Update app.py

app.py

@@ -49,11 +49,7 @@ logger = CSVLogger("out", name, flush_logs_every_n_steps=log_interval)
 
 fabric = L.Fabric(devices=1, strategy='auto', precision=None, loggers=logger)
 
-checkpoint_path = Path("out/redpajama/iter-023999-ckpt.pth")
-config = Config.from_name(model_name)
-model = GPT(config)
 
-load_checkpoint(fabric, model, checkpoint_path)
 
 #print(model.transformer.h[0].mlp.fc.weight)
 
@@ -92,7 +88,7 @@ def generate( model, config, idx, max_new_tokens, temperature=1.0, top_k=None):
 checkpoint_dir = Path('./checkpoints/meta-llama/Llama-2-7b-chat-hf')
 token = Tokenizer(checkpoint_dir = checkpoint_dir)
 
-def tsaigpt(start:str , model= model, max_new_tokens = 300, num_samples =2, tokeniser= token):
+def tsaigpt(start:str , max_new_tokens = 300, num_samples =2, tokeniser= token):
 
 
 
@@ -113,7 +109,11 @@ def tsaigpt(start:str , model= model, max_new_tokens = 300, num_samples =2, toke
     device_type = 'cuda' if 'cuda' in device else 'cpu' # for later use in torch.autocast
     ptdtype = {'float32': torch.float32, 'bfloat16': torch.bfloat16, 'float16': torch.float16}[dtype]
     ctx = nullcontext() if device_type == 'cpu' else torch.amp.autocast(device_type=device_type, dtype=ptdtype)
+    checkpoint_path = Path("out/redpajama/iter-023999-ckpt.pth")
+    config = Config.from_name(model_name)
+    model = GPT(config)
 
+    load_checkpoint(fabric, model, checkpoint_path)
     model.eval()
     model.to(device)
     if compile: