main

GGitHubfix(examples): replace FilesystemBackend with upload_files in nvidia example (#1973)

273c31d0创建于 3月19日历史提交

文件	最后提交记录	最后更新时间
AGENTS.md	feat(examples): add nvidia deep agent (#1822) Add a multi-model deep research agent example built on NVIDIA's AIQ deep research pattern. Uses Claude for orchestration and NVIDIA Nemotron for web research, with GPU-accelerated data analysis via Modal sandboxes and NVIDIA RAPIDS (cuDF/cuML). - Multi-model architecture: Frontier model (orchestrator + data processor) + Nemotron Super (researcher) - GPU sandbox execution via Modal with RAPIDS for accelerated analytics and ML - Skill-based agent memory with self-improving knowledge files - Tavily web search with full page content fetching	2 个月前
__init__.py	feat(examples): add nvidia deep agent (#1822) Add a multi-model deep research agent example built on NVIDIA's AIQ deep research pattern. Uses Claude for orchestration and NVIDIA Nemotron for web research, with GPU-accelerated data analysis via Modal sandboxes and NVIDIA RAPIDS (cuDF/cuML). - Multi-model architecture: Frontier model (orchestrator + data processor) + Nemotron Super (researcher) - GPU sandbox execution via Modal with RAPIDS for accelerated analytics and ML - Skill-based agent memory with self-improving knowledge files - Tavily web search with full page content fetching	2 个月前
agent.py	chore(examples): update nvidia model (#1855)	2 个月前
backend.py	fix(examples): replace FilesystemBackend with upload_files in nvidia example (#1973) Closes #1862 --- Remove FilesystemBackend and CompositeBackend from the nvidia deep agent example. The old pattern routed skill/memory reads to the local filesystem, which isn't representative of production — skills don't originate from the host machine in a real deployment. The backend now uploads skill and memory files directly into the Modal sandbox on creation via upload_files, so the agent operates entirely within the sandbox.	2 个月前
prompts.py	feat(examples): add nvidia deep agent (#1822) Add a multi-model deep research agent example built on NVIDIA's AIQ deep research pattern. Uses Claude for orchestration and NVIDIA Nemotron for web research, with GPU-accelerated data analysis via Modal sandboxes and NVIDIA RAPIDS (cuDF/cuML). - Multi-model architecture: Frontier model (orchestrator + data processor) + Nemotron Super (researcher) - GPU sandbox execution via Modal with RAPIDS for accelerated analytics and ML - Skill-based agent memory with self-improving knowledge files - Tavily web search with full page content fetching	2 个月前
tools.py	feat(examples): add nvidia deep agent (#1822) Add a multi-model deep research agent example built on NVIDIA's AIQ deep research pattern. Uses Claude for orchestration and NVIDIA Nemotron for web research, with GPU-accelerated data analysis via Modal sandboxes and NVIDIA RAPIDS (cuDF/cuML). - Multi-model architecture: Frontier model (orchestrator + data processor) + Nemotron Super (researcher) - GPU sandbox execution via Modal with RAPIDS for accelerated analytics and ML - Skill-based agent memory with self-improving knowledge files - Tavily web search with full page content fetching	2 个月前