GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
BRZ911
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
BRZ911
Model Tree
Input Modalities
Output Modalities
Supported Functionality
| Resource | Link |
|---|---|
| Code repository | https://github.com/BRZ911/Latent-VC |
| Training & evaluation data | https://huggingface.co/datasets/BRZ911/Latent-VC-Data |
| Model weights (this repo) | https://huggingface.co/BRZ911/Latent-VC-9B |
Qwen3_5ForConditionalGeneration (Qwen3.5-VL backbone)lvc_temperature=0.07, loss_lvc_fct=cosine)checkpoint-800.| File | Description |
|---|---|
model.safetensors | Model weights (~18.8 GB) |
config.json | Model configuration |
generation_config.json | Generation configuration |
tokenizer.json, tokenizer_config.json | Tokenizer |
processor_config.json, chat_template.jinja | Processor & chat template |
This model uses custom LVC components. For training and LVC reasoning inference, please use the code from the Latent-VC repository.
bash
pip install -U "huggingface_hub[cli]"hf download BRZ911/Latent-VC-9B --local-dir ./Latent-VC-9B
The included scripts evaluate the model on six video benchmarks (VideoMME, MVBench, TempCompass, VideoMMMU, VSIBench, MMVU). The evaluation data is hosted in the companion dataset BRZ911/Latent-VC-Data under Eval/ — download and extract it so that an Evaluation/ directory (with eval_<dataset>.json and the per-benchmark videos) is available, then run:
bash
# MODEL_PATH : path to the downloaded weights (this repo)# EVAL_DIR : path to the extracted Evaluation/ directory (default: ./Evaluation)# DATASETS : comma-separated subset of# videomme,mvbench,tempcompass,videommmu,vsibench,mmvuMODEL_PATH=./Latent-VC-9B \EVAL_DIR=./Evaluation \DATASETS=mmvu \bash eval_all_benchmarks.sh
See the Latent-VC repository for the full inference/training environment and dependencies.
The model is evaluated on the following benchmarks (data in BRZ911/Latent-VC-Data):
| Benchmark | Samples |
|---|---|
| VideoMME | 2700 |
| MVBench | 4000 |
| TempCompass | 7540 |
| VideoMMMU | 900 |
| VSIBench | 5130 |
| MMVU | 625 |
If you find this model or dataset useful, please cite the project:
bibtex
@misc{latentvc,title = {Latent-VC: Latent Visual Compression for Efficient Video Reasoning},author = {BRZ911},year = {2025},url = {https://github.com/BRZ911/Latent-VC}}
eval_all_benchmarks.py | Multi-benchmark evaluation script |
eval_all_benchmarks.sh | Evaluation launcher |