|
Download EMA_WEIGHTS.md from nvidia/cbottle: direct link, hf CLI and curl.
- Browser
- Download file 9.55 kB
-
https://huggingface.co/nvidia/cbottle/resolve/main/EMA_WEIGHTS.md
- Command line
-
hf download hf://nvidia/cbottle/EMA_WEIGHTS.md
-
curl -L -o EMA_WEIGHTS.md https://huggingface.co/nvidia/cbottle/resolve/main/EMA_WEIGHTS.md
9.55 kB
| # EMA checkpoints (`cBottle-3d-ema/`, `cBottle-3d-tc-ema/`) | |
| Exponential-moving-average (EMA) checkpoints for the `cbottle-3d-moe` denoiser | |
| and its paired `cbottle-3d-moe-tc` tropical-cyclone classifier, released to | |
| reproduce the results in **"Towards accurate extreme event likelihoods from | |
| diffusion model climate emulators"** (egusphere-2026-2610). These are the | |
| weights used for every figure and number in the revised manuscript; the | |
| originally submitted version used the raw (non-EMA) optimizer weights in | |
| `cBottle-3d/` and `cBottle-3d-tc/` instead. | |
| Inference code is public at [NVlabs/cBottle](https://github.com/NVlabs/cBottle) | |
| (`scripts/inference_coarse.py` for AMIP-style sampling, | |
| `scripts/inference_odds_ratio.py` for the guided-sampling / importance-sampling | |
| pipeline used in the likelihood paper). | |
| ## Contents | |
| - `cBottle-3d-ema/ema-checkpoint-000512000.checkpoint` β MoE expert A (Ο β₯ 100 active range), 512k images seen | |
| - `cBottle-3d-ema/ema-checkpoint-002048000.checkpoint` β MoE expert B (10 β€ Ο < 100), 2048k images seen | |
| - `cBottle-3d-ema/ema-checkpoint-009856000.checkpoint` β MoE expert C (Ο < 10), 9856k images seen | |
| - `cBottle-3d-tc-ema/ema-checkpoint-002176000.checkpoint` β paired TC classifier, 2176k images seen | |
| Same training run and same image counts (`nimg`) as `cBottle-3d/` and | |
| `cBottle-3d-tc/` β these are an EMA-averaged view of the identical training | |
| run, not a separately trained model. Do not mix these EMA weights with the | |
| classifier from the raw-weight run, or vice versa: the likelihood-paper | |
| reproduction used the EMA classifier throughout, paired only with the EMA MoE | |
| denoiser. | |
| ## Why EMA weights | |
| The manuscript's original submission used `net_state.pth` from the raw | |
| training-state checkpoints (the optimizer's live weights, i.e. what's in | |
| `cBottle-3d/` and `cBottle-3d-tc/`). A later evaluation found the EMA weights | |
| close part of the gap between cBottle's tropical-cyclone climatology and | |
| ERA5's, at no cost elsewhere. We gated the switch by re-running the original | |
| [cBottle paper](https://github.com/NVlabs/cBottle)'s own AMIP evaluation suite | |
| (spectra, seasonal cycle, precipitation PDF, variance maps, global-mean | |
| biases, Arctic Oscillation, ENSO teleconnection, TC frequency) comparing raw | |
| vs. EMA weights against ERA5. All figured evidence below is from that gate; | |
| diagnostics we evaluated but aren't shown as figures here are called out | |
| explicitly rather than asserted without a plot. | |
| ### Validation loss / log-probability | |
| 80 held-out samples per checkpoint β EMA is equal to or better than the raw | |
| weights at all three checkpoint steps, with the largest gain at the | |
| least-trained checkpoint (512k): | |
|  | |
| Breaking this down by noise level confirms EMA is not worse anywhere, and is | |
| consistently at least as good within each expert's own MoE-active Ο range | |
| (shaded): | |
|  | |
| ### Tropical cyclone frequency | |
| AMIP inference, 1980β2000, TempestExtremes detections, same pipeline as the | |
| original cBottle paper β EMA closes roughly half the gap to ERA5 without any | |
| guidance or classifier involved: | |
| | | detections/year | ΟΜ (detections vs. ERA5 ratio) | | |
| |---|---|---| | |
| | ERA5 (truth) | 1007 | 1.3 | | |
| | cBottle raw MoE | 707 | 0.95 | | |
| | cBottle EMA MoE | **873** | **1.2** | | |
|  | |
| ### Global-mean biases | |
| This is the most notable improvement outside of TC frequency. Global-mean, | |
| deseasonalized anomaly time series (1980β2000, shading = ensemble spread) | |
| against ERA5 (black): | |
| **Precipitation** β the raw-weight model has a persistent **~+0.3 mm/d global | |
| bias** relative to ERA5 that does not appear in the EMA weights, which track | |
| ERA5's inter-annual variability directly instead of just running parallel to | |
| it offset upward: | |
| | raw weights | EMA weights | | |
| |---|---| | |
| |  |  | | |
| **Z500** β similarly, the raw-weight model runs persistently above ERA5 for | |
| most of the 1980s and misses the 1984β85 dip entirely; the EMA weights track | |
| ERA5 much more closely throughout, including that dip: | |
| | raw weights | EMA weights | | |
| |---|---| | |
| |  |  | | |
| **2 m temperature** β both track ERA5's inter-annual variability similarly | |
| well; no meaningful difference between raw and EMA here: | |
| | raw weights | EMA weights | | |
| |---|---| | |
| |  |  | | |
| ### Modes of variability (no regression) | |
| **Arctic Oscillation** β daily AO index against ERA5, train-period CRPS is | |
| identical (5.0) for both raw and EMA weights; switching to EMA does not | |
| change AO skill: | |
| | raw weights | EMA weights | | |
| |---|---| | |
| |  |  | | |
| **ENSO precipitation teleconnection** β regression of precipitation anomaly | |
| onto the ENSO index, against ERA5 (a); raw (b, left) and EMA (b, right) both | |
| reproduce the tropical Pacific dipole pattern with no visible difference | |
| between them: | |
| | raw weights | EMA weights | | |
| |---|---| | |
| |  |  | | |
| ### Precipitation PDF and variance maps | |
| Effectively unchanged, and if anything marginally closer to ERA5 truth for | |
| precipitation and near-surface temperature RMS: | |
|  | |
|  | |
| ### Power spectra | |
| Spherical-harmonic power spectra against ERA5 truth (black) for Z500, pr, | |
| T850, tcwv, tas, and uas, raw weights (dashed) and EMA weights (teal) overlaid | |
| directly. The two are visually indistinguishable across the full wavenumber | |
| range for all six variables β EMA does not shift spectral content at any | |
| scale: | |
|  | |
| ### Seasonal cycle | |
| DJF-minus-JJA seasonal contrast maps for Z500, precipitation, 2 m temperature, | |
| and sea-ice concentration β truth (top row), raw weights (middle row), EMA | |
| weights (bottom row). All three rows are qualitatively very similar, | |
| including the Southern Ocean sea-ice edge and the Southern Hemisphere storm | |
| track pattern in panel (d)/(h)/(l): | |
|  | |
| **Conclusion:** EMA weights are climatologically as good as or better than | |
| the raw weights on every metric checked, with clear improvements in TC | |
| frequency and in the global-mean precipitation and Z500 biases, and no | |
| regression in validation loss, power spectra, seasonal cycle, precipitation | |
| PDF, variance maps, Arctic Oscillation skill, or ENSO teleconnection. This is | |
| why the revision uses EMA weights throughout. | |
| ## Usage | |
| Unconditional / AMIP-forced sampling (reproduces the climatology plots above): | |
| ```bash | |
| python3 scripts/inference_coarse.py \ | |
| --dataset amip --sample.mode sample --sample.sigma_max 200 \ | |
| --sample.sampler all --sample.batch_gpu 8 --sample.bf16 \ | |
| --end_time 2017-12-31 \ | |
| cBottle-3d-ema/ema-checkpoint-000512000.checkpoint,cBottle-3d-ema/ema-checkpoint-002048000.checkpoint,cBottle-3d-ema/ema-checkpoint-009856000.checkpoint \ | |
| <output_dir> | |
| ``` | |
| Guided sampling / odds-ratio calculation for the TC likelihood paper (uses the | |
| classifier checkpoint too): | |
| ```bash | |
| python3 scripts/inference_odds_ratio.py \ | |
| --moe cBottle-3d-ema/ema-checkpoint-000512000.checkpoint,cBottle-3d-ema/ema-checkpoint-002048000.checkpoint,cBottle-3d-ema/ema-checkpoint-009856000.checkpoint \ | |
| --classifier cBottle-3d-tc-ema/ema-checkpoint-002176000.checkpoint \ | |
| ... # see scripts/inference_odds_ratio.py --help for the full sampler/guidance CLI | |
| ``` | |
| ## Provenance | |
| - Converted from EMA network snapshots (`network-snapshot-*.pkl`) to | |
| inference-ready checkpoints via | |
| `scripts/cmip_eval_scripts/convert_ema_snapshot_to_checkpoint.py` in the | |
| cBottle training repo. | |
| - Denoiser: three-expert mixture (`SongUNet`, HEALPix HPX64), each expert | |
| responsible for a disjoint noise-level range (Ο<10, 10β€Ο<100, Οβ₯100). | |
| - Classifier: `cBottle-3d-tc`, a `SongUNet` with an HPX3-resolution | |
| classification head, trained jointly with a denoising objective, binary | |
| cross-entropy term (weight 0.1) on IBTrACS-derived tropical-cyclone labels. | |
| ## License | |
| Same terms as the rest of this repository: the | |
| [NVIDIA Software and Model Evaluation License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-software-and-model-evaluation-license/). | |
| ## Citation | |
| ```bibtex | |
| @article{manshausen2026towards, | |
| title={Towards accurate extreme event likelihoods from diffusion model climate emulators}, | |
| author={Manshausen, Peter and Brenowitz, Noah and Berner, Julius and Kashinath, Karthik and Pritchard, Mike}, | |
| journal={arXiv preprint arXiv:2605.03802}, | |
| year={2026} | |
| } | |
| ``` | |
| This is the arXiv preprint; update to the camera-ready EGUsphere/journal citation once egusphere-2026-2610 is formally published. | |