Sukhil-Patel commited on
Commit
20a0bda
·
verified ·
1 Parent(s): ff50642

Upload models - 3z7g9npe

Browse files
Files changed (4) hide show
  1. README.md +3 -7
  2. data_config.yaml +8 -49
  3. model_config.yaml +3 -17
  4. model_weights.safetensors +2 -2
README.md CHANGED
@@ -14,7 +14,7 @@ Do not remove elements like the above surrounded by two curly braces and do not
14
  ## Model Description
15
 
16
  <!-- e.g.
17
- This model uses ECMWF deterministc NWP (HRES) & GDM's GenCast forecast, it forecasts up to 48 hours ahead at 15 min intervals and is trained on regional wind power data in NW India. More information can be found in the model repo [1]. The model repo also includes links to our workshop paper on this model and some experimental notes.
18
  -->
19
 
20
  - **Developed by:** openclimatefix
@@ -27,7 +27,7 @@ This model uses ECMWF deterministc NWP (HRES) & GDM's GenCast forecast, it forec
27
  ## Data
28
 
29
  <!-- eg.
30
- The model is trained on data from 2020-2022 and validated on data from 2023. This model uses ECMWF deterministc NWP (HRES) & GDM's GenCast forecast, which has its ensemble member data summarised into some aggregate stats e.g. mean, std & percentiles.
31
 
32
  See the data_config.yaml file for more information on the channels and window-size used for each input data source.
33
  -->
@@ -43,7 +43,7 @@ Data is prepared with the `ocf_data_sampler/torch_datasets/datasets/site` Datase
43
 
44
  <!-- Do not remove the lines below -->
45
  The training logs for this model commit can be found here:
46
- - [https://wandb.ai/openclimatefix/gdm-gen/runs/7itrv8nl](https://wandb.ai/openclimatefix/gdm-gen/runs/7itrv8nl)
47
 
48
 
49
  <!-- The hardware section is also just nice to have -->
@@ -62,7 +62,3 @@ This model was trained using the following Open Climate Fix packages:
62
  The versions of these packages can be found below:
63
  - pvnet==5.2.3
64
  - ocf-data-sampler==0.6.2.post0+git.accd7422.dirty
65
-
66
-
67
- ---
68
- **Migration Note**: This model was migrated on 2025-12-15 to pvnet version 5.3.0.post1+git.615e1fa3.dirty
 
14
  ## Model Description
15
 
16
  <!-- e.g.
17
+ This model uses ECMWF deterministc NWP (HRES) it forecasts up to 48 hours ahead at 15 min intervals and is trained on regional wind power data in NW India. More information can be found in the model repo [1]. The model repo also includes links to our workshop paper on this model and some experimental notes.
18
  -->
19
 
20
  - **Developed by:** openclimatefix
 
27
  ## Data
28
 
29
  <!-- eg.
30
+ The model is trained on data from 2020-2022 and validated on data from 2023. This model uses ECMWF deterministc NWP (HRES).
31
 
32
  See the data_config.yaml file for more information on the channels and window-size used for each input data source.
33
  -->
 
43
 
44
  <!-- Do not remove the lines below -->
45
  The training logs for this model commit can be found here:
46
+ - [https://wandb.ai/openclimatefix/gdm-gen/runs/3z7g9npe](https://wandb.ai/openclimatefix/gdm-gen/runs/3z7g9npe)
47
 
48
 
49
  <!-- The hardware section is also just nice to have -->
 
62
  The versions of these packages can be found below:
63
  - pvnet==5.2.3
64
  - ocf-data-sampler==0.6.2.post0+git.accd7422.dirty
 
 
 
 
data_config.yaml CHANGED
@@ -2,49 +2,15 @@ general:
2
  description: Example config for producing PVNet samples
3
  name: example_config
4
  input_data:
 
 
 
 
 
 
 
 
5
  nwp:
6
- ecmwf:
7
- provider: ecmwf
8
- zarr_path: PLACEHOLDER.zarr
9
- interval_start_minutes: 0
10
- interval_end_minutes: 2880
11
- time_resolution_minutes: 60
12
- channels:
13
- - temperature_sl
14
- - wind_u_component_10m
15
- - wind_v_component_10m
16
- - wind_u_component_100m
17
- - wind_v_component_100m
18
- - wind_u_component_200m
19
- - wind_v_component_200m
20
- image_size_pixels_height: 60
21
- image_size_pixels_width: 60
22
- dropout_timedeltas_minutes:
23
- - -360
24
- dropout_fraction: 1.0
25
- max_staleness_minutes: null
26
- normalisation_constants:
27
- temperature_sl:
28
- mean: 283.48333740234375
29
- std: 3.692270040512085
30
- wind_u_component_10m:
31
- mean: 1.7677178382873535
32
- std: 5.531515598297119
33
- wind_u_component_100m:
34
- mean: 2.393547296524048
35
- std: 7.2320556640625
36
- wind_u_component_200m:
37
- mean: 2.7963004112243652
38
- std: 8.049470901489258
39
- wind_v_component_10m:
40
- mean: 0.985887885093689
41
- std: 5.411230564117432
42
- wind_v_component_100m:
43
- mean: 1.4244288206100464
44
- std: 6.944501876831055
45
- wind_v_component_200m:
46
- mean: 1.6010299921035767
47
- std: 7.561611652374268
48
  gencast:
49
  provider: gencast
50
  zarr_path: PLACEHOLDER.zarr
@@ -220,10 +186,3 @@ input_data:
220
  P90_mean_sea_level_pressure:
221
  mean: 100894.1015625
222
  std: 672.7207641601562
223
- generation:
224
- time_resolution_minutes: 15
225
- interval_start_minutes: 0
226
- interval_end_minutes: 2880
227
- dropout_timedeltas_minutes: []
228
- dropout_fraction: 0
229
- zarr_path: /mnt/disks/gdm_data/gdm_data/ruvnl_data/india_wind_data_data_sampler_schema_v2.nc
 
2
  description: Example config for producing PVNet samples
3
  name: example_config
4
  input_data:
5
+ site:
6
+ time_resolution_minutes: 15
7
+ interval_start_minutes: 0
8
+ interval_end_minutes: 2880
9
+ file_path: /mnt/disks/gdm_data/gdm_data/ruvnl_data/india_wind_data_data_sampler_schema_v2.nc
10
+ metadata_file_path: /mnt/disks/gdm_data/gdm_data/ruvnl_data/india_wind_metadata_data_sampler_schema.csv
11
+ dropout_timedeltas_minutes: []
12
+ dropout_fraction: 0
13
  nwp:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
  gencast:
15
  provider: gencast
16
  zarr_path: PLACEHOLDER.zarr
 
186
  P90_mean_sea_level_pressure:
187
  mean: 100894.1015625
188
  std: 672.7207641601562
 
 
 
 
 
 
 
model_config.yaml CHANGED
@@ -5,22 +5,10 @@ output_quantiles:
5
  - 0.5
6
  - 0.75
7
  - 0.9
 
8
  interval_minutes: 15
9
  embedding_dim: null
10
  nwp_encoders_dict:
11
- ecmwf:
12
- _target_: pvnet.models.late_fusion.encoders.encoders3d.DefaultPVNet
13
- _partial_: true
14
- in_channels: 7
15
- out_features: 32
16
- fc_features: 32
17
- number_of_conv3d_layers: 3
18
- conv3d_channels: 16
19
- image_size_pixels: 60
20
- stride:
21
- - 1
22
- - 2
23
- - 2
24
  gencast:
25
  _target_: pvnet.models.late_fusion.encoders.encoders3d.DefaultPVNet
26
  _partial_: true
@@ -40,15 +28,13 @@ output_network:
40
  dropout_frac: 0.2
41
  include_sun: false
42
  include_time: true
 
 
43
  forecast_minutes: 2880
44
  history_minutes: 0
45
  nwp_history_minutes:
46
  gencast: 0
47
- ecmwf: 0
48
  nwp_forecast_minutes:
49
  gencast: 2880
50
- ecmwf: 2880
51
  nwp_interval_minutes:
52
  gencast: 360
53
- ecmwf: 60
54
- include_generation_history: false
 
5
  - 0.5
6
  - 0.75
7
  - 0.9
8
+ target_key: site
9
  interval_minutes: 15
10
  embedding_dim: null
11
  nwp_encoders_dict:
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  gencast:
13
  _target_: pvnet.models.late_fusion.encoders.encoders3d.DefaultPVNet
14
  _partial_: true
 
28
  dropout_frac: 0.2
29
  include_sun: false
30
  include_time: true
31
+ include_gsp_yield_history: false
32
+ include_site_yield_history: false
33
  forecast_minutes: 2880
34
  history_minutes: 0
35
  nwp_history_minutes:
36
  gencast: 0
 
37
  nwp_forecast_minutes:
38
  gencast: 2880
 
39
  nwp_interval_minutes:
40
  gencast: 360
 
 
model_weights.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d21a0a811b0c151f6d07516f970e268d064338711d5ed650e1700a92db4903f5
3
- size 6605176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6eaa3406e8c8a34640f08fe8da833fbe68055459218e6d5d358b9c313b4d965
3
+ size 2915336