🧪 Gemma-2B-DolphinR1-TestV2 (Experimental Fine-Tune) 🧪

This is an experimental fine-tune of Google's Gemma-2B using the Dolphin-R1 dataset.

The goal is to enhance reasoning and chain-of-thought capabilities while maintaining efficiency with LoRA (r=32) and 4-bit quantization.

🚨 Disclaimer: This model is very much a work in progress and is still being tested for performance, reliability, and generalization. Expect quirks, inconsistencies, and potential overfitting in responses.

Balkon mode confirmed lol. It's broken.

base_model: YorkieOH10/Gemma-2B-DolphinR1-Testv2 tags: - llama-cpp - gguf-my-repo

YorkieOH10/Gemma-2B-DolphinR1-Testv2-Q8_0-GGUF

This model was converted to GGUF format from YorkieOH10/Gemma-2B-DolphinR1-Testv2 using llama.cpp via the ggml.ai's GGUF-my-repo space. Refer to the original model card for more details on the model.

Use with llama.cpp

Install llama.cpp through brew (works on Mac and Linux)

brew install llama.cpp

Invoke the llama.cpp server or the CLI.

CLI:

llama-cli --hf-repo YorkieOH10/Gemma-2B-DolphinR1-Testv2-Q8_0-GGUF --hf-file gemma-2b-dolphinr1-testv2-q8_0.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo YorkieOH10/Gemma-2B-DolphinR1-Testv2-Q8_0-GGUF --hf-file gemma-2b-dolphinr1-testv2-q8_0.gguf -c 2048

Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.

git clone https://github.com/ggerganov/llama.cpp

Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).

cd llama.cpp && LLAMA_CURL=1 make

Step 3: Run inference through the main binary.

./llama-cli --hf-repo YorkieOH10/Gemma-2B-DolphinR1-Testv2-Q8_0-GGUF --hf-file gemma-2b-dolphinr1-testv2-q8_0.gguf -p "The meaning to life and the universe is"

./llama-server --hf-repo YorkieOH10/Gemma-2B-DolphinR1-Testv2-Q8_0-GGUF --hf-file gemma-2b-dolphinr1-testv2-q8_0.gguf -c 2048