[FLOAT, 64x64x3x3]
%onnx::Conv_419[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 253703680,
"params": 2781770,
"val_accuracy": 92.64
} | {
"arch_str": "1617",
"identifier": "Hiaml_1617",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "1617"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 1617 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 253703680,
"val_accuracy": 92.64
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_301 | GraphArch:Hiaml:301 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_453[FLOAT, 64x3x3x3]
%onnx::Conv_454[FLOAT, 64]
%onnx::Conv_456[FLOAT, 64x64x1x3]
%onnx::Conv_459[FLOAT, 64x64x3x1]
%onnx::Conv_462[FLOAT, 64x64x1x3]
%onnx::Conv_465[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 241350144,
"params": 2832970,
"val_accuracy": 92.37
} | {
"arch_str": "301",
"identifier": "Hiaml_301",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "301"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 301 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 241350144,
"val_accuracy": 92.37
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_893 | GraphArch:Hiaml:893 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_347[FLOAT, 64x3x3x3]
%onnx::Conv_348[FLOAT, 64]
%onnx::Conv_350[FLOAT, 64x64x3x3]
%onnx::Conv_353[FLOAT, 64x64x1x3]
%onnx::Conv_356[FLOAT, 64x64x3x1]
%onnx::Conv_359[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 227259904,
"params": 2157130,
"val_accuracy": 92.38
} | {
"arch_str": "893",
"identifier": "Hiaml_893",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "893"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 893 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 227259904,
"val_accuracy": 92.38
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_855 | GraphArch:Hiaml:855 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_390[FLOAT, 64x3x3x3]
%onnx::Conv_391[FLOAT, 64]
%onnx::Conv_393[FLOAT, 64x64x3x3]
%onnx::Conv_396[FLOAT, 64x64x1x1]
%onnx::Conv_399[FLOAT, 64x64x3x3]
%onnx::Conv_402[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 203011584,
"params": 2586442,
"val_accuracy": 91.92
} | {
"arch_str": "855",
"identifier": "Hiaml_855",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "855"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 855 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 203011584,
"val_accuracy": 91.92
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_176 | GraphArch:Hiaml:176 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_377[FLOAT, 64x3x3x3]
%onnx::Conv_378[FLOAT, 64]
%onnx::Conv_380[FLOAT, 64x64x1x1]
%onnx::Conv_383[FLOAT, 64x64x1x3]
%onnx::Conv_386[FLOAT, 64x64x3x1]
%onnx::Conv_389[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 206583296,
"params": 2571978,
"val_accuracy": 92.57
} | {
"arch_str": "176",
"identifier": "Hiaml_176",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "176"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 176 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 206583296,
"val_accuracy": 92.57
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_600 | GraphArch:Hiaml:600 | graph torch_jit (
%input.1[FLOAT, 1x3x32x32]
%out_net.classifier.weight[FLOAT, 10x256]
%out_net.classifier.bias[FLOAT, 10]
%onnx::Conv_393[FLOAT, 64x3x3x3]
%onnx::Conv_394[FLOAT, 64]
%onnx::Conv_396[FLOAT, 64x64x1x3]
%onnx::Conv_399[FLOAT, 64x64x3x1]
%onnx::Conv_402[FLOAT, 64x64x1x3]
%onnx::Conv_405[F... | graph | {
"flops": "Floating-point operation count for the architecture.",
"params": "Number of trainable parameters in the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 218215936,
"params": 3227338,
"val_accuracy": 92.14
} | {
"arch_str": "600",
"identifier": "Hiaml_600",
"params_retained_not_default_target": true,
"search_space": "Hiaml",
"source_dataset": "grapharch_regression",
"source_target_metric": "val_accuracy",
"uid": "600"
} | {
"input_format": "serialized neural-network graph text",
"input_to_evaluate_column": "input_text",
"problem_context": "Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text.",
"search_space": "Hiaml"
} | [] | 600 | Predict neural architecture validation accuracy and compute cost from serialized neural-network graph text. | train | {
"flops": "Floating-point operation count for the architecture.",
"val_accuracy": "Validation accuracy reported for the neural architecture."
} | {
"flops": 218215936,
"val_accuracy": 92.14
} | full |
GraphArch | architecture_regression | GraphArch:Hiaml_1937 | GraphArch:Hiaml:1937 | "graph torch_jit (\n %input.1[FLOAT, 1x3x32x32]\n %out_net.classifier.weight[FLOAT, 10x256]\n %ou(...TRUNCATED) | graph | {"flops":"Floating-point operation count for the architecture.","params":"Number of trainable parame(...TRUNCATED) | {
"flops": 217888256,
"params": 1972682,
"val_accuracy": 92.91
} | {"arch_str":"1937","identifier":"Hiaml_1937","params_retained_not_default_target":true,"search_space(...TRUNCATED) | {"input_format":"serialized neural-network graph text","input_to_evaluate_column":"input_text","prob(...TRUNCATED) | [] | 1937 | "Predict neural architecture validation accuracy and compute cost from serialized neural-network gra(...TRUNCATED) | train | {"flops":"Floating-point operation count for the architecture.","val_accuracy":"Validation accuracy (...TRUNCATED) | {
"flops": 217888256,
"val_accuracy": 92.91
} | full |
End of preview. Expand in Data Studio
GraphArch
This dataset is the MO-RELISH graph architecture regression collection. Each row contains serialized neural-network graph text and execution/benchmark measurements from neural architecture search spaces.
Configs And Splits
| Config | Train | Validation | Test |
|---|---|---|---|
full |
475,257 | 10,000 | 10,000 |
subset_10k |
10,000 | 1,000 | 1,000 |
The full config keeps all strict-clean rows, with 10,000 validation rows and
10,000 test rows sampled deterministically and stratified by search space. The
subset_10k config is a deterministic 10,000/1,000/1,000 sample from the full
splits, also stratified by search space.
Columns
Input columns:
source_text: task-level context.input_text: serialized neural-network graph text.reference_outputs: empty list; this is a regression benchmark.prompt_components.problem_context: same task context assource_text.prompt_components.input_to_evaluate_column: points toinput_text.prompt_components.input_format: serialized neural-network graph text.prompt_components.search_space: NAS search-space name.
Prediction targets:
targets.val_accuracy: Validation accuracy reported for the neural architecture.targets.flops: Floating-point operation count for the architecture.
Retained measurements:
measurements.val_accuracy: Validation accuracy reported for the neural architecture.measurements.flops: Floating-point operation count for the architecture.measurements.params: Number of trainable parameters in the architecture.
params is retained for analysis but is not a default prediction target because
it is usually deterministic from the architecture.
The original zero-cost-proxy metadata is not included in rows because it may leak target values such as FLOPs, parameter count, or validation accuracy.
Loading
from datasets import load_dataset
full_ds = load_dataset("Samsoup/GraphArch", "full")
small_ds = load_dataset("Samsoup/GraphArch", "subset_10k")
- Downloads last month
- 9