GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
GoForceX
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
GoForceX
Model Tree
Qwen/Qwen3.5-0.8B
Input Modalities
Output Modalities
Supported Functionality
This repository contains a merged Hugging Face checkpoint: the frozen base-model
components plus the fine-tuned language_model and lm_head from best-model-stage2-0250.
It was exported for local inference / GGUF conversion. The decision task is not a normal free-form text-generation benchmark: inference should reproduce the same sentence-only prompt and score the six rating candidate tokens (1..5 quality, 6=skip).
Calibrated rating temperature: 1.409963883056206 (see temperature.json).
For GGUF conversion, use a recent llama.cpp checkout with Qwen3.5 support and first create an F16 GGUF as the fidelity reference. Re-fit/validate temperature for quantized variants.