Qwen2 MoE#

Qwen1.5-MoE is a Mixture-of-Experts variant from Alibaba Cloud that activates only a fraction of parameters per token, enabling efficient training and inference at scale.

Task

Text Generation (MoE)

Architecture

Qwen2MoeForCausalLM

Parameters

14.3B total / 2.7B active

HF Org

Qwen

Available Models#

  • Qwen1.5-MoE-A2.7B: 14.3B total parameters, 2.7B activated per token

Architecture#

  • Qwen2MoeForCausalLM

Example HF Models#

Model

HF ID

Qwen1.5 MoE A2.7B

Qwen/Qwen1.5-MoE-A2.7B

Qwen1.5 MoE A2.7B Chat

Qwen/Qwen1.5-MoE-A2.7B-Chat

Example Recipes#

Recipe

Description

qwen1_5_moe_a2_7b_qlora.yaml

QLoRA — Qwen1.5 MoE A2.7B

Try with NeMo AutoModel#

1. Install (full instructions):

pip install nemo-automodel

2. Clone the repo to get the example recipes:

git clone https://github.com/NVIDIA-NeMo/Automodel.git
cd Automodel

3. Run the recipe from inside the repo:

automodel --nproc-per-node=8 examples/llm_finetune/qwen/qwen1_5_moe_a2_7b_qlora.yaml
Run with Docker

1. Pull the container and mount a checkpoint directory:

docker run --gpus all -it --rm \
  --shm-size=8g \
  -v $(pwd)/checkpoints:/opt/Automodel/checkpoints \
  nvcr.io/nvidia/nemo-automodel:26.02.00

2. Navigate to the AutoModel directory (where the recipes are):

cd /opt/Automodel

3. Run the recipe:

automodel --nproc-per-node=8 examples/llm_finetune/qwen/qwen1_5_moe_a2_7b_qlora.yaml

See the Installation Guide and LLM Fine-Tuning Guide.

Fine-Tuning#

See the LLM Fine-Tuning Guide and the Large MoE Fine-Tuning Guide.

Hugging Face Model Cards#