Initial release: trained nanoGPT in safetensors format

README.md

@@ -21,7 +21,7 @@ A 10.7M parameter character-level GPT trained from scratch on the Tiny Shakespea
 ```python
 import json
 import torch
-from safetensors.torch import load_file
+from safetensors.torch import load_model  # <-- Changed to load_model
 from huggingface_hub import hf_hub_download
 from model import GPTConfig, GPT
 
@@ -43,7 +43,7 @@ decode = lambda l: "".join([itos[i] for i in l])
 # 3. Instantiate model
 config = GPTConfig(**cfg_dict)
 model = GPT(config)
-model.load_state_dict(load_file(weights_file))
+load_model(model, weights_file)  # <-- Handles tied weights automatically!
 model.eval()
 
 # 4. Generate text