Transformers
GGUF
mistral
text-generation-inference

generate gibberish.

#1
by shengluo - opened

The prompt is as following which works fine for context length of 8092:

Summarize following text concisely by listing the key points like I am 5.
<|end_of_turn|>
Human:
===BEGIN===
some text
===END===
<|end_of_turn|>
Assistant:
...

Do I miss anything?

I ran it with LM Studio on mbp 14 with m1 pro.

may due problem when using LM Studio.

shengluo changed discussion status to closed

Sign up or log in to comment