File size: 9,548 Bytes
368507e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
# 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 (Οƒ β‰₯ 100 active range), 512k images seen
- `cBottle-3d-ema/ema-checkpoint-002048000.checkpoint` β€” MoE expert B (10 ≀ Οƒ < 100), 2048k images seen
- `cBottle-3d-ema/ema-checkpoint-009856000.checkpoint` β€” MoE expert C (Οƒ < 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):

![EMA vs raw net, validation loss and log-prob](ema_gate_assets/ema_vs_raw_aggregate.png)

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 Οƒ range
(shaded):

![Denoising loss by noise level, raw vs EMA](ema_gate_assets/ema_vs_raw_by_sigma.png)

### Tropical cyclone frequency

AMIP inference, 1980–2000, 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 | ρ̄ (detections vs. ERA5 ratio) |
|---|---|---|
| ERA5 (truth) | 1007 | 1.3 |
| cBottle raw MoE | 707 | 0.95 |
| cBottle EMA MoE | **873** | **1.2** |

![TC detection density, ERA5 vs raw vs EMA](ema_gate_assets/TC_frequency_heatmap.png)

### Global-mean biases

This is the most notable improvement outside of TC frequency. Global-mean,
deseasonalized anomaly time series (1980–2000, 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 |
|---|---|
| ![Global-mean precip anomaly, raw](ema_gate_assets/global_mean_pr_net.png) | ![Global-mean precip anomaly, EMA](ema_gate_assets/global_mean_pr_ema.png) |

**Z500** β€” similarly, the raw-weight model runs persistently above ERA5 for
most of the 1980s and misses the 1984–85 dip entirely; the EMA weights track
ERA5 much more closely throughout, including that dip:

| raw weights | EMA weights |
|---|---|
| ![Global-mean Z500 anomaly, raw](ema_gate_assets/global_mean_Z500_net.png) | ![Global-mean Z500 anomaly, EMA](ema_gate_assets/global_mean_Z500_ema.png) |

**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 |
|---|---|
| ![Global-mean tas anomaly, raw](ema_gate_assets/global_mean_tas_net.png) | ![Global-mean tas anomaly, EMA](ema_gate_assets/global_mean_tas_ema.png) |

### 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 |
|---|---|
| ![AO index, raw](ema_gate_assets/ao_index_net.png) | ![AO index, EMA](ema_gate_assets/ao_index_ema.png) |

**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 |
|---|---|
| ![ENSO precip teleconnection, raw](ema_gate_assets/enso_pr_net.png) | ![ENSO precip teleconnection, EMA](ema_gate_assets/enso_pr_ema.png) |

### Precipitation PDF and variance maps

Effectively unchanged, and if anything marginally closer to ERA5 truth for
precipitation and near-surface temperature RMS:

![Precipitation PDF, ERA5 truth vs raw-gen vs EMA-gen](ema_gate_assets/precip_pdf.png)

![Global variance maps: Z500, precipitation, 2m temperature](ema_gate_assets/total_variance_maps.png)

### 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:

![Power spectra, ERA5 truth vs raw-gen vs EMA-gen, six variables](ema_gate_assets/spectra.png)

### 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):

![Seasonal cycle (DJF-JJA), ERA5 truth vs raw-gen vs EMA-gen](ema_gate_assets/seasonal_cycle.png)

**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 (Οƒ<10, 10≀σ<100, Οƒβ‰₯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.