import os # options INDEX_BY_DESC = { 'Dolma-v1.7 (2.6T tokens)': 'v4_dolma-v1_7_llama', 'RedPajama (1.4T tokens)': 'v4_rpj_llama_s4', 'Pile-train (380B tokens)': 'v4_piletrain_llama', 'C4-train (200B tokens)': 'v4_c4train_llama', 'Pile-val (390M tokens)': 'v4_pileval_llama', } INDEX_DESCS = list(INDEX_BY_DESC.keys()) # API limits and defaults MAX_QUERY_CHARS = int(os.environ.get('MAX_QUERY_CHARS', 1000)) NGRAM_LEN_MIN = int(os.environ.get('NGRAM_LEN_MIN', 5)) NGRAM_LEN_MAX = int(os.environ.get('NGRAM_LEN_MAX', 11)) MAX_DISP_CHARS_PER_LINE = int(os.environ.get('MAX_DISP_CHARS_PER_LINE', 100)) # HF demo API_URL = os.environ.get('API_URL', None) DEFAULT_CONCURRENCY_LIMIT = os.environ.get('DEFAULT_CONCURRENCY_LIMIT', 10) MAX_SIZE = os.environ.get('MAX_SIZE', 100) MAX_THREADS = os.environ.get('MAX_THREADS', 40) DEBUG = (os.environ.get('DEBUG', 'False') != 'False')