NexesQuants/TeeZee_Kyllene-Yi-34B-v1.1-iMat.GGUF

Quants with iMatrix for : https://huggingface.co/TeeZee/Kyllene-34B-v1.1

Non-iMatrix quants (more choice in higher bitrate quants) : https://huggingface.co/TeeZee/Kyllene-34B-v1.1-GGUF/tree/main

!image/jpeg

TeeZee's Kyllene 34B v1.1 model is one of the best Yi_34b merge around with those of BruceTheMoose.

But it has a little thing which distinguishes it :

It uses Gryphe's MergeMonster as a tool to trim out the GPTisms, Yisms, and Llamaisms, and give a more natural output.

The clearing of any problematic gptism, llamaism, or yiism which was specified to MergeMonster is noticeable
And it's like the model is freed of these sequences which represent some form of "EOS chains of tokens" in many models, this in the sense that they conclude many outputs, this ofc in an unwanted way
It's quite a step in the right direction which should become the standard practice.

That make me wonder about the future, when we'll get Miqu 70b models properly finetuned with the best datatsets AND with the Mistralisms trimmed out as well.

Available quants :

Full offload possible on 48GB VRAM with a huge context size : Q8_0

Full offload possible on 36 GB VRAM with a huge context size : Q5KS

Full offload possible on 24GB VRAM with a big to huge context size (from 12288 with Q4KM, for example) : Q4KM, Q4KS, Q3KM

Full offload possible on 16GB VRAM with a decent context size : IQ3XXS SOTA (which is equivalent to a Q3KS with more context!), Q2K, Q2KS

Full offload possible on 12GB VRAM with a decent context size : IQ2XS SOTA. lower quality : IQ2XXS SOTA

Full offload maybe possible on 8GB VRAM with a small context size : IQ1_S revision "even better" (b2404) (or v5).
All my IQ1S quant from the 13/03/2024 will be with this new IQ1S quantization base.

The merge parameters and logs are in the repo : https://huggingface.co/TeeZee/Kyllene-34B-v1.1/tree/main

After iMatrixing and quantizing Kyllene, I benched her thoroughly, and she proved herself worthy :

Q4KS :

• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Hellaswag,85,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Hellaswag,85.2,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Hellaswag,84.6,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,HellaswagBin,81,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,HellaswagBin,83.5,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,HellaswagBin,82.95,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Arc-Challenge,61.53846154,,299,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Arc-Easy,80.35087719,,570,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,MMLU,43.13099042,,313,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Thruthful-QA,35.00611995,,817,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KS.gguf,-,Winogrande,79.3212,,1267,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4K_S.gguf,-,wikitext,5.1703,512,512,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,

Q4KM :

• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Hellaswag,84.75,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Hellaswag,85.6,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Hellaswag,84.9,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,HellaswagBin,81,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,HellaswagBin,83.4,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,HellaswagBin,82.9,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Arc-Challenge,60.53511706,,299,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Arc-Easy,80.52631579,,570,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,MMLU,42.49201278,,313,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Thruthful-QA,34.39412485,,817,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,Winogrande,79.4791,,1267,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,wikitext,5.1679,512,512,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4KM.gguf,-,wikitext,4.3623,4096,4096,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q4K_M.gguf,-,wikitext,4.4061,8192,8192,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,

Q5KS :

• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Hellaswag,85.25,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Hellaswag,85.6,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Hellaswag,84.95,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,HellaswagBin,81.25,,400,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,HellaswagBin,83.3,,1000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,HellaswagBin,83,,2000,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Arc-Challenge,60.20066890,,299,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Arc-Easy,81.05263158,,570,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,MMLU,42.17252396,,313,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Thruthful-QA,36.96450428,,817,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5KS.gguf,-,Winogrande,79.5580,,1267,2024-01-28 05:40:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,
• Kyllene-34B-v1.1-b1989-iMat-c32ch3250-Q5K_S.gguf,-,wikitext,5.1806,512,512,2024-01-28 00:00:00,,34b,Yi,200000,,,GGUF,TeeZee,Nexesenex,

-----

IQ1_S V5 :

• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,Hellaswag,70.3,,1000,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,Arc-Challenge,40.46822742,299,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,Arc-Easy,62.28070175,,570,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,MMLU,32.90734824,,313,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,Thruthful-QA,29.37576499,,817,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,Winogrande,68.7451,,1267,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,wikitext,9.8761,512,512,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,
• TeeZeeKyllene-34B-v1.1-b2409-iMat-c32ch3250-IQ1Sv5.gguf,-,wikitext,7.8954,4096,4096,2024-03-12 00:00:00,,34b,Yi,2000000,,,GGUF,TeeZee,Nexesenex,

-----

Enjoy these quants!