🌍 A minimalist repository for training video world models based on diffusion-forcing. • 20 items • Updated • 7
NanoWM-L/2 · csgo · NanoWM-L/2 on CSGO
NanoWM-L/2 on CSGO (100k steps)
Run identity
- wandb run 1 (steps 0–50000): https://wandb.ai/better_guidance/nano-world-model-phase2/runs/cosr7ivu
- wandb run 2 (steps 50000–100000): https://wandb.ai/better_guidance/nano-world-model-phase2/runs/zr3k1djc
- launcher:
src/scripts/train/main/csgo.sh(initial),src/scripts/train/main/csgo_continued.sh(resume) - collection: https://huggingface.co/collections/knightnemo/nano-world-model
Training setup
| Key | Value |
|---|---|
| Architecture | NanoWM-L/2 |
| Dataset | csgo |
| Prediction | v |
| Noise schedule | squaredcos_cap_v2 (ZTSNR=True) |
| Steps | 100000 |
| Batch | 6/GPU |
Loading
import sys
from omegaconf import OmegaConf
from safetensors.torch import load_file
sys.path.insert(0, "src")
from models import get_models
cfg = OmegaConf.load("ckpt/config.yaml")
cfg.experiment.infra.compile = False
model = get_models(cfg).eval()
state_dict = load_file("ckpt/model.safetensors")
model.load_state_dict(state_dict, strict=True)
- Downloads last month
- 11
Safetensors
Model size
0.6B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support