Add logo, use case note, vLLM serving, and Quickstart

README.md

@@ -206,8 +206,16 @@ model-index:
 license: apache-2.0
 ---
 
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="mellum-logo-dark.svg">
+  <img alt="Mellum" src="mellum-logo.svg" width="320">
+</picture>
+
 # Mellum 2 Base
 
+> [!Note]
+> Use this checkpoint as the starting point for your own fine-tuning, alignment, or domain adaptation on top of the long-context base. For instruction-following or reasoning tasks out of the box, use [Instruct](https://huggingface.co/JetBrains/Mellum2-12B-A2.5B-Instruct) or [Thinking](https://huggingface.co/JetBrains/Mellum2-12B-A2.5B-Thinking) instead.
+
 ## Mellum 2 Base Highlights
 
 Mellum 2 Base is a long-context pretrained causal language model trained by JetBrains.
@@ -245,6 +253,34 @@ This repository contains one checkpoint from the Mellum 2 family.
 - Vocabulary Size: 98,304
 - Precision: bfloat16
 
+## Serving with vLLM
+
+```sh
+vllm serve JetBrains/Mellum2-12B-A2.5B-Base --max-model-len 131072
+```
+
+## Quickstart
+
+Text-Only Input (base model — use the completions endpoint, not chat)
+
+```python
+from openai import OpenAI
+# Configured by environment variables
+client = OpenAI()
+
+completion = client.completions.create(
+    model="JetBrains/Mellum2-12B-A2.5B-Base",
+    prompt="def fibonacci(n):\n    ",
+    max_tokens=81920,
+    temperature=0.6,
+    top_p=0.95,
+    extra_body={
+        "top_k": 20,
+    },
+)
+print("Completion:", completion)
+```
+
 ## Evaluation
 
 Evaluation results are available in the model card. All values are self-reported by JetBrains.