diff --git a/constants.py b/constants.py index 87d6b7a06..2f96fb960 100644 --- a/constants.py +++ b/constants.py @@ -36,19 +36,15 @@ KEY_ROPE_DIMENSION_COUNT = "{llm}.rope.dimension_count" KEY_ROPE_SCALE = "{llm}.rope.scale" # tokenization -KEY_TOKENIZER_MODEL = "tokenizer.ggml.model" -KEY_TOKENIZER_LIST = "tokenizer.ggml.tokens" -KEY_TOKENIZER_SCORES = "tokenizer.ggml.scores" -KEY_TOKENIZER_MERGES = "tokenizer.ggml.merges" -KEY_TOKENIZER_BOS_ID = "tokenizer.ggml.bos_token_id" -KEY_TOKENIZER_EOS_ID = "tokenizer.ggml.eos_token_id" -KEY_TOKENIZER_UNK_ID = "tokenizer.ggml.unknown_token_id" -KEY_TOKENIZER_SEP_ID = "tokenizer.ggml.seperator_token_id" -KEY_TOKENIZER_PAD_ID = "tokenizer.ggml.padding_token_id" -KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json" -KEY_TOKENIZER_RWKV = "tokenizer.rwkv.world" -KEY_TOKENIZER_BOS_ID = "tokenizer.ggml.bos_token_id" -KEY_TOKENIZER_EOS_ID = "tokenizer.ggml.eos_token_id" -KEY_TOKENIZER_UNK_ID = "tokenizer.ggml.unknown_token_id" -KEY_TOKENIZER_SEP_ID = "tokenizer.ggml.separator_token_id" -KEY_TOKENIZER_PAD_ID = "tokenizer.ggml.padding_token_id" +KEY_TOKENIZER_MODEL = "tokenizer.ggml.model" +KEY_TOKENIZER_LIST = "tokenizer.ggml.tokens" +KEY_TOKENIZER_TOKEN_TYPE = "tokenizer.ggml.token_type" +KEY_TOKENIZER_SCORES = "tokenizer.ggml.scores" +KEY_TOKENIZER_MERGES = "tokenizer.ggml.merges" +KEY_TOKENIZER_BOS_ID = "tokenizer.ggml.bos_token_id" +KEY_TOKENIZER_EOS_ID = "tokenizer.ggml.eos_token_id" +KEY_TOKENIZER_UNK_ID = "tokenizer.ggml.unknown_token_id" +KEY_TOKENIZER_SEP_ID = "tokenizer.ggml.seperator_token_id" +KEY_TOKENIZER_PAD_ID = "tokenizer.ggml.padding_token_id" +KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json" +KEY_TOKENIZER_RWKV = "tokenizer.rwkv.world"