Feat - Install

README.md

@@ -16,6 +16,37 @@ tags:
 
 A complete autoencoder implementation that integrates seamlessly with the Hugging Face Transformers ecosystem, providing all the standard functionality you expect from transformer models.
 
+
+### Install-and-Use from the Hub (code repo)
+
+If you want to use the implementation directly from the Hub code repository (without a packaged pip install), you can download the repo and add it to `sys.path`:
+
+```python
+from huggingface_hub import snapshot_download
+import sys, torch
+
+repo_dir = snapshot_download(
+    "amaye15/autoencoder",
+    repo_type="model",
+    allow_patterns=["*.py", "config.json", "*.safetensors"],
+)
+sys.path.append(repo_dir)
+
+from configuration_autoencoder import AutoencoderConfig
+from modeling_autoencoder import AutoencoderForReconstruction
+
+# Load placeholder weights from the same repo (or your own trained weights)
+model = AutoencoderForReconstruction.from_pretrained(
+    "amaye15/autoencoder",
+    trust_remote_code=True,
+)
+
+# Quick smoke test
+x = torch.randn(8, 20)
+outputs = model(input_values=x)
+print("Reconstructed:", tuple(outputs.reconstructed.shape), "Latent:", tuple(outputs.last_hidden_state.shape))
+```
+
 ## 🚀 Features
 
 - **Full Hugging Face Integration**: Compatible with `AutoModel`, `AutoConfig`, and `AutoTokenizer` patterns
@@ -86,6 +117,7 @@ print(f"Latent shape: {outputs.last_hidden_state.shape}")
 print(f"Reconstructed shape: {outputs.reconstructed.shape}")
 ```
 
+
 ### Training with Hugging Face Trainer
 
 ```python

config.json

@@ -0,0 +1,37 @@
+{
+  "activation": "gelu",
+  "architectures": [
+    "AutoencoderForReconstruction"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_autoencoder.AutoencoderConfig",
+    "AutoModel": "modeling_autoencoder.AutoencoderForReconstruction"
+  },
+  "autoencoder_type": "classic",
+  "beta": 1.0,
+  "bidirectional": true,
+  "dropout_rate": 0.1,
+  "flow_coupling_layers": 2,
+  "hidden_dims": [
+    16,
+    12
+  ],
+  "input_dim": 20,
+  "latent_dim": 8,
+  "learn_inverse_preprocessing": true,
+  "model_type": "autoencoder",
+  "noise_factor": 0.1,
+  "num_layers": 2,
+  "preprocessing_hidden_dim": 32,
+  "preprocessing_num_layers": 2,
+  "preprocessing_type": "robust_scaler",
+  "reconstruction_loss": "mse",
+  "rnn_type": "lstm",
+  "sequence_length": null,
+  "teacher_forcing_ratio": 0.5,
+  "tie_weights": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.55.2",
+  "use_batch_norm": true,
+  "use_learnable_preprocessing": true
+}