GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
fudan-generative-ai
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
fudan-generative-ai
Model Tree
Qwen/Qwen3-1.7B
Input Modalities
Output Modalities
Supported Functionality
| Property | Value |
|---|---|
| Role | Diffusion draft model; not the target verifier |
| Target verifier | Qwen/Qwen3-32B |
| Draft base architecture | Qwen3-1.7B |
| Transformer layers | 24 |
| Hidden size | 2048 |
| Vocabulary size | 151936 |
| Stored weight dtype | FP32 (original checkpoint tensors, no conversion) |
| Recommended inference dtype | BF16 |
Pruned to 24 transformer layers from Qwen3-1.7B.
The size in the model name describes the base draft family. The stored checkpoint
can include untied embedding/output weights, and pruned variants have fewer layers;
the exact tensor parameter count is reported in release_metadata.json.
Install the dependencies in the D3D repository, then run from its root:
bash
export TARGET_MODEL=Qwen/Qwen3-32Bexport DRAFT_MODEL=fudan-generative-ai/D3D-24D-p1.7B-32Bbash scripts/quickstart_single.shFor concurrent requests:
bash
CONCURRENCY=4 bash scripts/quickstart_concurrent.shThe target runs on GPU 0 and the draft on GPU 1. D3D uses its checked-in D3D compatibility classes and does not execute Hub-hosted Python during inference. Use the pinned D3D environment (PyTorch 2.8.0 / Transformers 4.57.3).