An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics
Updated 45 min ago · first seen 11 Sept 2026
paper_01M294WYE1A9ZBCTMZPP6AFRR5
- Published
- 9 Sept 2026
- T2 · 48 min ago
- arXiv
- 2609.10712
- T2 · 48 min ago
Abstract
We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.
Authors 6
Ivan Moshkov, Stephen Ge, George Armstrong, Wei Du, Sadegh Mahdavi, Igor Gitman
Specification
- Official page
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 48 min agomedium
- arXiv id
- 2609.10712
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 48 min agomedium
- Hf paper url
- https://huggingface.co/papers/2609.10712
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 48 min agomedium
- Hf comments
- 0
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 45 min agomedium
- Upvotes
- 5
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 45 min agomedium
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 48 min agomedium
- Published
- 9 Sept 2026
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 48 min agomedium
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
9
Source tiers
T29
Freshest observation
45 min ago
Conflicts
None
No models linked to this paper yet.
No relations recorded.
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history
Official pageofficial_url1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| https://arxiv.org/abs/2609.10712 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Abstractabstract1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems. | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
arXiv idarxiv_id1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 2609.10712 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Hf paper urlhf_paper_url1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| https://huggingface.co/papers/2609.10712 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Hf commentsmetric.hf_comments1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 0 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Upvotesmetric.upvotes1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 5 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
PDFpdf_url1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| https://arxiv.org/pdf/2609.10712 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Publishedpublished_at1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 9 Sept 2026 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
New paper: An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics
huggingface
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/papers | listing | T2· Quality secondary | 45 min ago | 2 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.