ExternalVendorCode/LocalAI
1
0
Fork 0
mirror of https://github.com/mudler/LocalAI.git synced 2024-12-21 13:37:51 +00:00
Code Issues Actions 12 Packages Projects Releases Wiki Activity

models(gallery): add darkest-muse-v1 (#3973)

Browse Source 
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
This commit is contained in:
Ettore Di Giacinto 2024-10-26 14:52:55 +02:00 committed by GitHub
parent 546dce68a6
commit 43bfdc9561
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 23 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
gallery/index.yaml
View File

@ -3182,6 +3182,29 @@
- filename: Apollo2-9B.Q4_K_M.gguf - filename: Apollo2-9B.Q4_K_M.gguf
sha256: 9fdb63f78e574558a4f33782eca88716eea28e90ea3ae36c381769cde6b81e0f sha256: 9fdb63f78e574558a4f33782eca88716eea28e90ea3ae36c381769cde6b81e0f
uri: huggingface://mradermacher/Apollo2-9B-GGUF/Apollo2-9B.Q4_K_M.gguf uri: huggingface://mradermacher/Apollo2-9B-GGUF/Apollo2-9B.Q4_K_M.gguf
- !!merge <<: *gemma
name: "darkest-muse-v1"
icon: https://cdn-uploads.huggingface.co/production/uploads/65ad56b4c2eef2ba1154618c/0AB6uPPuCvbNtRZb3Rdj1.png
urls:
- https://huggingface.co/sam-paech/Darkest-muse-v1
- https://huggingface.co/bartowski/Darkest-muse-v1-GGUF
description: |
This is a creative writing merge of two very different models that I trained on the brand new Gutenberg3 dataset, plus Ataraxy-v2 in the mix.
It's lost much of the slop and tryhard vocab flexing and positivity bias that's typical of these models and writes in its own voice.
The main source model in the merge, Quill-v1, inherited a natural, spare prose from the human writing in the gutenberg set. The other source model, Delirium-v1, got overcooked in SIMPO training; it has crazy panache, a really dark flair for the grotesque, and has some mental issues. These two source models balance each other out in the merge, resulting in something pretty unique.
It seems to be quite uncensored and creative. Since Delirium was pushed right to the edge during training, the merge may exhibit some of its weirdness and word / concept fixations. This may be mitigated by using custom anti-slop lists.
The payoff is a really creative, stream of consciousness style of writing, with punchy dialogue that I haven't seen in other models. Oh, it also scored around the top of the EQ-Bench creative writing leaderboard!
overrides:
parameters:
model: Darkest-muse-v1-Q4_K_M.gguf
files:
- filename: Darkest-muse-v1-Q4_K_M.gguf
sha256: a19ec9e3dc875511ea771bf363e71e7ae5578986b2f8cf50aeb50683d56e9b76
uri: huggingface://bartowski/Darkest-muse-v1-GGUF/Darkest-muse-v1-Q4_K_M.gguf
- &llama3 - &llama3
url: "github:mudler/LocalAI/gallery/llama3-instruct.yaml@master" url: "github:mudler/LocalAI/gallery/llama3-instruct.yaml@master"
icon: https://cdn-uploads.huggingface.co/production/uploads/642cc1c253e76b4c2286c58e/aJJxKus1wP5N-euvHEUq7.png icon: https://cdn-uploads.huggingface.co/production/uploads/642cc1c253e76b4c2286c58e/aJJxKus1wP5N-euvHEUq7.png