L16-V65536-chat-tokenizer / conversion.json
amphora's picture
L16-V65536 tokenizer with Llama-3 chat specials in reserved ids 4-8
1e037db verified
Raw History Blame Contribute Delete
156 Bytes
{
"source_spm": "L16_V65536/spm.model",
"vocab_size": 65536,
"token_size_in_bytes": 2,
"eos_token_id": 2,
"bos_token_id": 1,
"pad_token_id": 3
}