Upload folder using huggingface_hub

architecture.py

@@ -4,7 +4,9 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from typing import Optional, Dict
-
+from transformers import AutoConfig, AutoModelForCausalLM
+from transformers.models.phi3.configuration_phi3 import Phi3Config
+from transformers.models.phi3.modeling_phi3 import Phi3ForCausalLM
 # This file contains the custom nn.Module definitions required by the fine-tuned model.
 # By placing them here, the main training script is cleaner, and more importantly,
 # this file can be packaged with the model for easy loading from the Hugging Face Hub.
@@ -115,4 +117,5 @@ class GCVectorMemoryLayer(nn.Module):
 
         return corrected_activation
 
+AutoModelForCausalLM.register(Phi3Config, Phi3WithVectorMemoryForCausalLM)
 # --- END OF FILE architecture.py ---

config.json

@@ -1,12 +1,11 @@
 {
   "architectures": [
-    "Phi3ForCausalLM"
+    "Phi3WithVectorMemoryForCausalLM"
   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
   "auto_map": {
-    "AutoConfig": "microsoft/Phi-3-mini-4k-instruct--configuration_phi3.Phi3Config",
-    "AutoModelForCausalLM": "microsoft/Phi-3-mini-4k-instruct--modeling_phi3.Phi3ForCausalLM"
+    "AutoModelForCausalLM": "architecture.Phi3WithVectorMemoryForCausalLM"
   },
   "bos_token_id": 1,
   "embd_pdrop": 0.0,
@@ -32,4 +31,4 @@
   "transformers_version": "4.52.4",
   "use_cache": true,
   "vocab_size": 32064
-}
+}