Llama 3 Context-Obedient Models in a 3 x MoE configuration. The models experts are split into Understanding and summarizing input, following the format provided, and outputting only context relevant answers as the three experts involved
2ed3e35
verified
{ | |
"bos_token": { | |
"content": "<|begin_of_text|>", | |
"lstrip": false, | |
"normalized": false, | |
"rstrip": false, | |
"single_word": false | |
}, | |
"eos_token": { | |
"content": "<|im_end|>", | |
"lstrip": false, | |
"normalized": false, | |
"rstrip": false, | |
"single_word": false | |
}, | |
"pad_token": { | |
"content": "<|begin_of_text|>", | |
"lstrip": false, | |
"normalized": false, | |
"rstrip": false, | |
"single_word": false | |
} | |
} | |