Update app.py

app.py

@@ -49,58 +49,58 @@ import transformers
 
 # model_id = "mistralai/Mistral-7B-v0.3"
 
-# model_id = "microsoft/Phi-3-medium-4k-instruct"
+model_id = "microsoft/Phi-3-medium-4k-instruct"
 # model_id = "microsoft/phi-4"
 
-# # model_id = "Qwen/Qwen2-7B-Instruct"
+# model_id = "Qwen/Qwen2-7B-Instruct"
 
 
-# tokenizer = AutoTokenizer.from_pretrained(
-#     # model_id
-#     model_id,
-#     # use_fast=False
-#     token= token,
-# trust_remote_code=True)
+tokenizer = AutoTokenizer.from_pretrained(
+    # model_id
+    model_id,
+    # use_fast=False
+    token= token,
+trust_remote_code=True)
 
 
-# accelerator = Accelerator()
+accelerator = Accelerator()
 
-# model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
-#                                                  # torch_dtype= torch.uint8, 
-#                                              torch_dtype=torch.bfloat16,
-#                                               # load_in_8bit=True,
-#                                              # #  # torch_dtype=torch.fl,
-#                                              attn_implementation="flash_attention_2",
-#                                              low_cpu_mem_usage=True,
-#                                              trust_remote_code=True,
-#                                              device_map='cuda',
-#                                              # device_map=accelerator.device_map,
+model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
+                                                 # torch_dtype= torch.uint8, 
+                                             torch_dtype=torch.bfloat16,
+                                              # load_in_8bit=True,
+                                             # #  # torch_dtype=torch.fl,
+                                             attn_implementation="flash_attention_2",
+                                             low_cpu_mem_usage=True,
+                                             trust_remote_code=True,
+                                             device_map='cuda',
+                                             # device_map=accelerator.device_map,
                                              
-#                                             )
+                                            )
 
 
 
 
 
-# # 
-# model = accelerator.prepare(model)
-# from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
+# 
+model = accelerator.prepare(model)
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 
-# pipe = pipeline(
-#     "text-generation",
-#     model=model,
-#     tokenizer=tokenizer,
-# )
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+)
 
 
 
 
-pipeline = transformers.pipeline(
-    "text-generation",
-    model="microsoft/phi-4",
-    model_kwargs={"torch_dtype": "auto"},
-    device_map="auto",
-)
+# pipeline = transformers.pipeline(
+#     "text-generation",
+#     model="microsoft/phi-4",
+#     model_kwargs={"torch_dtype": "auto"},
+#     device_map="auto",
+# )
 
 
 # device_map = infer_auto_device_map(model, max_memory={0: "79GB", "cpu":"65GB" })