[Example] Add WOQ to code-generation example (#581)

* add woq for code-generation example

Signed-off-by: Cheng, Zixuan <zixuan.cheng@intel.com>

examples/huggingface/pytorch/code-generation/quantization/run_generation.py

@@ -60,7 +60,17 @@
                     help="Save accuracy results path.")
 parser.add_argument("--tasks", default="humaneval", type=str, \
                     help="tasks list for accuracy validation")
-
+# WeightOnlyQuant config
+parser.add_argument("--woq", action="store_true")
+parser.add_argument("--woq_algo", default="RTN", choices=['RTN', 'AWQ', 'TEQ'], 
+                    help="Weight-only parameter.")
+parser.add_argument("--woq_compute_dtype", type=str, default="fp32")
+parser.add_argument("--woq_weight_dtype", type=str, default="int8", 
+                    choices=["int8", "int4_clip", "int4_fullrange", "fp4_e2m1_bnb", "fp4_e2m1", "nf4"])
+parser.add_argument("--woq_group_size", type=int, default=-1)
+parser.add_argument("--woq_scheme", default="sym")
+parser.add_argument("--woq_enable_mse_search", action="store_true")
+parser.add_argument("--woq_enable_full_range", action="store_true")
 # Harness config
 parser.add_argument("--n_samples", default=200, type=int)
 parser.add_argument("--limit", default=None, type=int, help="Limit number of samples to eval")
@@ -100,7 +110,7 @@
 args = parser.parse_args()
 
 
-from intel_extension_for_transformers.transformers import AutoModelForCausalLM
+from intel_extension_for_transformers.transformers import AutoModelForCausalLM, WeightOnlyQuantConfig
 user_model = AutoModelForCausalLM.from_pretrained(
     args.model,
     torchscript=True
@@ -274,6 +284,14 @@ def calib_func(prepared_model):
             recipes=recipes,
             example_inputs=example_inputs,
         )
+    elif args.woq:
+        conf = WeightOnlyQuantConfig(
+            compute_dtype=args.woq_compute_dtype,
+            weight_dtype=args.woq_weight_dtype,
+            group_size=args.woq_group_size,
+            scheme=args.woq_scheme,
+            algorithm=args.woq_algo,
+        )
     else:
         conf = PostTrainingQuantConfig(
             backend="ipex" if args.ipex else "default",
@@ -283,10 +301,10 @@ def calib_func(prepared_model):
         )
     # save config
     user_model.config.save_pretrained(args.output_dir)
-    q_model = quantization.fit(
-        user_model,
-        conf,
-        calib_func=calib_func,
+    q_model = AutoModelForCausalLM.from_pretrained(
+        args.model,
+        quantization_config=conf,
+        use_llm_runtime=False
     )
     q_model.save(args.output_dir)