SwathiManikya commited on
Commit
6e2e088
·
verified ·
1 Parent(s): 3119764

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +77 -49
README.md CHANGED
@@ -1,75 +1,103 @@
1
- ---
2
- license: apache-2.0
3
- ---
4
 
5
- # HLO Feature Dataset for Deep Learning Workloads
6
 
7
- [![Dataset](https://img.shields.io/badge/HuggingFace-Dataset-yellow)](https://huggingface.co/datasets/your-username/hlo-feature-dataset)
8
 
9
  ## Dataset Summary
10
- The **HLO Feature Dataset** provides High-Level Optimizer (HLO) graph features extracted from various deep learning model training runs. Each sample represents a unique configuration of neural network training, paired with graph-based features suitable for machine learning tasks such as:
11
 
12
- - Runtime and resource prediction
13
- - AI workload optimization in HPC environments
14
- - Graph neural network (GNN) research on compiler-level representations
15
- - Scheduling and efficiency analysis for GPU-based training
16
 
17
- The dataset includes metadata for each training run and corresponding `.npz` files containing serialized HLO graph features.
18
 
19
  ---
20
 
21
- ## Supported Tasks and Benchmarks
22
- - 🕒 **Runtime Prediction**
23
- - 📊 **Resource Utilization Estimation**
24
- - ⚙️ **Graph-Based Neural Architecture Analysis**
25
- - 🚀 **HPC & GPU Scheduling Optimization**
26
-
27
- This dataset can be used to train models that predict execution time, memory usage, or optimize scheduling strategies based on compiler graph features.
28
 
29
  ---
30
 
31
  ## Dataset Structure
32
 
33
- Each sample consists of:
34
- - **Metadata**: Model configuration, hardware specs, and performance metrics (from `dataset-new.csv`).
35
- - **HLO Graph Features**: Stored in `.npz` files containing:
36
- - `node_opcode`: Operation codes
37
- - `node_feat`: Node feature matrix
38
- - `edge_index`: Graph topology
39
- - `node_config_ids`: Config identifiers
40
- - `node_splits`: Graph partition info
41
 
42
  ---
43
 
44
- ## Features
45
- | Feature | Type | Description |
46
- |--------------------|----------|--------------------------------------|
47
- | name | string | Model name |
48
- | optimizer | string | Optimizer used |
49
- | batch, epochs | int | Training parameters |
50
- | learn_rate | float | Learning rate |
51
- | gpu_name | string | GPU model |
52
- | fit_time | float | Total training time |
53
- | npz_path | string | Path to HLO feature `.npz` file |
54
- | ... | ... | Additional GPU & utilization metrics |
 
 
 
 
 
 
 
 
 
 
 
 
 
55
 
56
- The `.npz` file contains graph data relevant for GNNs or ML models.
 
 
 
 
57
 
58
  ---
59
 
60
- ## Usage Example
 
61
 
62
  ```python
63
- from datasets import load_dataset
64
- import numpy as np
 
 
 
 
 
 
 
 
 
65
 
66
- # Load metadata
67
- dataset = load_dataset("your-username/hlo-feature-dataset")
68
- sample = dataset['train'][0]
69
 
70
- # Load HLO features
71
- npz_file = sample['npz_path']
72
- data = np.load(npz_file)
 
 
 
 
 
 
 
 
 
 
 
 
73
 
74
- node_features = data['node_feat']
75
- edges = data['edge_index']
 
 
 
 
1
 
2
+ # HLO Feature Dataset for Deep Learning Resource Estimation
3
 
4
+ [![Dataset](https://img.shields.io/badge/HuggingFace-Dataset-blue)](https://huggingface.co/datasets/your-username/hlo-feature-dataset)
5
 
6
  ## Dataset Summary
7
+ The **HLO Feature Dataset** is a collection of compiler-level graph features (HLO graphs) extracted from deep learning training workloads. Alongside detailed metadata (model configs, GPU stats), this dataset enables machine learning approaches for:
8
 
9
+ - ⏱️ **Training Time Prediction**
10
+ - 📉 **Resource Consumption Estimation**
11
+ - **HPC and GPU Scheduling Optimization**
12
+ - 🧩 **Graph-based Neural Architecture Analysis**
13
 
14
+ This dataset is ideal for experimenting with regression models (e.g., XGBoost) and Graph Neural Networks (GNNs) using compiler features.
15
 
16
  ---
17
 
18
+ ## Supported Tasks
19
+ - **⚙️ Runtime & Resource Prediction**: Predict training time (`fit_time`) based on HLO features.
20
+ - **📊 ML for Systems Optimization**: Use tabular + graph data for AI workload management.
21
+ - **🔗 Graph Representation Learning**: Apply GNNs on HLO graphs (`node_feat`, `edge_index`).
 
 
 
22
 
23
  ---
24
 
25
  ## Dataset Structure
26
 
27
+ Each entry includes:
28
+ - **Metadata**: From `dataset-new.csv` (model, optimizer, GPU specs, timing metrics, etc.)
29
+ - **HLO Graph Features**: `.npz` files containing:
30
+ - `node_opcode`, `node_feat`, `edge_index`, `node_config_ids`, `node_splits`
 
 
 
 
31
 
32
  ---
33
 
34
+ ## Usage Example
35
+
36
+ This example demonstrates how to load metadata, preprocess features, and train an XGBoost model to predict training time (`fit_time`), as shown in the Colab notebook.
37
+
38
+ ```python
39
+ import pandas as pd
40
+ import numpy as np
41
+ from sklearn.model_selection import train_test_split
42
+ from sklearn.metrics import mean_squared_error
43
+ from xgboost import XGBRegressor
44
+
45
+ # Load metadata CSV
46
+ df = pd.read_csv('dataset-new.csv')
47
+
48
+ # Example feature selection (drop non-numeric/categorical handling needed)
49
+ X = df[['batch', 'epochs', 'learn_rate', 'gpu_core_count', 'gpu_memory_size']]
50
+ y = df['fit_time']
51
+
52
+ # Train-test split
53
+ X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
54
+
55
+ # Initialize XGBoost Regressor
56
+ xgb_model = XGBRegressor(n_estimators=100, learning_rate=0.1, max_depth=6, random_state=42)
57
+ xgb_model.fit(X_train, y_train)
58
 
59
+ # Evaluate
60
+ preds = xgb_model.predict(X_test)
61
+ rmse = mean_squared_error(y_test, preds, squared=False)
62
+ print(f"RMSE: {rmse}")
63
+ ```
64
 
65
  ---
66
 
67
+ ### Loading HLO Graph Features
68
+ For graph-based ML tasks, load the `.npz` files:
69
 
70
  ```python
71
+ npz_file = df.iloc[0]['npz_path']
72
+ graph_data = np.load(npz_file)
73
+
74
+ node_features = graph_data['node_feat']
75
+ edges = graph_data['edge_index']
76
+
77
+ print("Node Feature Shape:", node_features.shape)
78
+ print("Edge Index Shape:", edges.shape)
79
+ ```
80
+
81
+ ---
82
 
83
+ ## Citation
84
+ If you use this dataset, please cite:
 
85
 
86
+ ```
87
+ @misc{hlofeatures2025,
88
+ title={HLO Feature Dataset for AI Resource Estimation},
89
+ author={Your Name},
90
+ year={2025},
91
+ url={https://huggingface.co/datasets/your-username/hlo-feature-dataset}
92
+ }
93
+ ```
94
+
95
+ ---
96
+
97
+ ## License
98
+ Specify your license here (e.g., MIT, Apache-2.0).
99
+
100
+ ---
101
 
102
+ ## Contributions
103
+ Open to contributions! Feel free to suggest improvements or share your models trained on this dataset.