Borcherding/OARC_Commander_v001_beta

🎮 OARC_Commander_v0.1-Llama-3.2-3B 🎮

Part of the 🎮OARC_Commander🎮 Collection

Curated and trained by Leo Borcherding

OARC Discord: https://discord.gg/mNeQZzBHuW

OARC Github: https://github.com/Leoleojames1/ollama_agent_roll_cage

What is OARC_Commander_v0.1?

OARC_Commander_v0.1 is the first generation of the OARC_Commander series of models.

OARC_Commander aims to be a general purpose Llama-3.2-3B model, while also having been finetuned on set of user commands in Ollama Agent Roll Cage (OARC).

Chat Template

We use ChatML for the chat template.

<|im_start|>system
You are OARC_Commander, a helpful AI assistant that has been trained on the command list from ollama agent roll cage. You will guide the OARC user, and help them understand the meaning of any oarc information that they will request from you.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

System Prompt

In OARC_Commander, the system prompt is what you can use to set the tone and alignment of the responses for OARC_Commander if you wish.

Example use of system prompt:

<|im_start|>system
You are OARC_Commander, a helpful AI assistant that has been trained on the command list from ollama agent roll cage. You will guide the OARC user, and help them understand the meaning of any oarc information that they will request from you.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Sample Outputs

Here are some sample outputs from OARC_Commander_v001, showcasing the commanding data from the Borcherding/OARC_Commander_v001 training data, I am aware that there are like greek and korean tokens being randomly output, I am working on figuring out why this is, and will update the repo with the fixed version once I have ironed this out.

How to use

There are many ways to use a huggingface model including:

• ollama
• Huggingface Transformers library
• vllm
• LM Studio

ollama usage

• Install ollama

• ollama run hf.co/Borcherding/OARC_Commander_v001_beta:Q8_0

• instead of Q8_0, you can try F16, or Q4_K_M as well, these commands can be found in the top right under "Use this model"

• once you have ollama loaded try asking the mode something like:

• "what is the purpose of the /swap command?"

• you can also try:

• /set system <your system prompt>

Evals

TBD

Appreciation

Respect and thanks to the creators of the open source datasets that were used:

Special thanks to

• Meta, and @mlabonne who wrote papers and published models that were instrumental in creating OARC_Commander_v001.