This config file defines the settings for the model info script, which is responsible for inspecting and loading details about the model and its configuration, such as depth, model path, and configuration file paths.
The script reads its configuration from the file model_info_config.json
. Below is a list of the key parameters that can be configured:
- CUDA_VISIBLE_DEVICES: The GPU device(s) IDs to use for loading the model (e.g.,
2
). - max_RAM_usage: Maximum percentage of RAM to use during model inspection (e.g., 85%). If this limit is exceeded, an error will be raised.
- model_path: Path to the pre-trained model file (e.g.,
./pretrained/epoch-500_step-637037.pt
). - model_config_path: Path to the model configuration JSON file (e.g.,
./pretrained/params.json
). - depth: Specifies the depth of the model's architecture to be displayed (e.g.,
3
).
python scripts/python/testing/model_info.py --config config/model_info_config.json