Update README.md

README.md

@@ -14,12 +14,13 @@ datasets:
 
 # MeteoLibre Rectified Flow Model
 
-This is a rectified flow diffusion model trained for meteorological data forecasting using the MeteoLibre dataset. The model uses a 3D U-Net architecture with FiLM conditioning for efficient weather pattern generation.
+This is a repo with differents models used for doing weather forecasting :
+- epoch_126_mtg_meteofrance_.safetensors (model with sat + ground station) : config is model_v0_mtg_meteofrance
 
 ## Model Description
 
 - **Model type**: Rectified Flow Diffusion Model
-- **Architecture**: 3D DC-AE U-Net with FiLM conditioning
+- **Architecture**: 3D U-Net with FiLM conditioning
 - **Input**: Meteorological data patches (12 channels, 3D spatio-temporal)
 - **Output**: Generated weather forecast data
 - **Training data**: MeteoLibre meteorological dataset
@@ -34,17 +35,6 @@ This model is designed for:
 - Research in atmospheric science and weather prediction
 - Educational purposes in machine learning for climate modeling
 
-## Model Architecture
-
-The model consists of:
-- **UNet_DCAE_3D**: 3D convolutional U-Net with encoder-decoder architecture
-- **FiLM Conditioning**: Feature-wise linear modulation for temporal context
-- **Rectified Flow**: Efficient generative modeling approach
-- **Input channels**: 12 (meteorological variables)
-- **Output channels**: 12 (forecast variables)
-- **Features**: [64, 128, 256] channel progression
-- **Context frames**: 4 (temporal conditioning)
-
 ## Training
 
 The model was trained using:
@@ -55,53 +45,6 @@ The model was trained using:
 - **Precision**: Mixed precision (bf16)
 - **Distributed training**: Multi-GPU support
 
-## Usage
-
-### Loading the Model
-
-```python
-from safetensors.torch import load_file
-import torch
-from meteolibre_model.models.dc_3dunet_film import UNet_DCAE_3D
-
-# Load model weights
-state_dict = load_file("epoch_141_rectified_flow.safetensors")
-
-# Create model
-model = UNet_DCAE_3D(
-    in_channels=12,
-    out_channels=12,
-    features=[64, 128, 256],
-    context_dim=4,
-    context_frames=4,
-    num_additional_resnet_blocks=2
-)
-
-model.load_state_dict(state_dict)
-model.eval()
-```
-
-### Inference
-
-```python
-# Example inference code
-with torch.no_grad():
-    generated_data = model(input_batch)
-```
-
-## Performance
-
-The model checkpoints are saved at regular intervals:
-- epoch_1_rectified_flow.safetensors through epoch_141_rectified_flow.safetensors
-- Best performing checkpoints available for different training stages
-
-## Limitations
-
-- Model trained on specific meteorological dataset
-- May not generalize to all weather patterns or regions
-- Requires significant computational resources for inference
-- Temporal context limited to 4 frames
-
 ## Ethical Considerations
 
 - Weather forecasting models should be used responsibly