tico 0.1.0.dev251026__py3-none-any.whl → 0.1.0.dev251028__py3-none-any.whl

tico/__init__.py

@@ -29,7 +29,7 @@ __all__ = [
 ]
 
 # THIS LINE IS AUTOMATICALLY GENERATED BY setup.py
-__version__ = "0.1.0.dev251026"
+__version__ = "0.1.0.dev251028"
 
 MINIMUM_SUPPORTED_VERSION = "2.5.0"
 SECURE_TORCH_VERSION = "2.6.0"

tico/quantization/__init__.py

@@ -0,0 +1,6 @@
+from tico.quantization.public_interface import convert, prepare
+
+__all__ = [
+    "convert",
+    "prepare",
+]

tico/{experimental/quantization → quantization}/algorithm/gptq/gptq.py

@@ -25,7 +25,7 @@ from typing import Optional
 import torch
 import torch.nn as nn
 
-from tico.experimental.quantization.algorithm.gptq.quant import quantize, Quantizer
+from tico.quantization.algorithm.gptq.quant import quantize, Quantizer
 
 torch.backends.cuda.matmul.allow_tf32 = False
 torch.backends.cudnn.allow_tf32 = False

tico/{experimental/quantization → quantization}/algorithm/gptq/quantizer.py

@@ -19,15 +19,15 @@ from typing import Any, Callable, Dict, List, Optional
 import torch
 from tqdm.auto import tqdm
 
-from tico.experimental.quantization.algorithm.gptq.gptq import GPTQ
-from tico.experimental.quantization.algorithm.gptq.utils import (
+from tico.quantization.algorithm.gptq.gptq import GPTQ
+from tico.quantization.algorithm.gptq.utils import (
     find_layers,
     gather_single_batch_from_dict,
     gather_single_batch_from_list,
 )
-from tico.experimental.quantization.config.gptq import GPTQConfig
-from tico.experimental.quantization.quantizer import BaseQuantizer
-from tico.experimental.quantization.quantizer_registry import register_quantizer
+from tico.quantization.config.gptq import GPTQConfig
+from tico.quantization.quantizer import BaseQuantizer
+from tico.quantization.quantizer_registry import register_quantizer
 
 
 class StopForward(Exception):

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/annotator.py

@@ -25,14 +25,12 @@ from torch.ao.quantization.observer import MinMaxObserver, PerChannelMinMaxObser
 from torch.ao.quantization.quantizer import QuantizationSpec, Quantizer
 from torch.ao.quantization.quantizer.utils import _get_module_name_filter
 
-from tico.experimental.quantization.algorithm.pt2e.annotation.op import *
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
-from tico.experimental.quantization.algorithm.pt2e.transformation.convert_scalars_to_attrs import (
+from tico.quantization.algorithm.pt2e.annotation.op import *
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
+from tico.quantization.algorithm.pt2e.transformation.convert_scalars_to_attrs import (
     convert_scalars_to_attrs,
 )
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/adaptive_avg_pool2d.py

@@ -19,12 +19,10 @@ if TYPE_CHECKING:
 import torch
 from torch.ao.quantization.quantizer import SharedQuantizationSpec
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import AdaptiveAvgPool2dArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/add.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import AddTensorArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/conv2d.py

@@ -19,12 +19,10 @@ if TYPE_CHECKING:
 import torch
 from torch.ao.quantization.quantizer import DerivedQuantizationSpec
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import Conv2DArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/div.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import DivTensorArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/linear.py

@@ -19,12 +19,10 @@ if TYPE_CHECKING:
 import torch
 from torch.ao.quantization.quantizer import DerivedQuantizationSpec
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import LinearArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/mean.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import MeanDimArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/mul.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import MulTensorArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/relu6.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import Relu6Args
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/rsqrt.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import RsqrtArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/sub.py

@@ -18,12 +18,10 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
-import tico.experimental.quantization.algorithm.pt2e.annotation.utils as annot_utils
-import tico.experimental.quantization.algorithm.pt2e.utils as quant_utils
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.utils as annot_utils
+import tico.quantization.algorithm.pt2e.utils as quant_utils
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 from tico.utils.validate_args_kwargs import SubTensorArgs
 
 

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/spec.py

@@ -18,9 +18,7 @@ if TYPE_CHECKING:
     import torch.fx
 import torch
 
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 
 AnnotatorType = Callable[
     [

tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/utils.py

@@ -22,7 +22,7 @@ from torch.ao.quantization.quantizer import (
     SharedQuantizationSpec,
 )
 
-import tico.experimental.quantization.algorithm.pt2e.annotation.spec as annot_spec
+import tico.quantization.algorithm.pt2e.annotation.spec as annot_spec
 
 
 def annotate_input_qspec_map(node: torch.fx.Node, input_node: torch.fx.Node, qspec):

tico/{experimental/quantization → quantization}/algorithm/pt2e/quantizer.py

@@ -18,13 +18,13 @@ import torch
 
 from torch.ao.quantization.quantize_pt2e import convert_pt2e, prepare_pt2e
 
-from tico.experimental.quantization.algorithm.pt2e.annotation.annotator import (
+from tico.quantization.algorithm.pt2e.annotation.annotator import (
     get_asymmetric_quantization_config,
     PT2EAnnotator,
 )
-from tico.experimental.quantization.config.pt2e import PT2EConfig
-from tico.experimental.quantization.quantizer import BaseQuantizer
-from tico.experimental.quantization.quantizer_registry import register_quantizer
+from tico.quantization.config.pt2e import PT2EConfig
+from tico.quantization.quantizer import BaseQuantizer
+from tico.quantization.quantizer_registry import register_quantizer
 
 
 @register_quantizer(PT2EConfig)

tico/{experimental/quantization → quantization}/algorithm/pt2e/utils.py

@@ -20,9 +20,7 @@ import torch
 from torch.ao.quantization.quantizer import QuantizationSpec
 from torch.ao.quantization.quantizer.utils import _get_module_name_filter
 
-from tico.experimental.quantization.algorithm.pt2e.annotation.config import (
-    QuantizationConfig,
-)
+from tico.quantization.algorithm.pt2e.annotation.config import QuantizationConfig
 
 
 def get_module_type_filter(tp: Callable):

tico/{experimental/quantization → quantization}/algorithm/smoothquant/quantizer.py

@@ -16,16 +16,12 @@ from typing import Any, Dict, Optional
 
 import torch
 
-from tico.experimental.quantization.algorithm.smoothquant.observer import (
-    ChannelwiseMaxActsObserver,
-)
-
-from tico.experimental.quantization.algorithm.smoothquant.smooth_quant import (
-    apply_smoothing,
-)
-from tico.experimental.quantization.config.smoothquant import SmoothQuantConfig
-from tico.experimental.quantization.quantizer import BaseQuantizer
-from tico.experimental.quantization.quantizer_registry import register_quantizer
+from tico.quantization.algorithm.smoothquant.observer import ChannelwiseMaxActsObserver
+
+from tico.quantization.algorithm.smoothquant.smooth_quant import apply_smoothing
+from tico.quantization.config.smoothquant import SmoothQuantConfig
+from tico.quantization.quantizer import BaseQuantizer
+from tico.quantization.quantizer_registry import register_quantizer
 
 
 @register_quantizer(SmoothQuantConfig)

tico/{experimental/quantization → quantization}/config/gptq.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from tico.experimental.quantization.config.base import BaseConfig
+from tico.quantization.config.base import BaseConfig
 
 
 class GPTQConfig(BaseConfig):

tico/{experimental/quantization → quantization}/config/pt2e.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from tico.experimental.quantization.config.base import BaseConfig
+from tico.quantization.config.base import BaseConfig
 
 
 class PT2EConfig(BaseConfig):

tico/{experimental/quantization → quantization}/config/ptq.py

@@ -15,11 +15,11 @@
 from dataclasses import dataclass, field
 from typing import Any, Dict, Mapping, Type
 
-from tico.experimental.quantization.config.base import BaseConfig
-from tico.experimental.quantization.ptq.dtypes import DType
-from tico.experimental.quantization.ptq.observers.base import ObserverBase
-from tico.experimental.quantization.ptq.observers.minmax import MinMaxObserver
-from tico.experimental.quantization.ptq.qscheme import QScheme
+from tico.quantization.config.base import BaseConfig
+from tico.quantization.wrapq.dtypes import DType
+from tico.quantization.wrapq.observers.base import ObserverBase
+from tico.quantization.wrapq.observers.minmax import MinMaxObserver
+from tico.quantization.wrapq.qscheme import QScheme
 
 
 @dataclass
@@ -55,7 +55,7 @@ class PTQConfig(BaseConfig):
     Example
     -------
     ```python
-    from ptq.observers import PercentileObserver
+    from wrapq.observers import PercentileObserver
 
     cfg = PTQConfig(
         default_dtype   = DType.uint(8),

tico/{experimental/quantization → quantization}/config/smoothquant.py

@@ -14,7 +14,7 @@
 
 from typing import Dict, Literal, Optional
 
-from tico.experimental.quantization.config.base import BaseConfig
+from tico.quantization.config.base import BaseConfig
 
 
 class SmoothQuantConfig(BaseConfig):

tico/{experimental/quantization → quantization}/evaluation/evaluate.py

@@ -20,18 +20,12 @@ import torch
 from circle_schema import circle
 from torch.utils import _pytree as pytree
 
-from tico.experimental.quantization.evaluation.backend import BACKEND
-from tico.experimental.quantization.evaluation.executor.backend_executor import (
-    BackendExecutor,
-)
-from tico.experimental.quantization.evaluation.executor.circle_executor import (
-    CircleExecutor,
-)
-from tico.experimental.quantization.evaluation.executor.triv24_executor import (
-    Triv24Executor,
-)
-from tico.experimental.quantization.evaluation.metric import MetricCalculator
-from tico.experimental.quantization.evaluation.utils import (
+from tico.quantization.evaluation.backend import BACKEND
+from tico.quantization.evaluation.executor.backend_executor import BackendExecutor
+from tico.quantization.evaluation.executor.circle_executor import CircleExecutor
+from tico.quantization.evaluation.executor.triv24_executor import Triv24Executor
+from tico.quantization.evaluation.metric import MetricCalculator
+from tico.quantization.evaluation.utils import (
     ensure_list,
     find_invalid_types,
     get_graph_input_output,

tico/{experimental/quantization → quantization}/evaluation/executor/circle_executor.py

@@ -19,9 +19,7 @@ from typing import List
 import numpy as np
 import torch
 
-from tico.experimental.quantization.evaluation.executor.backend_executor import (
-    BackendExecutor,
-)
+from tico.quantization.evaluation.executor.backend_executor import BackendExecutor
 from tico.utils.model import CircleModel
 from tico.utils.utils import run_bash_cmd
 

tico/{experimental/quantization → quantization}/evaluation/executor/triv24_executor.py

@@ -20,10 +20,8 @@ import numpy as np
 import torch
 from circle_schema import circle
 
-from tico.experimental.quantization.evaluation.executor.backend_executor import (
-    BackendExecutor,
-)
-from tico.experimental.quantization.evaluation.utils import (
+from tico.quantization.evaluation.executor.backend_executor import BackendExecutor
+from tico.quantization.evaluation.utils import (
     dequantize,
     get_graph_input_output,
     quantize,

tico/{experimental/quantization → quantization}/public_interface.py

@@ -17,11 +17,11 @@ from typing import Any, Dict, Optional
 
 import torch
 
-from tico.experimental.quantization.algorithm.gptq.quantizer import GPTQQuantizer
-from tico.experimental.quantization.algorithm.pt2e.quantizer import PT2EQuantizer
-from tico.experimental.quantization.config.base import BaseConfig
-from tico.experimental.quantization.quantizer import BaseQuantizer
-from tico.experimental.quantization.quantizer_registry import get_quantizer
+from tico.quantization.algorithm.gptq.quantizer import GPTQQuantizer
+from tico.quantization.algorithm.pt2e.quantizer import PT2EQuantizer
+from tico.quantization.config.base import BaseConfig
+from tico.quantization.quantizer import BaseQuantizer
+from tico.quantization.quantizer_registry import get_quantizer
 
 
 QUANTIZER_ATTRIBUTE_NAME = "tico_quantizer"

tico/{experimental/quantization → quantization}/quantizer.py

@@ -17,7 +17,7 @@ from typing import Any, Dict, Optional
 
 import torch
 
-from tico.experimental.quantization.config.base import BaseConfig
+from tico.quantization.config.base import BaseConfig
 
 
 class BaseQuantizer(ABC):

tico/{experimental/quantization → quantization}/quantizer_registry.py

@@ -15,8 +15,8 @@
 import importlib
 from typing import Dict, Optional, Type, TypeVar
 
-from tico.experimental.quantization.config.base import BaseConfig
-from tico.experimental.quantization.quantizer import BaseQuantizer
+from tico.quantization.config.base import BaseConfig
+from tico.quantization.quantizer import BaseQuantizer
 
 TQ = TypeVar("TQ", bound=BaseQuantizer)
 
@@ -54,12 +54,10 @@ def get_quantizer(cfg: BaseConfig) -> BaseQuantizer:
     name = getattr(cfg, "name", None)
     if name:
         if name == "ptq":
-            importlib.import_module(f"tico.experimental.quantization.ptq.quantizer")
+            importlib.import_module(f"tico.quantization.wrapq.quantizer")
         else:
             try:
-                importlib.import_module(
-                    f"tico.experimental.quantization.algorithm.{name}.quantizer"
-                )
+                importlib.import_module(f"tico.quantization.algorithm.{name}.quantizer")
             except Exception as e:
                 raise RuntimeError(
                     f"Failed to import quantizer module for config name='{name}': {e}"

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/compare_ppl.py

@@ -28,9 +28,9 @@ import tqdm
 from datasets import load_dataset
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.ptq.utils.metrics import perplexity
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.wrapq.utils.metrics import perplexity
 
 # Token-budget presets for activation calibration
 TOKENS: dict[str, int] = {

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/debug_quant_outputs.py

@@ -38,14 +38,14 @@ import tqdm
 from datasets import load_dataset
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.ptq.utils.introspection import (
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.wrapq.utils.introspection import (
     build_fqn_map,
     compare_layer_outputs,
     save_fp_outputs,
 )
-from tico.experimental.quantization.ptq.wrappers.ptq_wrapper import PTQWrapper
+from tico.quantization.wrapq.wrappers.ptq_wrapper import PTQWrapper
 
 # Token-budget presets for activation calibration
 TOKENS: dict[str, int] = {

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_linear.py

@@ -29,12 +29,12 @@ import pathlib
 import torch
 import torch.nn as nn
 
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.evaluation.metric import compute_peir
-from tico.experimental.quantization.evaluation.utils import plot_two_outputs
-from tico.experimental.quantization.ptq.mode import Mode
-from tico.experimental.quantization.ptq.wrappers.nn.quant_linear import QuantLinear
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.evaluation.metric import compute_peir
+from tico.quantization.evaluation.utils import plot_two_outputs
+from tico.quantization.wrapq.mode import Mode
+from tico.quantization.wrapq.wrappers.nn.quant_linear import QuantLinear
 from tico.utils.utils import SuppressWarning
 
 

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_attn.py

@@ -17,14 +17,12 @@ import pathlib
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.evaluation.metric import compute_peir
-from tico.experimental.quantization.evaluation.utils import plot_two_outputs
-from tico.experimental.quantization.ptq.mode import Mode
-from tico.experimental.quantization.ptq.wrappers.llama.quant_attn import (
-    QuantLlamaAttention,
-)
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.evaluation.metric import compute_peir
+from tico.quantization.evaluation.utils import plot_two_outputs
+from tico.quantization.wrapq.mode import Mode
+from tico.quantization.wrapq.wrappers.llama.quant_attn import QuantLlamaAttention
 from tico.utils.utils import SuppressWarning
 
 name = "Maykeye/TinyLLama-v0"

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_decoder_layer.py

@@ -31,12 +31,12 @@ import pathlib
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.evaluation.metric import compute_peir
-from tico.experimental.quantization.evaluation.utils import plot_two_outputs
-from tico.experimental.quantization.ptq.mode import Mode
-from tico.experimental.quantization.ptq.wrappers.llama.quant_decoder_layer import (
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.evaluation.metric import compute_peir
+from tico.quantization.evaluation.utils import plot_two_outputs
+from tico.quantization.wrapq.mode import Mode
+from tico.quantization.wrapq.wrappers.llama.quant_decoder_layer import (
     QuantLlamaDecoderLayer,
 )
 from tico.utils.utils import SuppressWarning

tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_mlp.py

@@ -18,14 +18,14 @@ import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
 import tico
-from tico.experimental.quantization import convert, prepare
-from tico.experimental.quantization.config.ptq import PTQConfig
-from tico.experimental.quantization.evaluation.metric import compute_peir
-from tico.experimental.quantization.evaluation.utils import plot_two_outputs
-from tico.experimental.quantization.ptq.dtypes import INT16
-from tico.experimental.quantization.ptq.mode import Mode
-from tico.experimental.quantization.ptq.qscheme import QScheme
-from tico.experimental.quantization.ptq.wrappers.llama.quant_mlp import QuantLlamaMLP
+from tico.quantization import convert, prepare
+from tico.quantization.config.ptq import PTQConfig
+from tico.quantization.evaluation.metric import compute_peir
+from tico.quantization.evaluation.utils import plot_two_outputs
+from tico.quantization.wrapq.dtypes import INT16
+from tico.quantization.wrapq.mode import Mode
+from tico.quantization.wrapq.qscheme import QScheme
+from tico.quantization.wrapq.wrappers.llama.quant_mlp import QuantLlamaMLP
 from tico.utils.utils import SuppressWarning
 
 name = "Maykeye/TinyLLama-v0"