DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF

Text Generation • Updated about 22 hours ago • 387 • 9

Note Proof of concept performance upscale using 32 bit methods, merge adjustments and carry forward of all model precision through all points of the merge/model creation path. Vast increase in nuance, instruction following, output generation and emergence of new abilities. Q4km now operates above Q6 quality. Q6 and Q8 are also greatly improved.

DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF-imatrix

Text Generation • Updated about 22 hours ago • 455 • 6

Note Imatrix versions with further improvements - close to doubling perplexity drops noted in non-imatrix versions. Proof of concept performance upscale using 32 bit methods, merge adjustments and carry forward of all model precision through all points of the merge/model creation path. Vast increase in nuance, instruction following, output generation and emergence of new abilities. Q4km now operates above Q6 quality. Q6 and Q8 are also greatly improved. All versions are imatrixed in this repo.

DavidAU/TieFighter-Holodeck-Holomax-Mythomax-F1-V1-COMPOS-20B-gguf

Text Generation • Updated about 23 hours ago • 285 • 7

Note A proof of concept model designed to increase creativity and reduce the ability to predict it's output on purpose. This is a four model non mathematical merge also designed to keep all attributes of all models fully intact. It is scaled up to 20B to further enhance it's abilities.

DavidAU/D_AU-Tiefighter-OrcaMaid-V3-13B-32k-slerp

Text Generation • Updated May 2 • 21

Note 2nd attempt to make Tiefighter 32,768 context. Results pending.

DavidAU/D_AU-Tiefighter-Plus-Giraffe-13B-32k-slerp

Text Generation • Updated May 1 • 13

Note Step 2 in making Tiefighter 32k Context. Operational. GGUFs working. Uploading shortly.

DavidAU/D_AU-Tiefighter-Giraffe-13B-32k-slerp

Text Generation • Updated May 1 • 16

Note Step 1 in making Tiefighter 32k Context. Not for use.

DavidAU/D_AU-Orac-13B-Tiefighter-slerp

Text Generation • Updated Apr 29 • 12

Note Tested. Working 100%. GGUFs now added. Imatrix Plus GGUFs uploaded.

DavidAU/D_AU-Mistral-7B-Instruct-v0.2-Bagel-DarkSapling-DPO-7B-v2.0

Text Generation • Updated Apr 27 • 5

Note Full precision. Test to take on some of Bagel's attributes with fiction / role play model Dark Sapling. Imatrix, and Imatrix-Plus GGUFs to follow shortly.