recover scale_dtype and remove double_quant_scale_dtype

Signed-off-by: changwangss <chang1.wang@intel.com>

docs/weightonlyquant.md

@@ -68,7 +68,7 @@ input_ids = tokenizer(prompt, return_tensors="pt").input_ids
 ```bash
 from intel_extension_for_transformers.transformers import AutoModelForCausalLM, WeightOnlyQuantConfig
 # weight_dtype: int8/int4_fullrange/int4_clip/nf4/fp4_e2m1_bnb/fp4_e2m1/fp8_e5m2/fp8_e4m3
-# double_quant_scale_dtype: fp32/fp8, fp8 only used for weight_dtype "fp8_e5m2", "fp8_e4m3"
+# scale_dtype: fp32/fp8, fp8 only used for weight_dtype "fp8_e5m2", "fp8_e4m3"
 woq_config = WeightOnlyQuantConfig(weight_dtype="int4_fullrange", group_size=32)
 woq_model = AutoModelForCausalLM.from_pretrained(
                                                     model_name_or_path,

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

@@ -187,7 +187,7 @@
 elif args.woq:
     quantization_config = WeightOnlyQuantConfig(
         weight_dtype=args.woq_weight_dtype,
-        double_quant_scale_dtype=args.woq_scale_dtype,
+        scale_dtype=args.woq_scale_dtype,
         group_size=args.woq_group_size,
         scheme=args.woq_scheme,
         algorithm=args.woq_algo,

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

@@ -206,7 +206,7 @@
 elif args.woq:
     quantization_config = WeightOnlyQuantConfig(
         compute_dtype=args.woq_compute_dtype,
-        double_quant_scale_dtype=args.woq_scale_dtype,
+        scale_dtype=args.woq_scale_dtype,
         weight_dtype=args.woq_weight_dtype,
         scheme=args.woq_scheme,
         group_size=args.woq_group_size,

intel_extension_for_transformers/llm/quantization/utils.py

@@ -48,45 +48,45 @@ def replace_linear(model, modules_to_not_convert=None, current_key_name=None, qu
 
 def get_weight_type_from_config(config):
     if config.weight_dtype == "int8":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "s8_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "int4_fullrange":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "s4fullrange_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "int4_clip":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "s4clip_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "fp4_e2m1_bnb":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "fp4bnb_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "fp4_e2m1":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "fp4e2m1_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "nf4":
-        if config.double_quant_scale_dtype == "fp32":
+        if config.scale_dtype == "fp32":
             weight_type = "nf4_scalef32"
         else:
-            raise Exception("double_quant_scale_dtype only support fp32 now!")
+            raise Exception("scale_dtype only support fp32 now!")
     elif config.weight_dtype == "fp8_e5m2":
-        if config.double_quant_scale_dtype == "fp8":
+        if config.scale_dtype == "fp8":
             weight_type = "fp8e5m2_scalef8"
         else:
-            raise Exception("double_quant_scale_dtype only support fp8 now!")
+            raise Exception("scale_dtype only support fp8 now!")
     elif config.weight_dtype == "fp8_e4m3":
-        if config.double_quant_scale_dtype == "fp8":
+        if config.scale_dtype == "fp8":
             weight_type = "fp8e4m3_scalef8"
         else:
-            raise Exception("double_quant_scale_dtype only support fp8 now!")
+            raise Exception("scale_dtype only support fp8 now!")
     return weight_type
 
 

intel_extension_for_transformers/llm/runtime/graph/README.md

@@ -125,7 +125,7 @@ Argument description of WeightOnlyQuantConfig:
 | weight_dtype      | String      | Data type of quantized weight: int4/int8 (default int4)                                 |
 | alg               | String      | Quantization algorithm: sym/asym (default sym)                                          |
 | group_size        | Int         | Group size: Int (default: 32)                                                           |
-| double_quant_scale_dtype       | String      | Data type of scales: fp32/bf16 (default fp32)                                           |
+| scale_dtype       | String      | Data type of scales: fp32/bf16 (default fp32)                                           |
 | use_ggml          | Bool        | Enable ggml for quantization and inference (default: False)                             |
 | use_quant         | Bool        | Determine whether or not the model will be quantized. (default: True)                  |
 

intel_extension_for_transformers/transformers/modeling/modeling_auto.py

@@ -174,7 +174,7 @@ def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
                     weight_dtype=quantization_config.weight_dtype,
                     alg=quantization_config.scheme,
                     group_size=quantization_config.group_size,
-                    scale_dtype=quantization_config.double_quant_scale_dtype,
+                    scale_dtype=quantization_config.scale_dtype,
                     compute_dtype=quantization_config.compute_dtype,
                     use_ggml=quantization_config.use_ggml,
                     use_quant=quantization_config.use_quant,

intel_extension_for_transformers/transformers/utils/config.py

@@ -33,10 +33,10 @@ def __init__(
         llm_int8_skip_modules=None,
         compute_dtype=None,
         weight_dtype=None,
+        scale_dtype="fp32"
         mse_range=False,
         use_double_quant=False,
         double_quant_dtype="int8",  # reserve for double quant
-        double_quant_scale_dtype="fp32",  # reserve for double quant
         group_size=32,
         scheme="sym",
         algorithm="RTN",
@@ -56,7 +56,7 @@ def __init__(
         self.mse_range = mse_range
         self.use_double_quant = use_double_quant
         self.double_quant_dtype = double_quant_dtype
-        self.double_quant_scale_dtype = double_quant_scale_dtype
+        self.scale_dtype = scale_dtype
         self.scheme = scheme
         self.algorithm = algorithm
         self.group_size = group_size
@@ -107,9 +107,9 @@ def post_init(self):
                 f"'int8', 'int4_fullrange', 'int4_clip', 'nf4', 'fp4_e2m1_bnb', 'fp4_e2m1', 'fp8_e5m2, fp8_e4m3'"
             )
 
-        if self.double_quant_scale_dtype not in ["fp32", "fp8"]:
+        if self.scale_dtype not in ["fp32", "fp8"]:
             raise ValueError(
-                f"double_quant_scale_dtype must be a string in 'fp32', 'fp8' "
+                f"scale_dtype must be a string in 'fp32', 'fp8' "
                 f"and fp8 only used for weight_dtype 'fp8_e5m2', 'fp8_e4m3'")
 
         if not isinstance(self.mse_range, bool):
@@ -121,8 +121,8 @@ def post_init(self):
         if self.use_double_quant and not isinstance(self.double_quant_dtype, str):
             raise ValueError("double_quant_dtype must be a string")
 
-        if self.use_double_quant and not isinstance(self.double_quant_scale_dtype, str):
-            raise ValueError("double_quant_scale_dtype must be a string")
+        if self.use_double_quant and not isinstance(self.scale_dtype, str):
+            raise ValueError("scale_dtype must be a string")
 
         if not isinstance(self.group_size, int):
             raise ValueError("group_size must be a int")
@@ -150,8 +150,8 @@ def post_init_runtime(self):
         elif self.weight_dtype not in ["int4", "int8"]:
             raise ValueError(f"weight_dtype must be 'int4', 'int8'.")
 
-        if self.double_quant_scale_dtype not in ["fp32", "fp16"]:
-            raise ValueError("double_quant_scale_dtype must be 'fp32', 'fp16'.")
+        if self.scale_dtype not in ["fp32", "fp16"]:
+            raise ValueError("scale_dtype must be 'fp32', 'fp16'.")
 
         if self.group_size not in [-1, 32, 128]:
             raise ValueError("group_size must be an integer in [-1, 32, 128]")