inference

# Alpamayo R1 Inference Use this sub-skill when the user wants to run, inspect, or adapt the Alpamayo R1 inference path on a PhysicalAI-AV clip. ## What this covers - Load a gated PhysicalAI-AV clip and egomotion history/future. - Build the multimodal chat prompt from stacked camera frames. - Load `nvidia/Alpamayo-R1-10B` on CUDA with the Alpamayo tokenizer and processor. - Sample future trajectories plus Chain-of-Causation text traces. - Compare predicted trajectories with ground truth and visualize them in notebook style. - Troubleshoot HF gating, CUDA OOM, flash-attn / SDPA fallback, device placement, prompt rank, and early `t0_us` failures. ## What this does not cover - Training, SFT, RL post-training, or repo maintenance. - Import/export plumbing or package provenance. - Any workflow that depends on the original checkout at runtime. ## Fast route 1. Read `references/api-reference.md` for the public call contract. 2. Read `references/data-formats.md` for shapes, frames, and output semantics. 3. Follow `references/workflows.md` for the end-to-end inference flow and notebook-style visualization. 4. If anything fails, open `references/troubleshooting.md`. 5. Run `scripts/run_inference_smoke.py` as the bundled smoke test. ## Key defaults - Default model id: `nvidia/Alpamayo-R1-10B` - Default loader history / future / camera settings: 16 steps, 64 steps, 4 cameras - Default attention path: `flash_attention_2` - SDPA is a fallback for flash-attn incompatibility, not a replacement for the CUDA path. ## See also - `../../SKILL.md` for the root router - `../../references/repo-provenance.md` - `../../references/repo-routing-metadata.json` - `../../references/troubleshooting.md` for shared Alpamayo R1 troubleshooting - `references/api-reference.md` - `references/data-formats.md` - `references/workflows.md` - `references/troubleshooting.md` - `scripts/run_inference_smoke.py`