Upload folder using huggingface_hub

earthformer_model/README.md

@@ -0,0 +1,45 @@
+---
+language: en
+library_name: pytorch
+license: mit
+---
+
+
+# Cloudcasting
+
+## Model Description
+
+<!-- Provide a longer summary of what this model is/does. -->
+This model is trained to predict future frames of satellite data from past frames. It takes 3 hours
+of recent satellkite imagery at 15 minute intervals and predicts 3 hours into the future also at
+15 minute intervals. The satellite inputs and predictions are multispectral with 11 channels.
+
+
+See [1] for the repo used to train the model.
+
+- **Developed by:** Open Climate Fix and the Alan Turing Institute
+- **License:** mit
+
+
+# Training Details
+
+## Data
+
+<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+This was trained on EUMETSAT satellite imagery derived from the data stored in [this google public 
+dataset](https://console.cloud.google.com/marketplace/product/bigquery-public-data/eumetsat-seviri-rss?hl=en-GB&inv=1&invt=AbniZA&project=solar-pv-nowcasting&pli=1). 
+
+The data was processed using the protocol in [2]
+
+
+
+## Results
+
+The training logs for the current model can be found here:
+- https://wandb.ai/openclimatefix/sat_pred/runs/ckyb2l1s
+
+
+### Software
+
+- [1] https://github.com/openclimatefix/sat_pred
+- [2] https://github.com/alan-turing-institute/cloudcasting

earthformer_model/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8470382c40ca61bc382a74f5e34bfd22aa2531b9bfa633497d7d63f3081b732e
+size 34776668

earthformer_model/model_config.yaml

@@ -0,0 +1,72 @@
+_target_: sat_pred.models.earthformer_model.Earthformer
+attn_drop: 0.1
+attn_linear_init_mode: '0'
+base_units: 128
+block_units: null
+checkpoint_level: 0
+conv_init_mode: '0'
+dec_cross_attn_patterns: cross_1x1
+dec_cross_last_n_frames: null
+dec_depth:
+- 1
+- 1
+dec_hierarchical_pos_embed: false
+dec_self_attn_patterns: axial
+dec_self_update_global: true
+dec_use_first_self_attn: false
+dec_use_inter_ffn: true
+down_up_linear_init_mode: '0'
+downsample: 2
+downsample_type: patch_merge
+enc_attn_patterns: axial
+enc_depth:
+- 1
+- 1
+enc_use_inter_ffn: true
+ffn_activation: gelu
+ffn_drop: 0.1
+ffn_linear_init_mode: '0'
+gated_ffn: false
+global_dim_ratio: 1
+initial_downsample_activation: leaky
+initial_downsample_stack_conv_dim_list:
+- 16
+- 64
+- 128
+initial_downsample_stack_conv_downscale_list:
+- 3
+- 2
+- 2
+initial_downsample_stack_conv_num_conv_list:
+- 2
+- 2
+- 2
+initial_downsample_stack_conv_num_layers: 3
+initial_downsample_type: stack_conv
+input_shape:
+- 12
+- 372
+- 614
+- 11
+norm_init_mode: '0'
+norm_layer: layer_norm
+num_global_vectors: 8
+num_heads: 4
+padding_type: zeros
+pos_embed_type: t+h+w
+proj_drop: 0.1
+scale_alpha: 1.0
+self_attn_use_final_proj: true
+separate_global_qkv: true
+target_shape:
+- 12
+- 372
+- 614
+- 11
+upsample_type: upsample
+use_dec_cross_global: false
+use_dec_self_global: false
+use_global_self_attn: true
+use_global_vector_ffn: false
+use_relative_pos: true
+z_init_method: zeros

simvp_model/README.md

@@ -0,0 +1,45 @@
+---
+language: en
+license: mit
+library_name: pytorch
+---
+
+
+# Cloudcasting
+
+## Model Description
+
+<!-- Provide a longer summary of what this model is/does. -->
+This model is trained to predict future frames of satellite data from past frames. It takes 3 hours
+of recent satellkite imagery at 15 minute intervals and predicts 3 hours into the future also at
+15 minute intervals. The satellite inputs and predictions are multispectral with 11 channels.
+
+
+See [1] for the repo used to train the model.
+
+- **Developed by:** Open Climate Fix and the Alan Turing Institute
+- **License:** mit
+
+
+# Training Details
+
+## Data
+
+<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+This was trained on EUMETSAT satellite imagery derived from the data stored in [this google public 
+dataset](https://console.cloud.google.com/marketplace/product/bigquery-public-data/eumetsat-seviri-rss?hl=en-GB&inv=1&invt=AbniZA&project=solar-pv-nowcasting&pli=1). 
+
+The data was processed using the protocol in [2]
+
+
+
+## Results
+
+The training logs for the current model can be found here:
+- https://wandb.ai/openclimatefix/sat_pred/runs/ob9v9128
+
+
+### Software
+
+- [1] https://github.com/openclimatefix/sat_pred
+- [2] https://github.com/alan-turing-institute/cloudcasting

simvp_model/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc35b136a9b0383cc989af09f3b370693a5a147835132282d9086af9e1579e8e
+size 55940132

simvp_model/model_config.yaml

@@ -0,0 +1,4 @@
+_target_: sat_pred.models.simvp_model.SimVP
+forecast_len: 12
+history_len: 12
+num_channels: 11