chore: Add Ollama Modelfile

# Summary
Add an Ollama Modelfile demonstrating how to run this fine-tuned model locally using Ollama.

Modelfile

@@ -0,0 +1,28 @@
+# Ollama Modelfile to use this fine-tuned model with Ollama.
+# TODO (Ryan Parker, 7-May-2026): this Modelfile currently doesn't work.
+#    When using safetensors, the base model must be downloaded first, and
+#    Ollama raises an error stating that MLX is required for quantization.
+#    Note that the base model can be downloaded like this:
+#        uv tool run --from huggingface-hub -- hf download LiquidAI/LFM2.5-1.2B-Instruct
+#    When using GGUF, Ollama successfully creates the model, but then raises
+#    an error that LoRA is not supported.
+
+# safetensors version
+# Uses the adapter_model.safetensors trained in this repository.
+# Usage:
+# ollama create lfm-coder --file Modelfile --quantize q4_K_M --experimental
+
+# FROM hf.co/LiquidAI/LFM2.5-1.2B-Instruct
+# ADAPTER adapter_model.safetensors
+
+# GGUF version
+# Usage:
+# ollama create lfm-coder --file Modelfile
+FROM hf.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M
+ADAPTER hf.co/rparkr/LFM2.5-1.2B-Instruct-Coding-F16-GGUF
+
+# Recommended settings from: https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct#:~:text=Generation%20parameters%3A
+PARAMETER temperature 0.1
+PARAMETER top_k 50
+PARAMETER repeat_penalty 1.05
+PARAMETER num_ctx 128000