{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"#

Google Slow VS Fast AI Runtime

\n\n
\n\nThe `Google AI Fast or Slow? Predict AI Model Runtime` competition is a `machine learning challenge` targets to `develop a model to predict the runtime of AI models` on a `variety of hardware configurations`. The competition data consists of a `training set`/`test set`, with each set containing information about `AI models`, including their `characteristics`/`runtimes`. The `goal` is to `develop a model` that can `accurately predict` the `runtime of an AI model` based on its characteristics, such as the number of parameters, the number of layers, and the hardware configuration.\n\nThe total prize pool for the competition is $50,000$, with the top 3 teams winning prizes of $15,000$, $10,000$, and $5,000$, respectively.","metadata":{}},{"cell_type":"code","source":"import os ","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:22:17.965193Z","iopub.execute_input":"2023-08-30T08:22:17.965630Z","iopub.status.idle":"2023-08-30T08:22:17.970931Z","shell.execute_reply.started":"2023-08-30T08:22:17.965595Z","shell.execute_reply":"2023-08-30T08:22:17.969616Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\nOur model data is situtated acrros various domains ranging from $Computer-Vision$ to $Natural-Langauage-Processing$. A look at the models can be seen below","metadata":{}},{"cell_type":"code","source":"os.listdir('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/xla/default/train')","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:19:59.646097Z","iopub.execute_input":"2023-08-30T08:19:59.646794Z","iopub.status.idle":"2023-08-30T08:19:59.680278Z","shell.execute_reply.started":"2023-08-30T08:19:59.646748Z","shell.execute_reply":"2023-08-30T08:19:59.679386Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

1 | Data 📈

","metadata":{}},{"cell_type":"code","source":"import tqdm\nimport numpy as np \nimport pandas as pd ","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:22:21.330388Z","iopub.execute_input":"2023-08-30T08:22:21.330878Z","iopub.status.idle":"2023-08-30T08:22:21.336934Z","shell.execute_reply.started":"2023-08-30T08:22:21.330838Z","shell.execute_reply":"2023-08-30T08:22:21.335654Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\nOur files are divided in $4$ Dirs","metadata":{}},{"cell_type":"code","source":"ndt = os.listdir('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/nlp/default/train')\nnrt = os.listdir('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/nlp/random/train')\n\nxdt = os.listdir('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/xla/default/train')\nxrt = os.listdir('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/xla/random/train')","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:22:21.536864Z","iopub.execute_input":"2023-08-30T08:22:21.537302Z","iopub.status.idle":"2023-08-30T08:22:21.588853Z","shell.execute_reply.started":"2023-08-30T08:22:21.537264Z","shell.execute_reply":"2023-08-30T08:22:21.587401Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ndt = [\n np.load('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/nlp/default/train/' + value)\n for value \n in tqdm.tqdm(ndt , total = len(ndt) , desc = 'Loading NDT --->')\n]\nnrt = [\n np.load('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/nlp/random/train/' + value)\n for value \n in tqdm.tqdm(nrt , total = len(nrt) , desc = 'Loading NRT --->')\n]\nxdt = [\n np.load('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/xla/default/train/' + value)\n for value\n in tqdm.tqdm(xdt , total = len(xdt) , desc = 'Loading XDT --->')\n]\nxrt = [\n np.load('/kaggle/input/predict-ai-model-runtime/npz_all/npz/layout/xla/random/train/' + value)\n for value\n in tqdm.tqdm(xrt , total = len(xrt) , desc = 'Loading NRT --->')\n]\n\nfiles = ndt + nrt + xdt + xrt \nlen(files)","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:22:22.201733Z","iopub.execute_input":"2023-08-30T08:22:22.202198Z","iopub.status.idle":"2023-08-30T08:22:30.123815Z","shell.execute_reply.started":"2023-08-30T08:22:22.202160Z","shell.execute_reply":"2023-08-30T08:22:30.122487Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\nEach file contains \n\n|||\n|---|---|\n|Node Feature\n|Node OpCode|Specific operation or instruction that is performed by a node\n|Edge Index |Connections between nodes in a graph or a neural network\n|Node Config Features\n|Node Config IDS\n|Config Runtime\n|Node Splits\n\nWe will make these files to a DataFrame in export it in the output DIRs","metadata":{}},{"cell_type":"code","source":"data = pd.DataFrame({\n 'node_feat' : [\n val['node_feat']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Feats --->')\n ] , \n 'node_opcode' : [\n val['node_opcode']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Opcodes --->')\n ] , \n 'edge_index' : [\n val['edge_index']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Edge Index --->')\n ] , \n# 'node_config_feat' : [\n# val['node_config_feat']\n# for val \n# in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Config Feats --->')\n# ] , \n 'node_config_ids' : [\n val['node_config_ids']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Config Ids --->')\n ] , \n 'config_runtime' : [\n val['config_runtime']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Config Runtime --->')\n ] , \n 'node_splits' : [\n val['node_splits']\n for val \n in tqdm.tqdm(files , total = len(files) , desc = 'Reading Node Splits --->')\n ]\n})","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:22:57.716420Z","iopub.execute_input":"2023-08-30T08:22:57.716935Z","iopub.status.idle":"2023-08-30T08:23:06.370157Z","shell.execute_reply.started":"2023-08-30T08:22:57.716893Z","shell.execute_reply":"2023-08-30T08:23:06.368954Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data.head()","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:30:19.136903Z","iopub.execute_input":"2023-08-30T08:30:19.137375Z","iopub.status.idle":"2023-08-30T08:30:19.766156Z","shell.execute_reply.started":"2023-08-30T08:30:19.137337Z","shell.execute_reply":"2023-08-30T08:30:19.764793Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data.to_csv('/kaggle/working/Input Files.csv')","metadata":{"execution":{"iopub.status.busy":"2023-08-30T08:24:05.052300Z","iopub.execute_input":"2023-08-30T08:24:05.052756Z","iopub.status.idle":"2023-08-30T08:24:05.763734Z","shell.execute_reply.started":"2023-08-30T08:24:05.052721Z","shell.execute_reply":"2023-08-30T08:24:05.762089Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

2 | Ending 🏁

\n\n
\n\n**WE WILL GO DEEPER INTO THE DATA IN THE UPCOMING VERSIONS**\n\n**PLEASE COMMENT DOWN IF I DID ANY MISTAKES, OR IF CAN MAKE THIS MORE CONNECTED TO THE GROUND, OR SUGGESTIONS. YOUR ASSISTS ARE HIGHLY APPRECIABLE**\n\n**THATS IT FOR TODAY GUYS**\n\n**HOPE YOU UNDERSTOOD AND LIKED MY WORK**\n\n**DONT FORGET TO MAKE AN UPVOTE $:)$**\n \n\n \n**PEACE OUT**","metadata":{}}],"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}}