World models · text-to-video · image-to-video · text-image-to-video · embodied-world-model

Kairos Sensenova

Embodied World Model

Uses the in-tree Apache-2.0 official torchrun runtime under worldfoundry/synthesis/visual_generation/kairos/kairos_runtime.

Runner verified独立环境kairos-sensenova

构建运行命令

选择仓库中已记录的 variant，然后复制对应的准备、检查或运行命令。

worldfoundry.pipeline

Variant

任务text-to-video

环境kairos-sensenova

设备CUDA 12.8

worldfoundry-eval evaluate \
  --mode model \
  --model-id kairos-sensenova \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/kairos-sensenova \
  --metric artifact_count \
  --json

兼容性与版本

由 Manifest 生成

集成状态: verified
Runner 证据: verified
环境: kairos-sensenova
Python: 3.10
CUDA: CUDA 12.8
PyTorch: torch==2.12.0
源码 revision: —
Checkpoint revision: —

Runtime profile: kairos-sensenova
Pipeline binding: kairos-sensenova
Runner: worldfoundry.pipeline
Pipeline target: worldfoundry.pipelines.kairos.pipeline_kairos:KairosPipeline
Backend stage: in_tree_official_runtime
Runtime 状态: runnable
Driver 状态: compatible
环境类型: 独立环境

安装环境

环境解析器会读取已记录 profile，并选择此处显示的统一或独立环境。

bash scripts/setup/model_env_install.sh --model kairos-sensenova

依赖版本约束 14

torch==2.12.0
torchvision
einops
flash-linear-attention
huggingface-hub
imageio
imageio-ffmpeg
mmengine
modelscope
numpy
numpydantic
pillow
transformers
autoawq

Conda 依赖 3

python
pip
ffmpeg

Checkpoint 与资产

分配算力前先做本地检查；gated、private 与 license 字段直接来自 checkpoint manifest。

worldfoundry-eval zoo model-download --model-id kairos-sensenova --check-local --json

kairos-agi/kairos-sensenova-robot-4B-480P-distilled

Revision: —
License: —
Gated: —
Private: —

kairos-agi/kairos-sensenova-4B-480P-pretrained

Revision: —
License: —
Gated: —
Private: —

kairos-agi/kairos-sensenova-4B-720P

Revision: —
License: —
Gated: —
Private: —

运行与输出

生成的命令通过共享 evaluation 边界运行，并持久保存结果 manifest 与 artifact。

worldfoundry-eval evaluate \
  --mode model \
  --model-id kairos-sensenova \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/kairos-sensenova \
  --metric artifact_count \
  --json

输入契约

字段	已记录契约
`prompt`	Required
`image`	Required
`video`	Optional
`actions`	seed

Artifact 契约

Artifact 类型	文件名 / 路径
`generated_video`	`kairos-sensenova.mp4`

证据与来源

Catalog 集成、原生 demo parity 与 runner parity 是三条独立记录。

集成状态: verified
Runner 证据: verified
原生 Demo 证据: 未记录
验证 Imports: torch, mmengine, transformers, imageio, fla, apex, sageattention

配方溯源

worldfoundry/data/models/catalog/world_models/kairos-sensenova.yamlGitHubhttps://github.com/kairos-agi/kairos-sensenovaHugging Facehttps://huggingface.co/kairos-agi/kairos-sensenova-4B-480P-pretrainedHugging Facehttps://huggingface.co/kairos-agi/kairos-sensenova-robot-4B-480P-distilledHugging Facehttps://huggingface.co/kairos-agi/kairos-sensenova-4B-720P

构建运行命令

输入契约

Artifact 契约

相关配方