Azirine

> Can you attach the sgf or link it on a medium without a login page? This is the file: [__go4go_20230927_Park-Junghwan_Xu-Haohong.txt](https://github.com/sanderland/katrain/files/12778197/__go4go_20230927_Park-Junghwan_Xu-Haohong.txt) GitHub doesn't allow attaching sgfs so I changed the...

After crashing 3-4 times, I cannot reproduce the crash again on demand (it still crashed at some random time with some other game). However, after doing the following, I was...

./main -ins -s 0 --in-prefix "[INST] " --in-suffix "[/INST] " -c 0 --repeat-penalty 1 --temp 0 -m mistral-7b-instruct-v0.2.Q5_K_M.gguf b2444 > [INST] Who are you? [/INST] I am a large language...

**Options:** ./main -ins -s 0 --in-prefix "[INST] " --in-suffix "[/INST] " -c 0 --repeat-penalty 1 --temp 0 -m mistral-7b-instruct-v0.2.Q5_K_M.gguf **Prompt:** Who are you? **b2444** I am a large language model...

## A longer example **Options:** ./main -ins -s 0 --in-prefix "[INST] " --in-suffix "[/INST] " -c 0 --repeat-penalty 1 --temp 0 -m mistral-7b-instruct-v0.2.Q5_K_M.gguf **Prompt:** In a world ruled by dictatorship,...

Output improved with the patch but there are still differences compared to pre-b2447, what else could be causing it?

Fixed in new version https://huggingface.co/lmstudio-community/Meta-Llama-3.1-8B-Instruct-GGUF/commit/42de8c2795a4283d753795e84f5f7f1aa9fd267d

Most quants are still affected, it was only fixed in the one above. For example: https://huggingface.co/lmstudio-community/Meta-Llama-3.1-70B-Instruct-GGUF https://huggingface.co/legraphista/Meta-Llama-3.1-70B-Instruct-IMat-GGUF

Fixed in new quants made after b3472 https://huggingface.co/legraphista/Meta-Llama-3.1-70B-Instruct-IMat-GGUF