Nice job!
I think this might be one of my Llama3 faves. I have some EXL2 quants on my page if anyone is interested. I'm editing the jsons to use the eot_id
Yeah, I'm keen to give this a try now!
Pleased to see the example stories were "shivers down spine" free! :D Fingers crossed this brings out the best in Euryale-v2.1.
Yeah, I'm keen to give this a try now!
Pleased to see the example stories were "shivers down spine" free! :D Fingers crossed this brings out the best in
Euryale-v2.1.
@jukofyork I am eager to see if you have any success with vectors! I diddled around with this one off and on for about eight hours yesterday, and I was impressed. The first one of the Llama 3 varients I tried that didn't start to come apart as easily as the context grew. I didn't have a chance to actually put it through my normal paces because I was having too much fun with it.
I think this might be one of my Llama3 faves. I have some EXL2 quants on my page if anyone is interested. I'm editing the jsons to use the eot_id
Confirmed. This decendant of Smaug-Llama-3-32k is amazing. Impressed with RP capabilities that affects how it follows complex and huge system prompt (26k tokens). Thank's for exl2 quants.
I think this might be one of my Llama3 faves. I have some EXL2 quants on my page if anyone is interested. I'm editing the jsons to use the eot_id
Confirmed. This decendant of Smaug-Llama-3-32k is amazing. Impressed with RP capabilities that affects how it follows complex and huge system prompt (26k tokens). Thank's for exl2 quants.
Not THAT's a system prompt. Wow. You must have good genetics...
Seriously though, it's good to hear it isn't just me. I am constantly downloading and evaluating these models and sometimes I wonder if I am hallucinating when one jumps out at me as standing apart.