Jackrong/Qwopus3.6-27B-Coder-MTP-GGUF · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Machine Learning Model
Summary
Mellum2-12B-A2.5B-Thinking is a JetBrains reasoning-augmented code assistant utilizing a Mixture-of-Experts architecture with 12B total parameters and a 131K-token context. It explicitly emits chain-of-thought reasoning traces within tags, making it ideal for debugging and multi-step planning.
Highlights
- Supports a 131,072-token context window using a hybrid of sliding-window and full-attention layers.
- Requires vLLM nightly builds (post-v0.22.0) for MellumForCausalLM support and specific reasoning parsers like qwen3.
- JetBrains recommends the Instruct variant for low-latency direct answers, reserving the Thinking variant for complex reasoning tasks.
auto-generated
Context
Audience
Machine Learning Engineers
DomainMachine Learning
Formattechnical guide
Accessopen source
Source Overview
JetBrains/Mellum2-12B-A2.5B-Thinking is a Mixture-of-Experts code model with 12B total parameters, 2.5B active, and a 131K-token context.
It outputs explicit `<think>...</think>` reasoning traces for debugging, planning, and agentic coding. JetBrains recommends using the Instruct variant for direct low-latency answers without reasoning traces.
Topics
Related
vLLMMixture-of-ExpertsChain-of-ThoughtJetBrainsAgentic Coding
