Routes 3D ResNets PyTorch video action-recognition workflows across training, inference, and data preparation.
日本語の概要は準備中です。原文の説明を表示しています。
Use NASA's apod-api Flask service and standalone APOD parser for date, count, range, media, deployment, Docker, and troubleshooting workflows.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Use this repo skill when a task involves NASA's Astronomy Picture of the Day
(APOD) service, its /v1/apod/ Flask endpoint, the standalone
apod_object_parser helper, APOD HTML/media extraction, or running the service
locally or in Docker. This is an operating guide for the public apod-api
repository, not a replacement NASA API credential store.
Keep the routes distinct: the Flask service scrapes APOD website HTML, while
apod_object_parser.get_data(api_key) calls NASA's separate JSON API. Do not
pass NASA JSON API parameters such as api_key to the Flask endpoint; the
implemented Flask handler rejects unknown fields.
apod-api version 1.1.0, Python >=3.12.GET /v1/apod/ with a trailing slash. It supports one
date, a random count from 1–100, or an inclusive start_date/end_date
range, plus concept_tags, thumbs, and legacy hd flags.service_version,
msg, and code.https://apod.nasa.gov/apod/.
Concept tagging additionally depends on a protected alchemy_api.key; its
absence is an expected degraded mode, not a reason to publish a secret.For a real repository checkout, install the declared runtime dependencies and then run the bundled route inspector from this skill:
python sub-skills/api-service/scripts/run_service.py --help
python sub-skills/api-service/scripts/run_service.py --inspect-routes
The inspector imports the application and prints its routes without scraping
APOD. Keep the application working directory able to resolve its templates and
static assets. If the project metadata attempts an editable build and reports
multiple top-level packages, follow the dependency-only workaround in
installation
rather than claiming that pip install -e . succeeded.
Before applying this guide to a changed checkout, read repo provenance. If the commit, package version, dirty paths, or public entry points differ, use a refresh workflow instead of assuming these facts remain current.
Do not put NASA keys, Alchemy credentials, or private deployment values in commands, fixtures, Docker layers, or generated reports. Treat live endpoint requests, image downloads, HTML scraping, and Locust as external-network operations with explicit timeouts and bounded scope. Do not run a load test as an ordinary startup check. The review/test artifacts for this skill live outside this runtime tree and are not needed by a future Researcher.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Routes 3D ResNets PyTorch video action-recognition workflows across training, inference, and data preparation.
日本語の概要は準備中です。原文の説明を表示しています。
Guide 3DDFA Python inference, geometry rendering, training/evaluation, and optional C++ ONNX workflows for 3D dense face alignment.
日本語の概要は準備中です。原文の説明を表示しています。
Routes 3DDFA_V2 face-alignment setup, still-image demos, video tracking, and ONNX benchmarking workflows.
日本語の概要は準備中です。原文の説明を表示しています。
Operate AB3DMOT 3D multi-object tracking workflows for KITTI and nuScenes data, tracking, evaluation, and visualization.
日本語の概要は準備中です。原文の説明を表示しています。
Use Hugging Face Accelerate for PyTorch training-loop migration, distributed launch/configuration, DeepSpeed/FSDP/TPU backend setup, big-model inference/offload, checkpointing, tracking, and troubleshooting.
日本語の概要は準備中です。原文の説明を表示しています。
Route Acme reinforcement-learning framework tasks across core loops, replay/data, JAX agents, and TensorFlow/Sonnet agents.
日本語の概要は準備中です。原文の説明を表示しています。