Initial release: first OpenVINO IR of Gemma 4 E4B QAT (int4 sym g32, q4_0-matched, probe-validated)
f3e82a2 verified | <net name="Model9" version="11"> | |
| <layers> | |
| <layer id="0" name="input_ids" type="Parameter" version="opset1"> | |
| <data shape="?,?" element_type="i64" /> | |
| <output> | |
| <port id="0" precision="I64" names="input_ids"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="1" name="self.language_model.embed_tokens_per_layer.weight" type="Const" version="opset1"> | |
| <data element_type="i8" shape="262144, 10752" offset="0" size="2818572288" /> | |
| <output> | |
| <port id="0" precision="I8"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="2" name="Convert_2646531" type="Convert" version="opset1"> | |
| <data destination_type="f16" /> | |
| <input> | |
| <port id="0" precision="I8"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="1" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="3" name="self.language_model.embed_tokens_per_layer.weight/scale" type="Const" version="opset1"> | |
| <data element_type="f16" shape="262144, 1" offset="2818572288" size="524288" /> | |
| <output> | |
| <port id="0" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="4" name="self.language_model.embed_tokens_per_layer.weight/fq_weights_0" type="Multiply" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| <port id="1" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="5" name="self.language_model.embed_tokens_per_layer.weight/fq_weights_0/convert" type="Convert" version="opset1"> | |
| <data destination_type="f32" /> | |
| <input> | |
| <port id="0" precision="FP16"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="1" precision="FP32"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="6" name="Constant_209302" type="Const" version="opset1"> | |
| <data element_type="i64" shape="1, 1" offset="2819096576" size="8" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="7" name="aten::eq/Equal" type="Equal" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="7"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="8" name="8" type="Const" version="opset1"> | |
| <data element_type="i64" shape="" offset="2819096584" size="8" /> | |
| <output> | |
| <port id="0" precision="I64" names="8" /> | |
| </output> | |
| </layer> | |
| <layer id="9" name="aten::full_like/ShapeOf" type="ShapeOf" version="opset3"> | |
| <data output_type="i32" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="1" precision="I32"> | |
| <dim>2</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="10" name="aten::full_like/Broadcast" type="Broadcast" version="opset3"> | |
| <data mode="numpy" /> | |
| <input> | |
| <port id="0" precision="I64" /> | |
| <port id="1" precision="I32"> | |
| <dim>2</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="I64" names="14,24,34"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="11" name="aten::where/Select" type="Select" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="2" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="3" precision="I64" names="15,per_layer_inputs_tokens.3"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="12" name="Constant_209303" type="Const" version="opset1"> | |
| <data element_type="i64" shape="1, 1" offset="2819096592" size="8" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="13" name="aten::eq/Equal_1" type="Equal" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="17"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="14" name="aten::where/Select_1" type="Select" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="2" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="3" precision="I64" names="25,per_layer_inputs_tokens.5"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="15" name="Constant_209304" type="Const" version="opset1"> | |
| <data element_type="i64" shape="1, 1" offset="2819096600" size="8" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="16" name="aten::eq/Equal_2" type="Equal" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="27"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="17" name="aten::where/Select_2" type="Select" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="2" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="3" precision="I64" names="35,per_layer_inputs_tokens"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="18" name="Constant_209305" type="Const" version="opset1"> | |
| <data element_type="i64" shape="1, 1" offset="2819096584" size="8" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="19" name="aten::ge/GreaterEqual" type="GreaterEqual" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="37"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="20" name="Constant_209306" type="Const" version="opset1"> | |
| <data element_type="i64" shape="1, 1" offset="2819096608" size="8" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="21" name="aten::lt/Less" type="Less" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="39"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="22" name="aten::logical_and/LogicalAnd" type="LogicalAnd" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="BOOL" names="40,per_layer_inputs_mask"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="23" name="aten::zeros_like/Constant" type="Const" version="opset1"> | |
| <data element_type="f32" shape="" offset="2819096616" size="4" /> | |
| <output> | |
| <port id="0" precision="FP32" /> | |
| </output> | |
| </layer> | |
| <layer id="24" name="aten::zeros_like/Broadcast" type="Broadcast" version="opset3"> | |
| <data mode="numpy" /> | |
| <input> | |
| <port id="0" precision="FP32" /> | |
| <port id="1" precision="I32"> | |
| <dim>2</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="FP32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="25" name="aten::zeros_like/ConvertLike" type="Convert" version="opset1"> | |
| <data destination_type="i64" /> | |
| <input> | |
| <port id="0" precision="FP32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="1" precision="I64" names="46"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="26" name="aten::where/Select_3" type="Select" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="BOOL"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="2" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="3" precision="I64" names="47,input"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="27" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Convert" type="Convert" version="opset1"> | |
| <data destination_type="i32" /> | |
| <input> | |
| <port id="0" precision="I64"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="1" precision="I32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="28" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Constant" type="Const" version="opset1"> | |
| <data element_type="i32" shape="" offset="2819096616" size="4" /> | |
| <output> | |
| <port id="0" precision="I32" /> | |
| </output> | |
| </layer> | |
| <layer id="29" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Gather" type="Gather" version="opset8"> | |
| <data batch_dims="0" /> | |
| <input> | |
| <port id="0" precision="FP32"> | |
| <dim>262144</dim> | |
| <dim>10752</dim> | |
| </port> | |
| <port id="1" precision="I32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| </port> | |
| <port id="2" precision="I32" /> | |
| </input> | |
| <output> | |
| <port id="3" precision="FP32" names="68"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="30" name="Constant_209307" type="Const" version="opset1"> | |
| <data element_type="f32" shape="1, 1, 1" offset="2819096620" size="4" /> | |
| <output> | |
| <port id="0" precision="FP32"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="31" name="__module.language_model.embed_tokens_per_layer/aten::mul/Multiply" type="Multiply" version="opset1"> | |
| <data auto_broadcast="numpy" /> | |
| <input> | |
| <port id="0" precision="FP32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>10752</dim> | |
| </port> | |
| <port id="1" precision="FP32"> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| <dim>1</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="FP32" names="70"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>10752</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="32" name="Constant_209339" type="Const" version="opset1"> | |
| <data element_type="i64" shape="4" offset="2819096624" size="32" /> | |
| <output> | |
| <port id="0" precision="I64"> | |
| <dim>4</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="33" name="aten::reshape/Reshape" type="Reshape" version="opset1"> | |
| <data special_zero="true" /> | |
| <input> | |
| <port id="0" precision="FP32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>10752</dim> | |
| </port> | |
| <port id="1" precision="I64"> | |
| <dim>4</dim> | |
| </port> | |
| </input> | |
| <output> | |
| <port id="2" precision="FP32" names="text_embeds_per_layer"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>42</dim> | |
| <dim>256</dim> | |
| </port> | |
| </output> | |
| </layer> | |
| <layer id="34" name="Result_207002" type="Result" version="opset1" output_names="text_embeds_per_layer"> | |
| <input> | |
| <port id="0" precision="FP32"> | |
| <dim>-1</dim> | |
| <dim>-1</dim> | |
| <dim>42</dim> | |
| <dim>256</dim> | |
| </port> | |
| </input> | |
| </layer> | |
| </layers> | |
| <edges> | |
| <edge from-layer="0" from-port="0" to-layer="7" to-port="0" /> | |
| <edge from-layer="0" from-port="0" to-layer="9" to-port="0" /> | |
| <edge from-layer="0" from-port="0" to-layer="11" to-port="2" /> | |
| <edge from-layer="1" from-port="0" to-layer="2" to-port="0" /> | |
| <edge from-layer="2" from-port="1" to-layer="4" to-port="0" /> | |
| <edge from-layer="3" from-port="0" to-layer="4" to-port="1" /> | |
| <edge from-layer="4" from-port="2" to-layer="5" to-port="0" /> | |
| <edge from-layer="5" from-port="1" to-layer="29" to-port="0" /> | |
| <edge from-layer="6" from-port="0" to-layer="7" to-port="1" /> | |
| <edge from-layer="7" from-port="2" to-layer="11" to-port="0" /> | |
| <edge from-layer="8" from-port="0" to-layer="10" to-port="0" /> | |
| <edge from-layer="9" from-port="1" to-layer="10" to-port="1" /> | |
| <edge from-layer="9" from-port="1" to-layer="24" to-port="1" /> | |
| <edge from-layer="10" from-port="2" to-layer="11" to-port="1" /> | |
| <edge from-layer="10" from-port="2" to-layer="14" to-port="1" /> | |
| <edge from-layer="10" from-port="2" to-layer="17" to-port="1" /> | |
| <edge from-layer="11" from-port="3" to-layer="13" to-port="0" /> | |
| <edge from-layer="11" from-port="3" to-layer="14" to-port="2" /> | |
| <edge from-layer="12" from-port="0" to-layer="13" to-port="1" /> | |
| <edge from-layer="13" from-port="2" to-layer="14" to-port="0" /> | |
| <edge from-layer="14" from-port="3" to-layer="16" to-port="0" /> | |
| <edge from-layer="14" from-port="3" to-layer="17" to-port="2" /> | |
| <edge from-layer="15" from-port="0" to-layer="16" to-port="1" /> | |
| <edge from-layer="16" from-port="2" to-layer="17" to-port="0" /> | |
| <edge from-layer="17" from-port="3" to-layer="19" to-port="0" /> | |
| <edge from-layer="17" from-port="3" to-layer="21" to-port="0" /> | |
| <edge from-layer="17" from-port="3" to-layer="26" to-port="1" /> | |
| <edge from-layer="18" from-port="0" to-layer="19" to-port="1" /> | |
| <edge from-layer="19" from-port="2" to-layer="22" to-port="0" /> | |
| <edge from-layer="20" from-port="0" to-layer="21" to-port="1" /> | |
| <edge from-layer="21" from-port="2" to-layer="22" to-port="1" /> | |
| <edge from-layer="22" from-port="2" to-layer="26" to-port="0" /> | |
| <edge from-layer="23" from-port="0" to-layer="24" to-port="0" /> | |
| <edge from-layer="24" from-port="2" to-layer="25" to-port="0" /> | |
| <edge from-layer="25" from-port="1" to-layer="26" to-port="2" /> | |
| <edge from-layer="26" from-port="3" to-layer="27" to-port="0" /> | |
| <edge from-layer="27" from-port="1" to-layer="29" to-port="1" /> | |
| <edge from-layer="28" from-port="0" to-layer="29" to-port="2" /> | |
| <edge from-layer="29" from-port="3" to-layer="31" to-port="0" /> | |
| <edge from-layer="30" from-port="0" to-layer="31" to-port="1" /> | |
| <edge from-layer="31" from-port="2" to-layer="33" to-port="0" /> | |
| <edge from-layer="32" from-port="0" to-layer="33" to-port="1" /> | |
| <edge from-layer="33" from-port="2" to-layer="34" to-port="0" /> | |
| </edges> | |
| <rt_info> | |
| <info name="OpenVINO Runtime" value="2026.3.0-22098-367f60d8408" /> | |
| <Runtime_version value="2026.3.0-22098-367f60d8408" /> | |
| <conversion_parameters> | |
| <framework value="pytorch" /> | |
| <is_python_object value="True" /> | |
| </conversion_parameters> | |
| <nncf> | |
| <friendly_names_were_updated value="True" /> | |
| <version value="3.2.0" /> | |
| <weight_compression> | |
| <advanced_parameters value="{'statistics_path': None, 'lora_adapter_rank': 256, 'group_size_fallback_mode': 'error', 'min_adjusted_group_size': 32, 'awq_params': {'subset_size': 32, 'percent_to_apply': 0.002, 'alpha_min': 0.0, 'alpha_max': 1.0, 'steps': 100, 'prefer_data_aware_scaling': True}, 'scale_estimation_params': {'subset_size': 64, 'initial_steps': 5, 'scale_steps': 5, 'weight_penalty': -1.0}, 'gptq_params': {'damp_percent': 0.1, 'block_size': 128, 'subset_size': 128}, 'lora_correction_params': {'adapter_rank': 8, 'num_iterations': 3, 'apply_regularization': True, 'subset_size': 128, 'use_int8_adapters': True}, 'backend_params': {}, 'codebook': None, 'adaptive_codebook_params': {'value_type': 'f8e4m3', 'across_blocks': False, 'num_elements': 16}}" /> | |
| <all_layers value="False" /> | |
| <awq value="False" /> | |
| <backup_mode value="int8_asym" /> | |
| <compression_format value="dequantize" /> | |
| <gptq value="False" /> | |
| <group_size value="-1" /> | |
| <ignored_scope value="[]" /> | |
| <lora_correction value="False" /> | |
| <mode value="int8_sym" /> | |
| <ratio value="1.0" /> | |
| <scale_estimation value="False" /> | |
| <sensitivity_metric value="weight_quantization_error" /> | |
| </weight_compression> | |
| </nncf> | |
| <optimum> | |
| <nncf_version value="3.2.0" /> | |
| <optimum_intel_version value="1.27.0.dev0+e6f612a" /> | |
| <optimum_version value="2.1.0.dev0" /> | |
| <pytorch_version value="2.12.0+cpu" /> | |
| <transformers_version value="5.5.0" /> | |
| </optimum> | |
| </rt_info> | |
| </net> | |