Download models/admet_ft/_modules/utils.py from Hattie/PharmAI-models: direct link, hf CLI and curl.
- Browser
- Download file 2.68 kB
-
https://huggingface.co/Hattie/PharmAI-models/resolve/main/models/admet_ft/_modules/utils.py
- Command line
-
hf download hf://Hattie/PharmAI-models/models/admet_ft/_modules/utils.py
-
curl -L -o utils.py https://huggingface.co/Hattie/PharmAI-models/resolve/main/models/admet_ft/_modules/utils.py
2.68 kB
| """ | |
| utils.py - μμ μ μ λ° μ νΈλ¦¬ν° ν¨μ | |
| """ | |
| import os | |
| from typing import Dict, List | |
| # λ°μ΄ν°μ μ€μ μ μν Class name μ μ | |
| DIDB_FILTER_COLS = [ | |
| "logP", | |
| "pKa", | |
| "solubility", | |
| "permeability", | |
| "plasma_protein_binding", | |
| # "fm_in_vitro", | |
| "fu_in_vitro", | |
| # "CYP450_Enzyme" | |
| ] | |
| LOAD_DATA_PATH = { | |
| "didb":"./dataset_v2/categorized_dataset/preprocessed_DIDBset.csv", | |
| "portal":"./dataset_v2/categorized_dataset/preprocessed_portal.csv", | |
| "kist":"./dataset_v2/categorized_dataset/convert_kist_data.csv", | |
| "all":"./dataset_v2/categorized_dataset/total.csv" | |
| } | |
| LOAD_PREPARED_DATA_PATH = { | |
| "didb":"./dataset_v2/prepared_set/didb", | |
| "portal":"./dataset_v2/prepared_set/portal", | |
| "kist":"./dataset_v2/prepared_set/kist", | |
| "all":"./dataset_v2/prepared_set/all", | |
| "all_clean":"./dataset_v2/prepared_set_outline/all", # Outliers removed | |
| "all_rating":"./dataset_v2/prepared_set_rating/all", # Balanced distributions (test outliers removed) | |
| } | |
| def get_task_list(task_type: str) -> List[str]: | |
| """ | |
| νμ€ν¬ μ νκ³Ό λ°μ΄ν° μ νμ λ°λ₯Έ νμ€ν¬ λͺ©λ‘μ λ°νν©λλ€. | |
| Args: | |
| task_type: νμ€ν¬ μ ν ('cls' λλ 'reg' λλ 'multi_reg') | |
| Returns: | |
| νμ€ν¬ λͺ©λ‘ | |
| """ | |
| y_cols_base = DIDB_FILTER_COLS | |
| # νμ€ν¬ μ νμ λ°λ₯Έ μΆλ ₯ μ»¬λΌ μ€μ | |
| if task_type == 'cls': | |
| return [f'{x}.cls' for x in y_cols_base] | |
| else: # reg and multi_reg | |
| return y_cols_base | |
| def load_vocabs(data_folder: str, cls_targets: List[str]) -> Dict[str, Dict[str, int]]: | |
| """ | |
| λΆλ₯ νμ€ν¬μ© μ΄ν μ¬μ μ λ‘λν©λλ€. | |
| Args: | |
| data_folder: λ°μ΄ν° ν΄λ κ²½λ‘ | |
| cls_targets: λΆλ₯ νμ€ν¬ λ μ΄λΈ λͺ©λ‘ | |
| Returns: | |
| μ΄ν μ¬μ (νμ€ν¬λ³ {λ μ΄λΈ: μΈλ±μ€}) | |
| """ | |
| all_vocabs = {} | |
| for target in cls_targets: | |
| vocab_path = os.path.join(data_folder, f'vocabs/{target}.vocab') | |
| if target not in all_vocabs: | |
| all_vocabs[target] = {} | |
| try: | |
| with open(vocab_path, 'r', encoding='utf-8') as f: | |
| for line in f: | |
| line = line.rstrip() | |
| vocab, idx = line.split('\t') | |
| idx = int(idx) | |
| all_vocabs[target][vocab] = idx | |
| except FileNotFoundError: | |
| # νμΌμ΄ μλ κ²½μ° κΈ°λ³Έκ° μ¬μ© | |
| print(f"Warning: Vocabulary file not found for {target}. Using default values.") | |
| all_vocabs[target] = {'inhibitor': 1, 'non-inhibitor': 0} | |
| return all_vocabs |