tico 0.1.0.dev250904__py3-none-any.whl → 0.1.0.dev251109__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Potentially problematic release.
This version of tico might be problematic. Click here for more details.
- tico/__init__.py +1 -1
- tico/config/v1.py +5 -0
- tico/passes/cast_mixed_type_args.py +2 -0
- tico/passes/convert_expand_to_slice_cat.py +153 -0
- tico/passes/convert_matmul_to_linear.py +312 -0
- tico/passes/convert_to_relu6.py +1 -1
- tico/passes/decompose_fake_quantize_tensor_qparams.py +4 -3
- tico/passes/ops.py +0 -1
- tico/passes/remove_redundant_expand.py +3 -1
- tico/quantization/__init__.py +6 -0
- tico/quantization/algorithm/fpi_gptq/fpi_gptq.py +161 -0
- tico/quantization/algorithm/fpi_gptq/quantizer.py +179 -0
- tico/{experimental/quantization → quantization}/algorithm/gptq/gptq.py +24 -3
- tico/{experimental/quantization → quantization}/algorithm/gptq/quantizer.py +14 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/annotator.py +6 -8
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/adaptive_avg_pool2d.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/add.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/conv2d.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/div.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/linear.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/mean.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/mul.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/relu6.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/rsqrt.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/sub.py +4 -6
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/spec.py +1 -3
- tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/utils.py +1 -1
- tico/{experimental/quantization → quantization}/algorithm/pt2e/quantizer.py +5 -2
- tico/{experimental/quantization → quantization}/algorithm/pt2e/utils.py +1 -3
- tico/{experimental/quantization → quantization}/algorithm/smoothquant/observer.py +26 -8
- tico/{experimental/quantization → quantization}/algorithm/smoothquant/quantizer.py +28 -9
- tico/quantization/algorithm/smoothquant/smooth_quant.py +327 -0
- tico/quantization/config/base.py +26 -0
- tico/quantization/config/fpi_gptq.py +29 -0
- tico/quantization/config/gptq.py +29 -0
- tico/quantization/config/pt2e.py +25 -0
- tico/{experimental/quantization/ptq/quant_config.py → quantization/config/ptq.py} +18 -10
- tico/{experimental/quantization/config.py → quantization/config/smoothquant.py} +9 -37
- tico/{experimental/quantization → quantization}/evaluation/evaluate.py +6 -12
- tico/{experimental/quantization → quantization}/evaluation/executor/circle_executor.py +3 -4
- tico/{experimental/quantization → quantization}/evaluation/executor/triv24_executor.py +2 -4
- tico/{experimental/quantization → quantization}/evaluation/utils.py +1 -1
- tico/{experimental/quantization → quantization}/public_interface.py +11 -18
- tico/{experimental/quantization → quantization}/quantizer.py +1 -1
- tico/quantization/quantizer_registry.py +73 -0
- tico/quantization/wrapq/examples/compare_ppl.py +230 -0
- tico/quantization/wrapq/examples/debug_quant_outputs.py +224 -0
- tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_linear.py +11 -10
- tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_attn.py +10 -12
- tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_decoder_layer.py +10 -9
- tico/{experimental/quantization/ptq → quantization/wrapq}/examples/quantize_llama_mlp.py +13 -13
- tico/quantization/wrapq/examples/quantize_with_gptq.py +265 -0
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/affine_base.py +3 -3
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/base.py +2 -2
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/ema.py +2 -2
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/identity.py +1 -1
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/minmax.py +2 -2
- tico/{experimental/quantization/ptq → quantization/wrapq}/observers/mx.py +1 -1
- tico/quantization/wrapq/quantizer.py +179 -0
- tico/{experimental/quantization/ptq → quantization/wrapq}/utils/introspection.py +3 -5
- tico/{experimental/quantization/ptq → quantization/wrapq}/utils/metrics.py +3 -2
- tico/quantization/wrapq/wrappers/fairseq/__init__.py +5 -0
- tico/quantization/wrapq/wrappers/fairseq/decoder_export_single_step.py +234 -0
- tico/quantization/wrapq/wrappers/fairseq/quant_decoder.py +429 -0
- tico/quantization/wrapq/wrappers/fairseq/quant_decoder_layer.py +492 -0
- tico/quantization/wrapq/wrappers/fairseq/quant_encoder.py +331 -0
- tico/quantization/wrapq/wrappers/fairseq/quant_encoder_layer.py +163 -0
- tico/quantization/wrapq/wrappers/fairseq/quant_mha.py +381 -0
- tico/quantization/wrapq/wrappers/llama/__init__.py +1 -0
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/llama/quant_attn.py +58 -21
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/llama/quant_decoder_layer.py +21 -13
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/llama/quant_mlp.py +5 -7
- tico/quantization/wrapq/wrappers/nn/__init__.py +1 -0
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/nn/quant_layernorm.py +6 -7
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/nn/quant_linear.py +7 -8
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/nn/quant_silu.py +8 -9
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/ptq_wrapper.py +4 -6
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/quant_elementwise.py +55 -17
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/quant_module_base.py +10 -9
- tico/{experimental/quantization/ptq → quantization/wrapq}/wrappers/registry.py +17 -10
- tico/serialize/circle_serializer.py +11 -4
- tico/serialize/operators/op_constant_pad_nd.py +41 -11
- tico/serialize/operators/op_le.py +54 -0
- tico/serialize/operators/op_mm.py +15 -132
- tico/utils/convert.py +20 -15
- tico/utils/register_custom_op.py +6 -4
- tico/utils/signature.py +7 -8
- tico/utils/validate_args_kwargs.py +12 -0
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/METADATA +48 -2
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/RECORD +128 -108
- tico/experimental/quantization/__init__.py +0 -6
- tico/experimental/quantization/algorithm/smoothquant/smooth_quant.py +0 -164
- tico/experimental/quantization/ptq/examples/compare_ppl.py +0 -121
- tico/experimental/quantization/ptq/examples/debug_quant_outputs.py +0 -129
- tico/experimental/quantization/ptq/examples/quantize_with_gptq.py +0 -165
- /tico/{experimental/quantization → quantization}/algorithm/__init__.py +0 -0
- /tico/{experimental/quantization/algorithm/gptq → quantization/algorithm/fpi_gptq}/__init__.py +0 -0
- /tico/{experimental/quantization/algorithm/pt2e → quantization/algorithm/gptq}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/algorithm/gptq/quant.py +0 -0
- /tico/{experimental/quantization → quantization}/algorithm/gptq/utils.py +0 -0
- /tico/{experimental/quantization/algorithm/pt2e/annotation → quantization/algorithm/pt2e}/__init__.py +0 -0
- /tico/{experimental/quantization/algorithm/pt2e/transformation → quantization/algorithm/pt2e/annotation}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/config.py +0 -0
- /tico/{experimental/quantization → quantization}/algorithm/pt2e/annotation/op/__init__.py +0 -0
- /tico/{experimental/quantization/algorithm/smoothquant → quantization/algorithm/pt2e/transformation}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/algorithm/pt2e/transformation/convert_scalars_to_attrs.py +0 -0
- /tico/{experimental/quantization/evaluation → quantization/algorithm/smoothquant}/__init__.py +0 -0
- /tico/{experimental/quantization/evaluation/executor → quantization/config}/__init__.py +0 -0
- /tico/{experimental/quantization/passes → quantization/evaluation}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/evaluation/backend.py +0 -0
- /tico/{experimental/quantization/ptq → quantization/evaluation/executor}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/evaluation/executor/backend_executor.py +0 -0
- /tico/{experimental/quantization → quantization}/evaluation/metric.py +0 -0
- /tico/{experimental/quantization/ptq/examples → quantization/passes}/__init__.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/fold_quant_ops.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/insert_quantize_on_dtype_mismatch.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/propagate_qparam_backward.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/propagate_qparam_forward.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/quantize_bias.py +0 -0
- /tico/{experimental/quantization → quantization}/passes/remove_weight_dequant_op.py +0 -0
- /tico/{experimental/quantization/ptq/observers → quantization/wrapq}/__init__.py +0 -0
- /tico/{experimental/quantization/ptq → quantization/wrapq}/dtypes.py +0 -0
- /tico/{experimental/quantization/ptq/utils → quantization/wrapq/examples}/__init__.py +0 -0
- /tico/{experimental/quantization/ptq → quantization/wrapq}/mode.py +0 -0
- /tico/{experimental/quantization/ptq/wrappers → quantization/wrapq/observers}/__init__.py +0 -0
- /tico/{experimental/quantization/ptq → quantization/wrapq}/qscheme.py +0 -0
- /tico/{experimental/quantization/ptq/wrappers/llama → quantization/wrapq/utils}/__init__.py +0 -0
- /tico/{experimental/quantization/ptq → quantization/wrapq}/utils/reduce_utils.py +0 -0
- /tico/{experimental/quantization/ptq/wrappers/nn → quantization/wrapq/wrappers}/__init__.py +0 -0
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/LICENSE +0 -0
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/WHEEL +0 -0
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/entry_points.txt +0 -0
- {tico-0.1.0.dev250904.dist-info → tico-0.1.0.dev251109.dist-info}/top_level.txt +0 -0
|
File without changes
|
|
File without changes
|