Downloads · 30 days
46
4% of all-time downloads
skyerx/gemma-3-270m-prompt-rewriter-GGUF
gemma-3-270m-prompt-rewriter-GGUF is a text generation model from skyerx. Use it when you need the model to write or continue text. The card lists the license as mit.
This repository hosts a lightweight, edge-optimized f16 GGUF conversion of a fine-tuned Gemma 3 270M model. It is specifically engineered for raw text-completion sequences, transforming brief, casual user commands int…
Downloads · 30 days
46
4% of all-time downloads
All-time downloads
1K
Public
Repo size
551 MB
Likes
1
Public
Click a slice to open those files.
.gguf1.1 GB · 100%
From the Hugging Face model README
This repository hosts a lightweight, edge-optimized f16 GGUF conversion of a fine-tuned Gemma 3 270M model. It is specifically engineered for raw text-completion sequences, transforming brief, casual user commands into comprehensive, structured, production-ready system prompts for LLMs.
Unlike conversational models that maintain ongoing multi-turn chat dynamics, this model was fine-tuned specifically for direct completion tasks. By bypassing traditional conversational tokens (<start_of_turn>, assistant, etc.), the architecture learns a direct mapping from a short input directly to the desired complex prompt structure.
f16 GGUF execution layers.262,144 structural tensor configuration, eliminating vision-encoder padding tokens (<image_soft_token> native to Gemma 3's multimodal architecture) to streamline local CPU/GPU text-only inference.Follow these steps to pull down your files and register your prompt rewriter model locally using Ollama.
Make sure you have Ollama installed and running on your machine:
ollama --versionDownload both rewriter.gguf and the companion Modelfile from the Files and versions tab of this repository, and place them together in a single folder on your local machine (e.g., C:\AI\Models\ or ~/models/).
Navigate into the directory where you saved the files:
cd C:\AI\Models
cd ~/models
Run the create command to parse your local Modelfile layout and compile a clean model configuration image within Ollama's local registry.
(Note: If you are replacing an older execution layer, run ollama rm gemma-3-270m-prompt-rewriter first to clear historical caches).
ollama create gemma-3-270m-prompt-rewriter -f ./Modelfile
Once the build completes successfully, initialize your model inside the local inference runner:
ollama run gemma-3-270m-prompt-rewriter
Because this model is configured strictly for raw text completion rather than conversational chat loops, it expects a short string input.
Once the interactive >>> prompt appears, simply type your short intent:
>>> build a todo api
The model will process the input sequence instantly and output a structured, production-ready system prompt layout before cleanly terminating the generation stream via the <eos> boundary.