GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
yifanouyang
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
yifanouyang
Model Tree
Input Modalities
Output Modalities
Supported Functionality
The GitHub toolkit supports resumable JSONL completion, duplicate-ID and settings checks, CPU/CUDA/MPS selection and offline loading. Standard Qwen3 architecture; no Hub Python execution is required.
A domain-adaptation trainer saves separate checkpoints. The training pipeline was smoke-tested with a tiny local Qwen3. Released 1.7B weights are unchanged; full-checkpoint inference and benchmark evaluation have not been validated here.
Outputs may repeat or contain incorrect information. CPU FP32 inference requires substantially more memory than the checkpoint size; full-parameter training also needs gradients and optimizer states.