load processed model automatically (#86)

Signed-off-by: zhenwei-intel <[email protected]>
intel · Jan 24, 2024 · 6625539 · 6625539
1 parent 51088a2
commit 6625539
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 16 deletions.
diff --git a/neural_speed/__init__.py b/neural_speed/__init__.py
@@ -77,7 +77,7 @@ def get_model_type(model_config):
             model_type = "chatglm2"
         return model_type
 
-    def init(self, model_name, use_quant=True, use_cache=False, use_gptq=False, use_awq=False,
+    def init(self, model_name, use_quant=True, use_gptq=False, use_awq=False,
             weight_dtype="int4", alg="sym", group_size=32,
             scale_dtype="fp32", compute_dtype="int8", use_ggml=False):
         self.config = AutoConfig.from_pretrained(model_name, trust_remote_code=True)
@@ -107,14 +107,17 @@ def init(self, model_name, use_quant=True, use_cache=False, use_gptq=False, use_
             self.bin_file = fp32_bin
         else:
             self.bin_file = quant_bin
-        if use_cache and os.path.exists(self.bin_file):
+
+        if os.path.exists(self.bin_file):
+            print("{} existed, will use cache file. Otherwise please remove the file".
+                  format(self.bin_file))
             return
 
         if use_gptq or use_awq:
             convert_model(model_name, quant_bin, "f32")
             return
 
-        if not use_cache or not os.path.exists(fp32_bin):
+        if not os.path.exists(fp32_bin):
             convert_model(model_name, fp32_bin, "f32")
             assert os.path.exists(fp32_bin), "Fail to convert pytorch model"
 
@@ -127,8 +130,7 @@ def init(self, model_name, use_quant=True, use_cache=False, use_gptq=False, use_
         assert os.path.exists(quant_bin), "Fail to quantize model"
 
         # clean
-        if not use_cache:
-            os.remove(fp32_bin)
+        os.remove(fp32_bin)
 
     def init_from_bin(self, model_type, model_path, **generate_kwargs):
         self.__import_package(model_type)

diff --git a/scripts/cal_diff.py b/scripts/cal_diff.py
@@ -35,10 +35,10 @@ def cmpData(numa, numb):
     args = parser.parse_args()
 
     woq_configs = {
-        "fp32": {"use_cache":True, "not_quant":True},
-        # "ggml_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_cache":True, "use_ggml":True},
-        "jblas_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_cache":True},
-        # "jblas_int8": {"compute_dtype":"bf16", "weight_dtype":"int8", "use_cache":True},
+        "fp32": {"not_quant":True},
+        # "ggml_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_ggml":True},
+        "jblas_int4": {"compute_dtype":"int8", "weight_dtype":"int4"},
+        # "jblas_int8": {"compute_dtype":"bf16", "weight_dtype":"int8"},
     }
     prompt = "What is the meaning of life?"
 

diff --git a/scripts/perplexity.py b/scripts/perplexity.py
@@ -105,7 +105,7 @@ def perplexity(model_name, dataset_name, **kwargs):
         init_kwargs = {
             k: kwargs[k]
             for k in kwargs
-            if k in ['use_cache', 'compute_dtype', 'weight_dtype', 'scale_dtype', 'group_size', 'use_ggml']
+            if k in ['compute_dtype', 'weight_dtype', 'scale_dtype', 'group_size', 'use_ggml']
         }
         model.init(model_name, **init_kwargs)
 
@@ -186,7 +186,6 @@ def add_quant_args(parser: argparse.ArgumentParser):
                        type=str,
                        help="path to quantized weight; other quant args will be ignored if specified",
                        default="")
-    group.add_argument('--use_cache', action="store_true", help="Use local quantized model if file exists")
     group.add_argument(
         "--weight_dtype",
         choices=["int4", "int8"],

diff --git a/tests/test_python_api.py b/tests/test_python_api.py
@@ -53,7 +53,7 @@ def test_llm_runtime(self):
         print(tokenizer.decode(pt_generate_ids))
 
         # check output ids
-        woq_config_fp32 = {"use_quant":False, "compute_dtype":"fp32", "weight_dtype":"fp32", "use_cache":True, "use_ggml":False, "group_size":128}
+        woq_config_fp32 = {"use_quant":False, "compute_dtype":"fp32", "weight_dtype":"fp32", "use_ggml":False, "group_size":128}
         itrex_model = Model()
 
         itrex_model.init(model_name, use_quant=False)
@@ -65,10 +65,10 @@ def test_llm_runtime(self):
 
         # check diff of logits
         woq_configs = {
-            "fp32": {"use_cache":True, "use_quant":False},
-            # "ggml_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_cache":True, "use_ggml":True},
-            "jblas_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_cache":True},
-            # "jblas_int8": {"compute_dtype":"bf16", "weight_dtype":"int8", "use_cache":True},
+            "fp32": {"use_quant":False},
+            # "ggml_int4": {"compute_dtype":"int8", "weight_dtype":"int4", "use_ggml":True},
+            "jblas_int4": {"compute_dtype":"int8", "weight_dtype":"int4"},
+            # "jblas_int8": {"compute_dtype":"bf16", "weight_dtype":"int8"},
         }
         for config_type in woq_configs:
             itrex_model = Model()