From 9ef2acd25e2f860ea66d2c47540abadc16e1ed63 Mon Sep 17 00:00:00 2001
From: Clement Dumas <clement.dumas@ens-paris-saclay.fr>
Date: Fri, 22 Nov 2024 13:45:09 +0100
Subject: [PATCH] testing

---
 dictionary_learning/dictionary.py | 47 +++++++++++++++----------------
 1 file changed, 23 insertions(+), 24 deletions(-)

diff --git a/dictionary_learning/dictionary.py b/dictionary_learning/dictionary.py
index 1e82d31..2b02909 100644
--- a/dictionary_learning/dictionary.py
+++ b/dictionary_learning/dictionary.py
@@ -11,7 +11,6 @@
 from torch.nn.functional import relu
 import einops
 from warnings import warn
-import tempfile
 
 
 class Dictionary(ABC, nn.Module, PyTorchModelHubMixin):
@@ -593,29 +592,29 @@ def forward(self, x: th.Tensor, output_features=False):
         else:
             return x_hat
 
-    @classmethod
-    def from_pretrained(
-        cls,
-        path: str,
-        dtype: th.dtype = th.float32,
-        device: th.device | None = None,
-    ):
-        """
-        Load a pretrained cross-coder from a file.
-        """
-        state_dict = th.load(path, map_location="cpu", weights_only=True)
-        if "encoder.weight" not in state_dict:
-            warn(
-                "Cross-coder state dict was saved while torch.compiled was enabled. Fixing..."
-            )
-            state_dict = {k.split("_orig_mod.")[1]: v for k, v in state_dict.items()}
-        num_layers, activation_dim, dict_size = state_dict["encoder.weight"].shape
-        cross_coder = cls(activation_dim, dict_size, num_layers)
-        cross_coder.load_state_dict(state_dict)
-
-        if device is not None:
-            cross_coder = cross_coder.to(device)
-        return cross_coder.to(dtype=dtype)
+    # @classmethod
+    # def from_pretrained(
+    #     cls,
+    #     path: str,
+    #     dtype: th.dtype = th.float32,
+    #     device: th.device | None = None,
+    # ):
+    #     """
+    #     Load a pretrained cross-coder from a file.
+    #     """
+    #     state_dict = th.load(path, map_location="cpu", weights_only=True)
+    #     if "encoder.weight" not in state_dict:
+    #         warn(
+    #             "Cross-coder state dict was saved while torch.compiled was enabled. Fixing..."
+    #         )
+    #         state_dict = {k.split("_orig_mod.")[1]: v for k, v in state_dict.items()}
+    #     num_layers, activation_dim, dict_size = state_dict["encoder.weight"].shape
+    #     cross_coder = cls(activation_dim, dict_size, num_layers)
+    #     cross_coder.load_state_dict(state_dict)
+
+    #     if device is not None:
+    #         cross_coder = cross_coder.to(device)
+    #     return cross_coder.to(dtype=dtype)
 
     def resample_neurons(self, deads, activations):
         # https://transformer-circuits.pub/2023/monosemantic-features/index.html#appendix-autoencoder-resampling