Views
No views yet
[!Note] Research artifact: the SFT-only intermediate checkpoint of the Instruct training pipeline. Use it to study or build on the SFT stage in isolation, run post-training experiments (preference tuning, RLVR, etc.), or compare against the final RL-tuned variant. For production-quality instruction following use Instruct instead.
Mellum2-12B-A2.5B-Base with three epochs of SFT on a mixed corpus of coding, agentic, tool use, instruction following, general chat, identity, and safety data. It is the starting point of the RLVR stage that produces the final Mellum2-12B-A2.5B-Instruct.| Checkpoint | Description |
|---|---|
| Base Pretrain | Base checkpoint before long-context extension |
| Base | Final base model |
| Instruct SFT | Supervised instruction-tuned checkpoint |
| Thinking SFT | Supervised thinking checkpoint |
| Instruct | RL-tuned instruction model |
| Thinking | RL-tuned thinking model |
1# Without tool calling
2vllm serve JetBrains/Mellum2-12B-A2.5B-Instruct-SFT --max-model-len 131072
3
4# With tool calling
5vllm serve JetBrains/Mellum2-12B-A2.5B-Instruct-SFT \
6 --max-model-len 131072 \
7 --enable-auto-tool-choice \
8 --tool-call-parser hermes1from openai import OpenAI
2# Configured by environment variables
3client = OpenAI()
4
5messages = [
6 {"role": "user", "content": "Write a Python function to reverse a string."},
7]
8
9chat_response = client.chat.completions.create(
10 model="JetBrains/Mellum2-12B-A2.5B-Instruct-SFT",
11 messages=messages,
12 max_tokens=81920,
13 temperature=0.6,
14 top_p=0.95,
15 extra_body={
16 "top_k": 20,
17 },
18)
19print("Chat response:", chat_response)