This is a friendly reminder - the current text generation call will exceed the model's predefined maximum length (4096). Depending on the model, you may observe exceptions, performance degradation, or nothing at all.
Llama2: [What is the pass key? The pass key is 。。。. 。]
SelfExtend: [What is the pass key? The pass key is 。。。. 。]
Could you please elaborate more on your environment and your running scripts? It seems that both in our and other folks' tests, the current codes work well.
Tokens of Prompt: 5144 Passkey target: 89427
This is a friendly reminder - the current text generation call will exceed the model's predefined maximum length (4096). Depending on the model, you may observe exceptions, performance degradation, or nothing at all. Llama2: [What is the pass key? The pass key is 。。。. 。] SelfExtend: [What is the pass key? The pass key is 。。。. 。]