From cad7e9a1cd103321dd87c30b1ed079bcebd086e4 Mon Sep 17 00:00:00 2001 From: Ettore Di Giacinto Date: Thu, 9 Jan 2025 15:55:56 +0100 Subject: [PATCH] chore(model gallery): add 14b-qwen2.5-freya-x1 (#4566) Signed-off-by: Ettore Di Giacinto --- gallery/index.yaml | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/gallery/index.yaml b/gallery/index.yaml index 5678e759..7a71e137 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -2849,6 +2849,30 @@ - filename: Dolphin3.0-Qwen2.5-3b-Q4_K_M.gguf sha256: 0cb1908c5f444e1dc2c5b5619d62ac4957a22ad39cd42f2d0b48e2d8b1c358ab uri: huggingface://bartowski/Dolphin3.0-Qwen2.5-3b-GGUF/Dolphin3.0-Qwen2.5-3b-Q4_K_M.gguf +- !!merge <<: *qwen25 + name: "14b-qwen2.5-freya-x1" + icon: https://huggingface.co/Sao10K/14B-Qwen2.5-Freya-x1/resolve/main/sad.png + urls: + - https://huggingface.co/Sao10K/14B-Qwen2.5-Freya-x1 + - https://huggingface.co/DevQuasar/Sao10K.14B-Qwen2.5-Freya-x1-GGUF + description: | + I decided to mess around with training methods again, considering the re-emegence of methods like multi-step training. Some people began doing it again, and so, why not? Inspired by AshhLimaRP's methology but done it my way. + Freya-S1 + + LoRA Trained on ~1.1GB of literature and raw text over Qwen 2.5's base model. + Cleaned text and literature as best as I could, still, may have had issues here and there. + + Freya-S2 + + The first LoRA was applied over Qwen 2.5 Instruct, then I trained on top of that. + Reduced LoRA rank because it's mainly instruct and other details I won't get into. + overrides: + parameters: + model: Sao10K.14B-Qwen2.5-Freya-x1.Q4_K_M.gguf + files: + - filename: Sao10K.14B-Qwen2.5-Freya-x1.Q4_K_M.gguf + sha256: 790953e2ffccf2f730d52072f300fba9d1549c7762f5127b2014cdc82204b509 + uri: huggingface://DevQuasar/Sao10K.14B-Qwen2.5-Freya-x1-GGUF/Sao10K.14B-Qwen2.5-Freya-x1.Q4_K_M.gguf - &smollm ## SmolLM url: "github:mudler/LocalAI/gallery/chatml.yaml@master"