riprap-nyc / app /inference.py

Commit History

feat(emissions): real GPU power from NVML on the L4 inference Space
c368c91

seriffic Claude Opus 4.7 (1M context) commited on

fix(emissions): default hardware to NVIDIA L4
d48454d

seriffic Claude Opus 4.7 (1M context) commited on

feat(emissions): per-call inference energy + token tracker
b84be35

seriffic Claude Opus 4.7 (1M context) commited on

feat: terramind_synthesis now routes through droplet remote inference
eea4d6e

seriffic Claude Opus 4.7 (1M context) commited on

deploy: sync all changes from main at 6904684
b9a10ad

seriffic Claude Sonnet 4.6 commited on

feat: route all GPU-accelerable inference through MI300X (Phase 1+2 of full GPU)
abcf7cd

seriffic Claude Opus 4.7 (1M context) commited on