Upload V2: stackformer GPT-2 + sparse cross-attention vision model (128 visual tokens)
4666d47 verified | { | |
| "bos_token": "<|endoftext|>", | |
| "eos_token": "<|endoftext|>", | |
| "pad_token": { | |
| "content": "<|pad|>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false | |
| }, | |
| "unk_token": "<|endoftext|>" | |
| } | |