Add figures to the card

.gitattributes

@@ -48,3 +48,4 @@ training_contrasts/no_templates_8k/run/PLAN.json filter=lfs diff=lfs merge=lfs -
 training_contrasts/site_encoder_only_8k/run/PLAN.json filter=lfs diff=lfs merge=lfs -text
 training_contrasts/summaries_only_88k/run/PLAN.json filter=lfs diff=lfs merge=lfs -text
 training_contrasts/summaries_only_8k/run/PLAN.json filter=lfs diff=lfs merge=lfs -text
+figures/model_training.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -34,6 +34,7 @@ networks the paper compares it with.
 | `final/snapshots/epoch06/`, `epoch09/` | states saved after epochs 6 and 9 | 47 MB each |
 | `training_contrasts/` | 7 networks trained on 8,000 or 88,000 windows | 0.36 GB |
 | `ablations/` | 5 networks trained on 12,500 windows | 0.25 GB |
+| `figures/` | the figure on this card | small |
 | `SHA256SUMS` | a checksum for every file except this card | small |
 
 ## The network
@@ -59,6 +60,14 @@ baseline curve: the window's Watterson estimate of log10 Ne plus a fixed offset
 - **Test error.** 0.287: the root mean squared error of log10 Ne over all 64 intervals of the
   500 test windows taken together, from `final/test_predictions.npz`.
 
+![Training and test error](https://huggingface.co/Larrivhan/NeCast/resolve/main/figures/model_training.png)
+
+**a**, Training error and validation error after each of the 12 epochs, from
+`final/DEVELOPMENT.json`; the validation windows are the 500 further training windows. **b**,
+Test error in each of the 64 intervals (5-interval running mean), with the error over all
+intervals as a dotted line. **c**, The test windows at the 10th, 50th and 90th percentiles of
+per-window error: truth (black step) and NeCast (blue).
+
 ## Running it
 
 The network reads inputs built by the NeCast code (summary statistics, PSMC+ posteriors and the

SHA256SUMS

@@ -73,6 +73,7 @@ f5d8b30b1d297929b731ea7a01e1ac6c71b1fbedf22a36ad7616dbf8d2a0ea09  ablations/summ
 c6fdaf5df9e564232d8b28929d1a749288d926b65dc45ee15015aaf5b1c472fa  ablations/summaries_only_12500/run/summaries_only/checkpoint.pt
 34f1a287906e64937cdf73fa481bb7164dbc26ac82beb592a23c8ff890713a9b  ablations/summaries_only_12500/run/summaries_only/mechanism_evaluation.npz
 fb5ec87b84a657081926b218cef0fa59455d3ef411c6ff913bd8e45448bdc82b  ablations/summaries_only_12500/run/summaries_only/report.json
+d0b8d9dcd1315e62716d41b2ebbceb7318b055593dcf2a114d274923138779c0  figures/model_training.png
 49a40ada70aaf3390cbc15945ee38c28edd1e122760f7851d0769c098263f6a8  final/DEVELOPMENT.json
 39caf385d0ffcf1c243a4c7369c849a8ec3fc65e643e6038606fac0497e49cc7  final/DONE.json
 ce8c2e983648b28bb997f8a7c4d26cc239c7379a0fbf81371d7becdac0114e76  final/INTAKE_VERIFY.json