Papers · Proving inference · Artemis / Apollo
Artemis / Apollo
Artemis: Efficient Commit-and-Prove SNARKs for zkML
Reported benchmarks
| model | params | proving time s | proof size mb | verification time s |
|---|---|---|---|---|
| GPT-2 | 124M | 12000–14400 | 0.015 | 0.014 |
proving_time_note: 200-240 minutes
Notes
Attacks a different bottleneck: consistency checks on committed model params and inputs. Cuts commitment-verification overhead on VGG from 11.5x to 1.1x. Apollo is the KZG / white-box variant; Artemis is generic and works in transparent Halo2/IPA settings.
Our reading
Citation neighbourhood
Builds on
- none recorded
Cited by, in this corpus
Edges are a proxy: paper A's text mentions B anywhere (body or bibliography). See the full graph.
Discussed in
The hybrid cellThe prior surveys, and where we disagree with themSystem by systemThe landscapeProof systemsWhat is actually being provenProving a scoreThe hardest claim in the table
Other recorded fields
claim_kind: pass objectives: - inference - testing