Diplomat2-Fused-3B

The full 4-bit model for Diplomat2 — a fine-tuned delegate that writes formal Model United Nations position papers. This repo contains the LoRA adapter fused directly into mlx-community/Llama-3.2-3B-Instruct-4bit, so no separate adapter step is needed.

• Base + adapter: LoRA (rank 16, alpha 32, scale 10.0, 8 layers) fused into the 4-bit base
• Format: MLX, Apple-Silicon ready
• Rating: CRA-1 (Human-Centric Reasoning) — see the Diplomat2 model card for the full alignment framework and training details.

Quickstart

pip install mlx-lm

from mlx_lm import generate, load
from mlx_lm.sample_utils import make_sampler

model, tok = load("Ahmad170412/Diplomat2-Fused-3B")

out = generate(
    model, tok,
    prompt="Write a Model United Nations position paper for Nigeria on the topic: Ocean Plastic Pollution",
    max_tokens=600,
    sampler=make_sampler(temp=0.45),
)
print(out)

CLI

python3 -m mlx_lm generate \
  --model Ahmad170412/Diplomat2-Fused-3B \
  --prompt "Write a Model United Nations position paper for Nigeria on the topic: Ocean Plastic Pollution" \
  --max-tokens 600 \
  --temp 0.45

Notes

• Use temp 0.4–0.5; higher temperatures cause repetition loops.
• For the small adapter upload (~30 MB) instead, use Diplomat2.
• Output is a drafting aid — verify facts before real committee use.

Fused with MLX. Base model under the Llama 3.2 Community License.