test_tiktoken.py 320 B

123456789101112
  1. import os
  2. import tiktoken
  3. from core.model_runtime.model_providers.__base.tokenizers.gpt2_tokenzier import GPT2Tokenizer
  4. def test_tiktoken():
  5. os.environ["TIKTOKEN_CACHE_DIR"] = "/tmp/.tiktoken_cache"
  6. GPT2Tokenizer.get_num_tokens("Hello, world!")
  7. assert tiktoken.registry.ENCODING_CONSTRUCTORS is not None