Video · text-to-image · text-to-video · text-image-to-video · video-refinement

LingBot-Video

robbyant

The inference package is flattened directly under worldfoundry/synthesis/visual_generation/lingbot_video.

Integrated独立环境lingbot-video

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-image

环境lingbot-video

设备CUDA 12.8

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

兼容性与版本

由 Manifest 生成

集成状态: integrated
Runner 证据: pending
环境: lingbot-video
Python: 3.10
CUDA: CUDA 12.8
PyTorch: torch==2.12.0.dev20260220+cu130
源码 revision: —
Checkpoint revision: f9789a7d9b4772a47aba62d4eb5282ddefd1da21

Runtime profile: lingbot-video
Pipeline binding: lingbot-video
Runner: worldfoundry.pipeline
Pipeline target: worldfoundry.pipelines.lingbot_video.pipeline_lingbot_video:LingBotVideoGenerationPipeline
Backend stage: in_tree_runtime
Runtime 状态: dense_quality_and_moe_quality_smoke_verified_a100
Driver 状态: requires_cuda_13_compatible_driver
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 lingbot-video

依赖版本约束 13

torch==2.12.0.dev20260220+cu130
torchvision==0.26.0.dev20260220+cu130
accelerate>=1.8.0
decord>=0.6.0
diffusers==0.39.0
imageio>=2.35.0
imageio-ffmpeg>=0.5.1
numpy>=1.26
peft==0.19.1
pillow>=10.4.0
safetensors>=0.4.5
scipy>=1.11
transformers==5.8.1

Conda 依赖 3

python=3.10
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 lingbot-video --check-local --json

robbyant/lingbot-video-dense-1.3b

Revision: —
License: apache-2.0
Gated: —
Private: —

robbyant/lingbot-video-moe-30b-a3b

Revision: —
License: apache-2.0
Gated: —
Private: —

robbyant/lingbot-video-dense-1.3b

Revision: f9789a7d9b4772a47aba62d4eb5282ddefd1da21
License: —
Gated: —
Private: —

robbyant/lingbot-video-moe-30b-a3b

Revision: f2e538f64afe00cc4ae674db2aeb52e2945edfd5
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 lingbot-video \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/lingbot-video \
  --metric artifact_count \
  --json

输入契约

字段	已记录契约
`prompt`	Required
`image`	Required
`prompt_json`	Required

Artifact 契约

Artifact 类型	文件名 / 路径
`generated_image`	`—`
`generated_video`	`—`
`generated_video`	`lingbot_video.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: integrated
Runner 证据: pending
原生 Demo 证据: 未记录
验证 Imports: torch, torchvision, diffusers, transformers, accelerate, worldfoundry.synthesis.visual_generation.lingbot_video.inference_backend

配方溯源

worldfoundry/data/models/catalog/video/lingbot-video.yamlProjecthttps://technology.robbyant.com/lingbot-videoGitHubhttps://github.com/Robbyant/lingbot-videoHugging Facehttps://huggingface.co/robbyant/lingbot-video-dense-1.3bHugging Facehttps://huggingface.co/robbyant/lingbot-video-moe-30b-a3b

构建运行命令

输入契约

Artifact 契约

相关配方