Upload app.py with huggingface_hub

app.py

@@ -22,7 +22,7 @@ logger = logging.getLogger(__name__)
 class GPTConfig:
     """GPT model configuration"""
     def __init__(self, vocab_size=32000, n_layer=6, n_head=8, n_embd=512,
-                 block_size=1024, dropout=0.1, bias=True, **kwargs):
+                 block_size=1024, dropout=0.1, bias=False, **kwargs):
         # Accept any additional kwargs to handle extra config fields
         self.vocab_size = vocab_size
         self.n_layer = n_layer
@@ -49,8 +49,8 @@ class GPT(nn.Module):
             ln_f = nn.LayerNorm(config.n_embd),
         ))
         
-        # Language model head - MUST have bias to match saved model
-        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=True)
+        # Language model head - Use bias=False to match saved models
+        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
         
         # Initialize weights
         self.apply(self._init_weights)
@@ -328,7 +328,7 @@ class RealOpenLLMInference:
                     n_embd=512,
                     block_size=1024,
                     dropout=0.1,
-                    bias=True
+                    bias=False
                 )
             
             # Load model weights