From af33483687affc05cfaea69db2b8846efd63f6bc Mon Sep 17 00:00:00 2001 From: Ettore Di Giacinto Date: Sat, 14 Dec 2024 11:27:11 +0100 Subject: [PATCH] chore(model gallery): add fusechat-qwen-2.5-7b-instruct (#4380) Signed-off-by: Ettore Di Giacinto --- gallery/index.yaml | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/gallery/index.yaml b/gallery/index.yaml index 90564eaeb1b0..9b1f5ea67c3a 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -2018,6 +2018,21 @@ - filename: Evathene-v1.3-Q4_K_M.gguf sha256: 0f54909b3ddca514994ee16417da8750f56e7bd59581b46ac47625c230e29d1f uri: huggingface://bartowski/Evathene-v1.3-GGUF/Evathene-v1.3-Q4_K_M.gguf +- !!merge <<: *qwen25 + name: "fusechat-qwen-2.5-7b-instruct" + icon: https://huggingface.co/FuseAI/FuseChat-Qwen-2.5-7B-Instruct/resolve/main/FuseChat-3.0.png + urls: + -https://huggingface.co/FuseAI/FuseChat-Qwen-2.5-7B-Instruct + - https://huggingface.co/bartowski/FuseChat-Qwen-2.5-7B-Instruct-GGUF + description: | + We present FuseChat-3.0, a series of models crafted to enhance performance by integrating the strengths of multiple source LLMs into more compact target LLMs. To achieve this fusion, we utilized four powerful source LLMs: Gemma-2-27B-It, Mistral-Large-Instruct-2407, Qwen-2.5-72B-Instruct, and Llama-3.1-70B-Instruct. For the target LLMs, we employed three widely-used smaller models—Llama-3.1-8B-Instruct, Gemma-2-9B-It, and Qwen-2.5-7B-Instruct—along with two even more compact models—Llama-3.2-3B-Instruct and Llama-3.2-1B-Instruct. The implicit model fusion process involves a two-stage training pipeline comprising Supervised Fine-Tuning (SFT) to mitigate distribution discrepancies between target and source LLMs, and Direct Preference Optimization (DPO) for learning preferences from multiple source LLMs. The resulting FuseChat-3.0 models demonstrated substantial improvements in tasks related to general conversation, instruction following, mathematics, and coding. Notably, when Llama-3.1-8B-Instruct served as the target LLM, our fusion approach achieved an average improvement of 6.8 points across 14 benchmarks. Moreover, it showed significant improvements of 37.1 and 30.1 points on instruction-following test sets AlpacaEval-2 and Arena-Hard respectively. We have released the FuseChat-3.0 models on Huggingface, stay tuned for the forthcoming dataset and code. + overrides: + parameters: + model: FuseChat-Qwen-2.5-7B-Instruct-Q4_K_M.gguf + files: + - filename: FuseChat-Qwen-2.5-7B-Instruct-Q4_K_M.gguf + sha256: 8cd8c317769f03125ac753c836ac92c5a76ee0b35502811d0e65bcbb8df9d55c + uri: huggingface://bartowski/FuseChat-Qwen-2.5-7B-Instruct-GGUF/FuseChat-Qwen-2.5-7B-Instruct-Q4_K_M.gguf - &archfunct license: apache-2.0 tags: