opencl-llmperf / README.md
minhkhoi1026's picture
Upload README.md with huggingface_hub
ff365ce verified
|
raw
history blame
7.12 kB
metadata
license: apache-2.0
dataset_info:
  - config_name: benchmark-loop_with_constant
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 848082
        num_examples: 2115
      - name: validation
        num_bytes: 289853
        num_examples: 735
    download_size: 98021830
    dataset_size: 1137935
  - config_name: benchmark-loop_with_variable_v2
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 848082
        num_examples: 2115
      - name: validation
        num_bytes: 292399
        num_examples: 608
    download_size: 98057898
    dataset_size: 1140481
  - config_name: benchmark-loop_with_variable_v2_no_k
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 2157459
        num_examples: 5452
      - name: validation
        num_bytes: 242278
        num_examples: 610
    download_size: 97989419
    dataset_size: 2399737
  - config_name: benchmark-rodinia-bfs
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 848082
        num_examples: 2115
      - name: validation
        num_bytes: 289853
        num_examples: 735
      - name: org_standardized
        num_bytes: 1457649
        num_examples: 1024
    download_size: 98130091
    dataset_size: 2595584
  - config_name: benchmark-rodinia-gaussian
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 848082
        num_examples: 2115
      - name: validation
        num_bytes: 289853
        num_examples: 735
      - name: sub_size_t
        num_bytes: 43712
        num_examples: 64
    download_size: 98034559
    dataset_size: 1181647
  - config_name: github-200k
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 499610570
        num_examples: 177892
      - name: validation
        num_bytes: 499610570
        num_examples: 177892
    download_size: 97669540
    dataset_size: 999221140
  - config_name: github-350k
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 2157459
        num_examples: 5452
      - name: validation
        num_bytes: 29945960
        num_examples: 36195
    download_size: 101263343
    dataset_size: 32103419
  - config_name: github-400k
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 2157459
        num_examples: 5452
      - name: validation
        num_bytes: 37244128
        num_examples: 39975
    download_size: 101546755
    dataset_size: 39401587
  - config_name: github-600k
    features:
      - name: code
        dtype: string
      - name: gsize
        dtype: int64
      - name: lsize
        dtype: int64
      - name: execution_time
        dtype: float64
      - name: input_sizes
        dtype: string
      - name: __index_level_0__
        dtype: int64
    splits:
      - name: train
        num_bytes: 2157459
        num_examples: 5452
      - name: validation
        num_bytes: 97565395
        num_examples: 61393
    download_size: 109431969
    dataset_size: 99722854
configs:
  - config_name: benchmark-loop_with_constant
    data_files:
      - split: train
        path: benchmark-loop_with_constant/train-*
      - split: validation
        path: benchmark-loop_with_constant/validation-*
  - config_name: benchmark-loop_with_variable_v2
    data_files:
      - split: train
        path: benchmark-loop_with_variable_v2/train-*
      - split: validation
        path: benchmark-loop_with_variable_v2/validation-*
  - config_name: benchmark-loop_with_variable_v2_no_k
    data_files:
      - split: train
        path: benchmark-loop_with_variable_v2_no_k/train-*
      - split: validation
        path: benchmark-loop_with_variable_v2_no_k/validation-*
  - config_name: benchmark-rodinia-bfs
    data_files:
      - split: org
        path: benchmark-rodinia-bfs/org-*
      - split: sub_no_of_nodes
        path: benchmark-rodinia-bfs/sub_no_of_nodes-*
      - split: sub_no_of_nodes_standardized
        path: benchmark-rodinia-bfs/sub_no_of_nodes_standardized-*
      - split: org_standardized
        path: benchmark-rodinia-bfs/org_standardized-*
  - config_name: benchmark-rodinia-gaussian
    data_files:
      - split: sub_size
        path: benchmark-rodinia-gaussian/sub_size-*
      - split: org
        path: benchmark-rodinia-gaussian/org-*
      - split: sub_size_standardized_norm_seq
        path: benchmark-rodinia-gaussian/sub_size_standardized_norm_seq-*
      - split: sub_size_t_standardized
        path: benchmark-rodinia-gaussian/sub_size_t_standardized-*
      - split: sub_size_t
        path: benchmark-rodinia-gaussian/sub_size_t-*
  - config_name: github-200k
    data_files:
      - split: train
        path: github-200k/train-*
      - split: validation
        path: github-200k/validation-*
  - config_name: github-350k
    data_files:
      - split: train
        path: github-350k/train-*
      - split: validation
        path: github-350k/validation-*
  - config_name: github-400k
    data_files:
      - split: train
        path: github-400k/train-*
      - split: validation
        path: github-400k/validation-*
  - config_name: github-600k
    data_files:
      - split: train
        path: github-600k/train-*
      - split: validation
        path: github-600k/validation-*

A collection of datasets and benchmarks for large-scale Performance Modeling with LLMs.

This collection includes these dataset:

  1. github-200K: a first version dataset that contains ~1300 kernel's execution time with input size - global size correlation and imbalanced data.
  2. github-350k: a continued second version dataset that fix the imbalanced data problem.
  3. github-400k: a continued third version dataset that relax the input - global size correlation.
  4. github-600K: a fourth version dataset that contains ~6000 kernel's execution time with input size - global size correlation and imbalanced data.
  5. benchmark-[]: benchmarks for LLMs performance on Perfomance Modeling task.