Layers that will be compiled: group_conv_0 single_conv_0 group_conv_1 single_conv_1 group_pre_2 group_post_2 single_pre_2 single_post_2 group_conv_3 single_conv_3 group_conv_4 single_conv_4 group_pre_5 group_post_5 single_pre_5 single_post_5 group_conv_6 single_conv_6 group_conv_7 single_conv_7 group_pre_8 group_post_8 single_pre_8 single_post_8 group_conv_9 single_conv_9 group_pre_10 group_post_10 single_pre_10 single_post_10 group_conv_11 single_conv_11 group_pre_12 group_post_12 single_pre_12 single_post_12 group_conv_13 single_conv_13 group_pre_14 group_post_14 single_pre_14 single_post_14 group_conv_15 single_conv_15 group_cache_0 group_cache_128 group_cache_256 group_cache_384 group_cache_512 group_cache_640 group_cache_768 group_cache_896 group_cache_1024 group_cache_1152 group_cache_1280 group_cache_1408 group_cache_1536 group_cache_1664 group_cache_1792 group_cache_1920 single_cache_127 single_cache_255 single_cache_383 single_cache_511 single_cache_639 single_cache_767 single_cache_895 single_cache_1023 single_cache_1151 single_cache_1279 single_cache_1407 single_cache_1535 single_cache_1663 single_cache_1791 single_cache_1919 single_cache_2047 conv_post_final_15 2026-03-07 05:02:37,869 - sima_lmm.model.vision_language_model - INFO - Generating FileGenMode.DEVKIT files... Generated all mode=DEVKIT files 2026-03-07 05:02:37,880 - sima_lmm.model.vision_language_model - INFO - Generating FileGenMode.SOURCE_TO_ONNX files... 2026-03-07 05:02:38,464 - sima_lmm.model.vision_language_model - INFO - FileGenMode.SOURCE_TO_ONNX files generation completed. Generated all mode=SOURCE_TO_ONNX files 2026-03-07 05:02:38,465 - sima_lmm.model.vision_language_model - INFO - Generating FileGenMode.ONNX_TO_QUANT files... 2026-03-07 05:03:01,607 - sima_lmm.model.vision_language_model - INFO - FileGenMode.ONNX_TO_QUANT files generation completed. Generated all mode=ONNX_TO_QUANT files 2026-03-07 05:03:01,607 - sima_lmm.model.vision_language_model - INFO - Generating FileGenMode.MODEL_SDK_COMPILE files... 2026-03-07 05:03:21,377 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer0_conv.sima 2026-03-07 05:03:21,423 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:21,423 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer0_conv" 2026-03-07 05:03:21,427 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:21,427 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:21,439 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:21,439 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:21,440 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:21,440 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:21,446 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer1_conv.sima 2026-03-07 05:03:21,481 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:21,481 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer1_conv" 2026-03-07 05:03:21,484 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:21,484 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:21,497 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:21,497 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:21,498 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:21,498 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:21,511 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:21,538 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:21,559 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:21,565 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer2.sima 2026-03-07 05:03:21,570 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:21,590 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:21,590 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer2" 2026-03-07 05:03:21,594 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:21,595 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:21,598 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:21,599 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:21,600 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:21,600 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:21,600 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:21,604 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer1_conv.sima 2026-03-07 05:03:21,613 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer0_conv.sima 2026-03-07 05:03:21,619 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:21,620 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:21,620 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:21,635 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:21,653 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:21,653 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer1_conv" 2026-03-07 05:03:21,657 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:21,657 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer0_conv" 2026-03-07 05:03:21,658 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:21,658 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:21,662 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:21,662 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:21,678 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:21,679 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:21,679 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:21,679 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:21,679 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:21,680 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:21,683 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:21,683 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:21,683 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:21,683 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:21,693 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:24,026 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer2.sima 2026-03-07 05:03:24,064 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:24,064 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer2" 2026-03-07 05:03:24,068 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:24,068 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:24,084 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:24,085 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:24,085 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:24,085 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:24,617 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:24,637 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:24,714 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:24,734 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:26,360 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:26,459 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:26,942 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:27,032 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:27,041 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:27,046 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:27,132 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:27,146 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:27,213 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:27,258 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:27,489 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:27,635 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:27,683 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:27,690 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:27,737 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:27,829 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:27,829 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:27,843 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:27,877 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:27,877 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:27,890 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:27,890 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:27,928 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:28,029 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:28,029 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:28,045 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:28,655 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:28,722 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:28,764 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:28,952 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:28,953 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:28,963 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:29,118 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.981 2026-03-07 05:03:29,119 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:29,219 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.98 2026-03-07 05:03:29,220 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:29,700 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:29,759 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:29,759 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:29,804 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:29,863 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:29,863 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:35,686 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:03:35,776 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:03:37,215 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:37,352 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:37,382 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:37,468 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:37,474 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:37,603 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:37,710 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:37,820 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:44,373 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:45,050 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:45,447 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:45,519 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:46,644 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:03:46,685 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:03:46,897 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:47,135 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:47,800 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:48,026 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:48,303 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:48,382 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:48,527 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,593 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer0_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer0_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer0_conv_mpk.json, llima-compile 2026-03-07 05:03:48,607 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:48,613 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:03:48,614 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer1_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer1_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer1_conv_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:03:48,786 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer2.sima 2026-03-07 05:03:48,793 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.97 2026-03-07 05:03:48,793 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:48,813 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:48,813 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer2" 2026-03-07 05:03:48,814 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:48,815 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:48,819 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:48,819 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:48,819 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:48,819 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:48,861 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:48,861 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:48,865 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:48,877 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:48,923 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:48,924 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:48,933 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:48,974 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer2.sima 2026-03-07 05:03:48,995 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:03:48,996 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer2" 2026-03-07 05:03:48,999 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:03:48,999 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:03:49,004 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:03:49,004 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:03:49,004 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:03:49,004 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:03:49,100 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:49,100 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:49,125 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:03:49,138 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:03:49,172 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:03:49,188 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:03:49,189 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:03:49,197 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:03:49,907 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:50,497 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:50,516 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:50,913 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:03:50,929 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:03:50,966 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:51,520 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:51,569 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:52,027 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.955 2026-03-07 05:03:52,027 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:52,037 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:52,191 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:52,191 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:52,227 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:52,265 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:03:52,334 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.974 2026-03-07 05:03:52,334 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:52,417 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:52,513 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.974 2026-03-07 05:03:52,513 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:52,553 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:52,595 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:52,618 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:52,624 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:52,644 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:03:52,683 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:52,683 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:52,712 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:03:52,724 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:03:52,864 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:52,864 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:52,881 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.966 2026-03-07 05:03:52,882 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:52,899 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:52,924 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:52,924 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:53,957 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:03:54,394 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.978 2026-03-07 05:03:54,394 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:03:54,458 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:03:54,502 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:03:54,502 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:03:56,231 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:03:56,235 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:03:59,263 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:00,661 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer2_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_pre_layer2_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer2_mpk.json, llima-compile 2026-03-07 05:04:01,124 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer3_conv.sima 2026-03-07 05:04:01,163 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:01,163 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer3_conv" 2026-03-07 05:04:01,167 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:01,168 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:01,187 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:01,188 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:01,188 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:01,188 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:01,294 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:01,294 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:04:05,818 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:04:06,115 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:04:06,409 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,699 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer2_mpk.json, models--LiquidAI--LFM2-350M_language_n1_pre_layer2_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_pre_layer2_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:04:06,837 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:06,891 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:06,915 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:06,916 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer2_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_post_layer2_mpk.json, models--LiquidAI--LFM2-350M_language_n128_post_layer2_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:04:06,971 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer3_conv.sima 2026-03-07 05:04:07,000 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:07,000 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer3_conv" 2026-03-07 05:04:07,003 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:07,003 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:07,009 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:07,010 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:07,010 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:07,010 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:07,026 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:07,027 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:07,040 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:07,082 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:07,110 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:07,124 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer2_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_post_layer2_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer2_stage1_mla.elf, llima-compile 2026-03-07 05:04:07,131 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:07,190 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:07,191 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:07,204 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:07,384 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer4_conv.sima 2026-03-07 05:04:07,395 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:07,400 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:07,403 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 6s. 2026-03-07 05:04:07,408 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:04:07,420 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:07,420 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer4_conv" 2026-03-07 05:04:07,425 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:07,425 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:07,446 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:07,446 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:07,446 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:07,447 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:07,804 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer4_conv.sima 2026-03-07 05:04:07,831 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:07,831 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer4_conv" 2026-03-07 05:04:07,834 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:07,834 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:07,839 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:07,840 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:07,840 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:07,840 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:07,911 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:07,939 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:07,959 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:08,000 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:08,000 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:08,013 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:10,237 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:10,258 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:10,989 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:11,010 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:11,990 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:12,570 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:12,664 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:12,680 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:12,735 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:13,018 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:13,320 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:13,413 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:13,428 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:13,438 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:13,492 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:13,624 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:13,624 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:13,637 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:14,724 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.979 2026-03-07 05:04:14,724 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,744 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer0_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer0_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_layer0_conv_mpk.json, llima-compile 2026-03-07 05:04:14,752 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:14,753 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer1_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer1_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer1_conv_stage1_mla.elf, llima-compile 2026-03-07 05:04:14,803 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:14,862 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:14,862 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:15,100 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer5.sima 2026-03-07 05:04:15,121 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer5.sima 2026-03-07 05:04:15,133 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:15,133 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer5" 2026-03-07 05:04:15,137 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:15,137 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:15,138 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:15,138 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer5" 2026-03-07 05:04:15,142 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:15,142 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:15,146 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:15,147 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:15,484 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.973 2026-03-07 05:04:15,485 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:15,563 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:15,622 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:15,623 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:16,593 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:16,704 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:19,815 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:19,885 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:19,928 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:20,111 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:20,112 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:20,121 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:20,629 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:20,805 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:21,031 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:21,067 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:21,166 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:21,167 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:21,181 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:21,436 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:23,069 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:23,180 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:25,379 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:26,211 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:04:26,284 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:26,821 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:26,906 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:27,270 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:04:27,981 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:28,071 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:28,114 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer3_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer3_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer3_conv_stage1_mla.elf, llima-compile 2026-03-07 05:04:28,495 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer5.sima 2026-03-07 05:04:28,512 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:28,512 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer5" 2026-03-07 05:04:28,515 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:28,516 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:28,520 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:28,520 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:28,520 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:28,520 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:28,642 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:28,655 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:28,688 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:28,704 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:28,705 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:28,713 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:29,077 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer4_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer4_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer4_conv_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:04:29,203 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer5.sima 2026-03-07 05:04:29,223 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:29,223 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer5" 2026-03-07 05:04:29,225 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:29,225 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:29,229 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:29,229 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:29,229 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:29,229 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:29,271 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:29,275 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:29,287 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:29,327 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:29,331 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:29,343 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:30,467 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:30,486 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:30,858 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.973 2026-03-07 05:04:30,858 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:30,940 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:31,021 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:31,162 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:31,209 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:31,209 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:31,772 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:31,778 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:31,787 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:32,062 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:32,128 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:32,135 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:32,389 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.964 2026-03-07 05:04:32,389 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:32,405 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:32,431 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:32,432 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:33,069 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:33,193 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:33,463 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:33,495 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:33,564 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:33,575 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:34,205 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:34,761 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:34,858 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:34,904 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:35,253 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.977 2026-03-07 05:04:35,253 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:35,317 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:35,361 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:35,361 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:35,577 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:35,986 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:36,054 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:39,154 - mlc.test_util.test_context - INFO - Compression done in 4s. Compression ratio: 0.964 2026-03-07 05:04:39,154 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:39,244 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:39,376 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.969 2026-03-07 05:04:39,377 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:39,445 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:39,525 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:39,525 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:39,678 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:39,678 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:40,155 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:42,757 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:43,802 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:44,341 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:44,388 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:44,841 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.954 2026-03-07 05:04:44,841 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:04:44,860 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:04:45,015 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:04:45,015 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:04:45,661 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:45,661 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:04:49,105 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:49,107 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:04:51,917 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:04:51,930 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:04:52,322 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,323 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer3_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer3_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer3_conv_stage1_mla.elf, llima-compile 2026-03-07 05:04:52,395 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:52,401 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:52,494 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer5_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_pre_layer5_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_pre_layer5_mpk.json, llima-compile 2026-03-07 05:04:52,783 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer6_conv.sima 2026-03-07 05:04:52,791 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer6_conv.sima 2026-03-07 05:04:52,808 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:52,808 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer6_conv" 2026-03-07 05:04:52,811 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:52,811 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:52,817 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:52,818 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:52,818 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:52,818 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:52,824 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:52,824 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer6_conv" 2026-03-07 05:04:52,828 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:52,828 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:52,840 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:52,841 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:52,841 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:52,841 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:52,890 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:52,918 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:52,938 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:52,984 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:52,985 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:52,998 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,256 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer5_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_post_layer5_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer5_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:04:53,430 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:53,430 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:53,431 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer5_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer5_mpk.json, models--LiquidAI--LFM2-350M_language_n128_pre_layer5_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:04:53,705 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer7_conv.sima 2026-03-07 05:04:53,721 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer7_conv.sima 2026-03-07 05:04:53,734 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:53,734 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer7_conv" 2026-03-07 05:04:53,737 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:53,737 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:53,748 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:53,749 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:53,749 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:53,749 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:53,758 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:53,758 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer7_conv" 2026-03-07 05:04:53,762 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:53,762 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:53,774 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:53,774 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:53,774 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:53,774 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:53,820 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:53,848 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:53,869 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:53,925 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:53,926 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:53,939 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:54,225 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:04:54,235 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:04:56,432 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:56,453 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:56,941 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:04:56,962 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:04:58,198 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:04:58,274 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer5_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_post_layer5_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_post_layer5_mpk.json, llima-compile 2026-03-07 05:04:58,371 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:58,648 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer8.sima 2026-03-07 05:04:58,665 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:04:58,665 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer8" 2026-03-07 05:04:58,669 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:04:58,669 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:04:58,673 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:04:58,674 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:04:58,674 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:04:58,674 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:04:58,679 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:04:58,776 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:58,794 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:58,849 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:58,871 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:58,887 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:58,975 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:58,975 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:58,988 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:04:59,260 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:04:59,318 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:04:59,355 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:04:59,371 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:04:59,746 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:04:59,801 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:04:59,927 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:04:59,927 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:04:59,940 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:00,922 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.978 2026-03-07 05:05:00,922 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:00,978 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer4_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_layer4_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer4_conv_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:05:01,001 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:01,060 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:01,060 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:01,291 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer8.sima 2026-03-07 05:05:01,322 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:01,322 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer8" 2026-03-07 05:05:01,325 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:01,326 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:01,335 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:01,336 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:01,336 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:01,336 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:01,422 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.972 2026-03-07 05:05:01,422 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:01,502 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:01,563 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:01,563 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:04,316 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:04,727 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:04,765 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:04,872 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:04,873 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:04,887 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:06,014 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:06,085 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:06,128 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:06,301 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:06,301 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:06,311 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:06,860 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:07,505 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:08,728 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:08,840 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:09,846 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:09,957 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:11,960 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 2s. 2026-03-07 05:05:12,511 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:13,823 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer6_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer6_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer6_conv_stage1_mla.elf, llima-compile 2026-03-07 05:05:14,201 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer8.sima 2026-03-07 05:05:14,217 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:14,217 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer8" 2026-03-07 05:05:14,220 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:14,220 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:14,224 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:14,225 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:14,225 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:14,225 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:14,303 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:14,346 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:14,359 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:14,390 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:14,392 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:14,407 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:14,408 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:14,417 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:14,417 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:14,418 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer7_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer7_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer7_conv_stage1_mla.elf, llima-compile 2026-03-07 05:05:14,470 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:14,542 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer8.sima 2026-03-07 05:05:14,565 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:14,565 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer8" 2026-03-07 05:05:14,566 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:14,567 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:14,571 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:14,571 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:14,571 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:14,571 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:14,612 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:14,612 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:14,616 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:14,628 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:14,663 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:14,668 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:14,680 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:16,133 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:16,151 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:17,110 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:17,126 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:17,449 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:17,725 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:17,729 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:17,795 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:17,802 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:18,055 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.969 2026-03-07 05:05:18,056 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:18,073 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:18,098 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:18,098 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:18,429 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:18,627 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:18,758 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:18,860 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:18,932 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:18,944 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:19,139 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:19,150 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:19,233 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:19,655 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:20,168 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:20,249 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:20,627 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.977 2026-03-07 05:05:20,628 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:20,693 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:20,738 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:20,739 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:21,418 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:22,103 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:22,512 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:22,583 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:23,226 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.972 2026-03-07 05:05:23,226 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:23,309 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:23,584 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:23,585 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:24,219 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.963 2026-03-07 05:05:24,220 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:24,306 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:24,578 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:24,579 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:25,543 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:25,840 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.969 2026-03-07 05:05:25,840 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:25,908 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:26,145 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:26,146 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:26,172 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:27,226 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:27,766 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:27,814 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:28,269 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.958 2026-03-07 05:05:28,269 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:28,279 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:28,433 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:28,433 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:31,071 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:05:31,400 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:05:31,652 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:31,653 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer8_mpk.json, models--LiquidAI--LFM2-350M_language_n1_pre_layer8_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_pre_layer8_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:05:32,115 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer9_conv.sima 2026-03-07 05:05:32,147 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:32,147 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer9_conv" 2026-03-07 05:05:32,152 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:32,152 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:32,164 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:32,165 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:32,165 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:32,165 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:32,581 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:32,584 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:32,706 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer8_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer8_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_post_layer8_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:05:33,001 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer9_conv.sima 2026-03-07 05:05:33,034 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:33,034 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer9_conv" 2026-03-07 05:05:33,037 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:33,037 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:33,043 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:33,043 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:33,043 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:33,043 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:33,116 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:33,144 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:33,165 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:33,208 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:33,210 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:33,604 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:36,586 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:36,607 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:36,943 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:36,944 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer8_mpk.json, models--LiquidAI--LFM2-350M_language_n128_pre_layer8_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer8_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:05:37,673 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:37,691 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer10.sima 2026-03-07 05:05:37,707 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:37,708 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer10" 2026-03-07 05:05:37,711 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:37,711 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:37,715 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:37,716 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:37,716 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:37,716 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:38,096 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:38,152 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:38,284 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:38,285 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:38,298 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:38,325 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:38,560 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:38,560 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:05:38,631 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:38,638 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:05:38,871 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:38,871 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:05:38,913 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:39,009 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:39,024 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:41,064 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.969 2026-03-07 05:05:41,064 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:41,144 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:41,204 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:41,205 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:43,177 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:43,580 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:43,618 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:43,720 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:43,721 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:43,735 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:44,978 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer8_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_post_layer8_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_post_layer8_mpk.json, llima-compile 2026-03-07 05:05:45,291 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer10.sima 2026-03-07 05:05:45,318 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:45,319 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer10" 2026-03-07 05:05:45,322 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:45,322 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:45,332 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:45,332 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:45,332 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:45,333 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,177 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer6_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer6_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer6_conv_stage1_mla.elf, llima-compile 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:46,318 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:46,319 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:46,319 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer7_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer7_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer7_conv_stage1_mla.elf, llima-compile 2026-03-07 05:05:46,443 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer10.sima 2026-03-07 05:05:46,462 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:46,462 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer10" 2026-03-07 05:05:46,464 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:46,464 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:46,468 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:46,468 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:46,468 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:46,468 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:46,510 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:46,514 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:46,525 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:46,561 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer10.sima 2026-03-07 05:05:46,565 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:46,566 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:46,573 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:46,578 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:46,578 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer10" 2026-03-07 05:05:46,582 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:46,582 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:46,586 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:46,587 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:46,587 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:46,587 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:46,716 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:46,730 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:46,764 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:46,783 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:46,783 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:46,792 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:47,123 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:47,773 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:47,885 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:48,143 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:48,163 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:48,943 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:48,958 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:49,482 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:49,774 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:49,844 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:49,851 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:49,955 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:05:50,026 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:05:50,068 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:05:50,109 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.957 2026-03-07 05:05:50,109 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:50,126 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:50,153 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:50,153 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:50,241 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:50,246 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:05:50,247 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:05:50,256 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:05:50,660 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:50,729 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:50,739 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:05:51,213 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:52,481 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.969 2026-03-07 05:05:52,481 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:05:53,019 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:05:53,063 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:05:53,064 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:05:53,720 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:53,859 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:56,110 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:05:57,418 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:05:57,727 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:05:58,001 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:05:58,002 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer9_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer9_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer9_conv_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:05:58,184 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:05:58,275 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:05:58,312 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:05:58,470 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer11_conv.sima 2026-03-07 05:05:58,506 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:05:58,506 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer11_conv" 2026-03-07 05:05:58,510 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:05:58,510 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:05:58,522 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:05:58,523 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:05:58,523 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:05:58,523 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:05:58,822 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:05:58,907 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:02,834 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.959 2026-03-07 05:06:02,834 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:02,916 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:03,185 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:03,186 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:04,049 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:04,150 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:06:04,153 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:04,472 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:04,526 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:04,658 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:04,658 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:04,671 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:04,731 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:04,732 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:06:04,732 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:04,732 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:04,732 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer10_mpk.json, models--LiquidAI--LFM2-350M_language_n1_pre_layer10_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_pre_layer10_stage1_mla.elf, llima-compile 2026-03-07 05:06:05,003 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer11_conv.sima 2026-03-07 05:06:05,027 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:05,027 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer11_conv" 2026-03-07 05:06:05,030 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:05,030 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:05,036 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:05,036 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:05,036 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:05,036 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:05,109 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:05,137 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:05,158 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:05,214 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:05,215 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:05,476 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:05,477 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:06:05,477 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:05,477 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:05,477 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer10_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_post_layer10_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer10_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:06:05,568 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:05,657 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:05,853 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer12.sima 2026-03-07 05:06:05,873 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:05,874 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer12" 2026-03-07 05:06:05,877 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:05,877 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:05,882 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:05,882 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:05,882 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:05,882 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:06,193 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:06,620 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:06,881 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:07,154 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:07,202 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:07,292 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:07,365 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:07,658 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.945 2026-03-07 05:06:07,658 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:07,668 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:07,821 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:07,821 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:08,830 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:08,851 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:10,583 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:10,636 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.959 2026-03-07 05:06:10,636 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:10,756 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:10,996 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:10,997 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:11,170 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:11,266 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:11,281 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:11,316 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:11,714 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:11,750 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:11,849 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:11,850 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:11,864 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:11,938 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:11,942 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:13,351 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.975 2026-03-07 05:06:13,351 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:13,431 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:13,491 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:13,491 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:13,986 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:14,098 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:16,583 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:06:16,584 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:16,584 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:16,584 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer10_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer10_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_pre_layer10_mpk.json, llima-compile 2026-03-07 05:06:16,911 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer12.sima 2026-03-07 05:06:16,970 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:16,970 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer12" 2026-03-07 05:06:16,978 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:16,979 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:16,993 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:16,993 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:16,993 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:16,993 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:17,477 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:17,488 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:06:19,384 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:21,821 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:21,858 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:21,930 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:21,960 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:21,974 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:22,160 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:22,161 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:22,171 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:23,155 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:23,162 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:06:23,305 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:23,566 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:24,284 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer9_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_layer9_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer9_conv_mpk.json, llima-compile 2026-03-07 05:06:24,457 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:24,659 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer12.sima 2026-03-07 05:06:24,677 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:24,677 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer12" 2026-03-07 05:06:24,680 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:24,680 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:24,684 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:24,685 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:24,685 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:24,685 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:24,805 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:24,818 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:24,851 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:24,868 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:24,868 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:24,876 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:24,974 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:25,056 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:25,157 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer11_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer11_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer11_conv_mpk.json, llima-compile 2026-03-07 05:06:25,287 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer12.sima 2026-03-07 05:06:25,310 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:25,310 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer12" 2026-03-07 05:06:25,312 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:25,312 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:25,316 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:25,316 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:25,316 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:25,316 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:25,358 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:25,363 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:25,374 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:25,421 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:25,421 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:25,430 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:26,188 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:26,207 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:27,949 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:27,965 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:27,974 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:28,252 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:28,320 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:28,327 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:28,582 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.957 2026-03-07 05:06:28,582 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:28,599 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:28,624 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:28,625 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:28,970 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.966 2026-03-07 05:06:28,970 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:29,054 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:29,240 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer10_mpk.json, models--LiquidAI--LFM2-350M_language_n128_post_layer10_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_post_layer10_stage1_mla.elf, llima-compile 2026-03-07 05:06:29,266 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:29,324 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:29,324 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:29,659 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:29,700 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:29,709 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer13_conv.sima 2026-03-07 05:06:29,742 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:29,742 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer13_conv" 2026-03-07 05:06:29,746 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:29,747 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:29,759 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:29,759 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:29,759 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:29,760 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:29,774 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:29,786 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:30,370 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:30,470 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:31,522 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.98 2026-03-07 05:06:31,522 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:31,588 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:31,634 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:31,634 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:33,884 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:34,921 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:35,023 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:35,444 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:35,447 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:35,491 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:35,501 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:35,637 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:35,638 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:35,651 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:35,943 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.948 2026-03-07 05:06:35,943 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:35,953 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:36,104 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:36,104 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:36,436 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:37,408 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:38,085 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:38,490 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:38,558 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:40,143 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:40,148 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:41,807 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.973 2026-03-07 05:06:41,807 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:41,877 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:42,113 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:42,113 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:42,799 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:43,251 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:43,380 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer12_mpk.json, models--LiquidAI--LFM2-350M_language_n1_pre_layer12_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_pre_layer12_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:06:43,645 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:43,652 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:06:43,661 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer13_conv.sima 2026-03-07 05:06:43,688 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:43,688 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer13_conv" 2026-03-07 05:06:43,690 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:43,690 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:43,696 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:43,697 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:43,697 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:43,697 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:43,770 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:43,797 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:43,818 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:43,869 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:43,869 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:43,883 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,516 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer12_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer12_mpk.json, models--LiquidAI--LFM2-350M_language_n128_pre_layer12_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:44,575 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:06:44,576 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:44,576 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:44,576 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer12_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_post_layer12_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer12_stage1_mla.elf, llima-compile 2026-03-07 05:06:44,898 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_pre_layer14.sima 2026-03-07 05:06:44,916 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:44,916 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_pre_layer14" 2026-03-07 05:06:44,917 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_post_layer14.sima 2026-03-07 05:06:44,920 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:44,920 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:44,924 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:44,924 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:44,925 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:44,925 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:44,949 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:44,949 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_post_layer14" 2026-03-07 05:06:44,952 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:44,953 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:44,964 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:44,964 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:44,964 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:44,964 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:45,185 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:45,296 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:46,954 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:46,976 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:48,716 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:49,296 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:49,392 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:49,407 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:49,747 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:49,817 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:49,859 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:50,043 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:50,044 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:50,054 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:50,460 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:06:50,461 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:06:50,461 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:06:50,461 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:06:50,461 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:06:50,461 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer11_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer11_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_layer11_conv_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:06:50,547 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:50,835 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_pre_layer14.sima 2026-03-07 05:06:50,851 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:06:50,851 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_pre_layer14" 2026-03-07 05:06:50,854 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:06:50,855 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:06:50,859 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:06:50,859 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:06:50,860 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:06:50,860 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:06:50,947 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:50,980 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:06:50,984 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:50,994 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:06:51,027 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:06:51,042 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:51,043 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:51,051 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:51,083 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:06:51,084 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:06:51,099 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:06:51,430 - mlc.test_util.test_context - INFO - Compression done in 2s. Compression ratio: 0.978 2026-03-07 05:06:51,430 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:51,509 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:51,568 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:51,569 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:52,401 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:52,420 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:06:53,701 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:53,980 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:54,047 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:54,054 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:54,310 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.94 2026-03-07 05:06:54,310 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:06:54,326 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:06:54,353 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:06:54,353 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:06:54,451 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:54,451 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:06:54,918 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:06:55,429 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:55,887 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:06:56,427 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:06:56,515 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:06:57,522 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:06:58,423 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:06:58,512 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:00,238 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:00,377 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:00,496 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.971 2026-03-07 05:07:00,496 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:00,507 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer12_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_post_layer12_mpk.json, models--LiquidAI--LFM2-350M_language_n128_post_layer12_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:07:00,580 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:00,631 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer14.sima 2026-03-07 05:07:00,651 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:00,651 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer14" 2026-03-07 05:07:00,653 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:00,653 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:00,657 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:00,657 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:00,657 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:00,657 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:00,699 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:00,704 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:00,715 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:00,758 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:00,758 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:00,766 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:00,855 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:00,855 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:01,994 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:03,186 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:03,201 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:03,821 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer13_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer13_conv_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_layer13_conv_mpk.json, llima-compile 2026-03-07 05:07:03,983 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_layer15_conv.sima 2026-03-07 05:07:03,997 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:03,997 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_layer15_conv" 2026-03-07 05:07:03,999 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:03,999 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:04,001 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:04,001 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:04,001 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:04,001 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:04,496 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:04,920 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:04,990 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:05,002 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:05,286 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:05,520 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:05,651 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:05,667 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:05,748 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:05,749 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:05,754 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:06,216 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:06,654 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:06,677 - mlc.test_util.test_context - INFO - Compression done in 1s. Compression ratio: 0.977 2026-03-07 05:07:06,677 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:06,727 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:06,742 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:06,788 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:06,788 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:08,482 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:08,520 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:08,792 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:07:09,362 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:09,363 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:09,363 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_pre_layer14_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_pre_layer14_mpk.json, models--LiquidAI--LFM2-350M_language_n1_pre_layer14_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:07:09,954 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_layer15_conv.sima 2026-03-07 05:07:09,966 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:09,966 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_layer15_conv" 2026-03-07 05:07:09,968 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:09,968 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:09,971 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:09,971 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:09,971 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:09,971 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:09,981 - mlc.test_util.test_context - INFO - Compression done in 3s. Compression ratio: 0.969 2026-03-07 05:07:09,981 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:10,016 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:10,041 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:10,049 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:10,056 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:10,072 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:10,073 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:10,082 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:10,281 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:10,281 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:10,950 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:10,958 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:11,584 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:11,606 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:11,894 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:11,954 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:11,988 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:11,992 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:12,185 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:12,464 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.979 2026-03-07 05:07:12,464 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:12,480 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:12,500 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:12,500 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:12,637 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:12,787 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:12,805 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:12,947 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:13,478 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:13,526 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:13,696 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.972 2026-03-07 05:07:13,696 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:13,715 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:13,775 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:13,775 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:13,968 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:13,993 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.923 2026-03-07 05:07:13,993 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:14,003 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:14,154 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:14,154 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:15,606 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:15,614 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:07:17,029 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:17,345 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:18,269 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:18,272 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:18,695 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer14_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_post_layer14_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_post_layer14_mpk.json, llima-compile 2026-03-07 05:07:18,893 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token0.sima 2026-03-07 05:07:18,904 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:18,904 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token0" 2026-03-07 05:07:18,908 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:18,908 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:18,909 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:18,909 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:18,909 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:18,909 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:21,861 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:22,066 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:22,078 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:22,221 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:22,222 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:22,227 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:22,417 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:22,417 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,418 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer13_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer13_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer13_conv_stage1_mla.elf, llima-compile 2026-03-07 05:07:22,425 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:22,426 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - EV74: 7 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:22,584 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_pre_layer14_mpk.json, models--LiquidAI--LFM2-350M_language_n128_pre_layer14_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_pre_layer14_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:07:22,612 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token128.sima 2026-03-07 05:07:22,621 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:22,621 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token128" 2026-03-07 05:07:22,625 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:22,625 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:22,626 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:22,626 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:22,626 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:22,626 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:22,781 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token256.sima 2026-03-07 05:07:22,789 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:22,790 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token256" 2026-03-07 05:07:22,793 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:22,793 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:22,794 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:22,794 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:22,794 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:22,794 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:24,237 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 2s. 2026-03-07 05:07:25,809 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:26,014 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:26,025 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:26,213 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_layer15_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_layer15_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n128_layer15_conv_stage1_mla.elf, llima-compile 2026-03-07 05:07:26,230 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:26,231 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:26,237 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:26,291 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:26,407 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token384.sima 2026-03-07 05:07:26,416 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:26,416 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token384" 2026-03-07 05:07:26,419 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:26,419 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:26,420 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:26,420 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:26,421 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:26,421 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:26,491 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:26,503 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:26,774 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:26,775 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:26,780 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:28,066 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - EV74: 3 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,230 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_post_layer14_mpk.json, models--LiquidAI--LFM2-350M_language_n128_post_layer14_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_post_layer14_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:07:28,422 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token512.sima 2026-03-07 05:07:28,431 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:28,431 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token512" 2026-03-07 05:07:28,434 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:28,434 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:28,435 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:28,436 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:28,436 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:28,436 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:28,564 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_layer15_conv_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_layer15_conv_mpk.json, models--LiquidAI--LFM2-350M_language_n1_layer15_conv_stage1_mla.elf, llima-compile 2026-03-07 05:07:28,761 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token640.sima 2026-03-07 05:07:28,771 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:28,771 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token640" 2026-03-07 05:07:28,774 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:28,774 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:28,775 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:28,776 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:28,776 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:28,776 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:29,587 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:29,796 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:29,807 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:30,079 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:30,129 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:30,140 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:30,141 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:30,147 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:31,595 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:07:31,688 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:31,700 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:32,100 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:32,104 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:32,109 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:32,313 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 413 2026-03-07 05:07:32,313 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:07:34,496 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:35,059 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:35,247 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:35,262 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:35,279 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.555 2026-03-07 05:07:35,279 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:35,281 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:35,331 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:35,331 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:36,315 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:36,998 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:37,073 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:37,277 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:07:37,438 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:37,454 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:37,933 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:37,935 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:37,942 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:38,977 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:07:38,993 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:39,071 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:40,414 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token0_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token0_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token0_stage1_mla.elf, llima-compile 2026-03-07 05:07:40,606 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token768.sima 2026-03-07 05:07:40,617 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:40,617 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token768" 2026-03-07 05:07:40,619 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:40,619 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:40,620 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:40,620 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:40,620 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:40,620 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:42,155 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:42,232 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:43,822 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 382 2026-03-07 05:07:43,823 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:07:44,248 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:44,912 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:45,208 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:45,212 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:45,240 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:45,261 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.391 2026-03-07 05:07:45,261 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:45,267 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:45,295 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:45,355 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:45,355 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:46,229 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:46,939 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:47,228 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:47,230 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:07:47,237 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:47,265 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:47,288 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.323 2026-03-07 05:07:47,288 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:47,299 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:47,385 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:47,385 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:48,636 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:48,863 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:07:49,004 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:49,021 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:49,276 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:49,277 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:49,287 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:49,579 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:49,579 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:49,587 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:07:49,593 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:49,620 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:49,646 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.292 2026-03-07 05:07:49,647 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:49,653 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:49,740 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:49,740 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:50,246 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token128_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token128_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token128_mpk.json, llima-compile 2026-03-07 05:07:50,448 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token896.sima 2026-03-07 05:07:50,460 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:50,460 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token896" 2026-03-07 05:07:50,461 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:50,461 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:50,462 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:50,463 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:50,463 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:50,463 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:51,623 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:51,625 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:52,576 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token256_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token256_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token256_mpk.json, llima-compile 2026-03-07 05:07:52,771 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1024.sima 2026-03-07 05:07:52,789 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:52,789 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1024" 2026-03-07 05:07:52,796 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:52,797 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:52,801 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:52,802 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:52,802 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:52,802 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:52,943 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:07:53,041 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:07:53,153 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:07:53,279 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 384 2026-03-07 05:07:53,279 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:07:53,866 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:07:54,202 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:07:54,233 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:07:54,269 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.269 2026-03-07 05:07:54,269 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:07:54,275 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:07:54,370 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:07:54,370 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:07:54,918 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token384_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token384_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token384_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:07:55,110 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1152.sima 2026-03-07 05:07:55,119 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:07:55,119 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1152" 2026-03-07 05:07:55,121 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:07:55,121 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:07:55,122 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:07:55,122 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:07:55,122 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:07:55,122 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:07:55,472 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 370 2026-03-07 05:07:55,472 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:07:56,392 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:07:56,394 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 2s. 2026-03-07 05:07:58,217 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 528 2026-03-07 05:07:58,217 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:07:58,807 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:07:58,915 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:07:58,932 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:07:59,559 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:07:59,560 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:07:59,567 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:00,106 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:00,106 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:00,106 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:00,106 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:00,107 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token512_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token512_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token512_mpk.json, llima-compile 2026-03-07 05:08:00,298 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1280.sima 2026-03-07 05:08:00,307 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:00,307 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1280" 2026-03-07 05:08:00,309 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:00,309 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:00,310 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:00,310 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:00,311 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:00,311 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:01,122 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:01,852 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:02,220 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:02,252 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:02,286 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.258 2026-03-07 05:08:02,286 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:02,292 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:02,394 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:02,395 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:03,028 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 436 2026-03-07 05:08:03,028 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:03,143 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:03,263 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:03,285 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:03,963 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:03,964 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:03,973 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:04,552 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:04,553 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 2s. 2026-03-07 05:08:04,784 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:04,880 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:05,615 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:05,721 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:05,743 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:06,491 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:06,491 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:06,500 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:08,628 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:08,629 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token640_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token640_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token640_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:08:08,821 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1408.sima 2026-03-07 05:08:08,830 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:08,830 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1408" 2026-03-07 05:08:08,832 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:08,832 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:08,833 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:08,833 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:08,834 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:08,834 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:10,556 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:10,646 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:10,669 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:11,516 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:11,521 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:11,538 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:11,913 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 451 2026-03-07 05:08:11,913 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:13,130 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:13,850 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:14,233 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:14,266 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:14,304 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.246 2026-03-07 05:08:14,304 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:14,318 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:14,420 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:14,420 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:14,763 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:14,855 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:16,626 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:16,628 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 3s. 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:20,721 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token768_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token768_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token768_mpk.json, llima-compile 2026-03-07 05:08:20,908 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:20,913 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1536.sima 2026-03-07 05:08:20,922 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:20,923 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1536" 2026-03-07 05:08:20,924 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:20,924 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:20,925 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:20,925 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:20,926 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:20,926 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:21,128 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:21,157 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:22,044 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:22,045 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:22,056 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:23,271 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:23,423 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:23,433 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:23,658 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 420 2026-03-07 05:08:23,658 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:24,176 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:24,554 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:24,588 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:24,629 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.241 2026-03-07 05:08:24,629 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:24,636 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:24,739 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:24,739 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:26,141 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:26,291 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:26,949 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:26,951 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 3s. 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:31,095 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token896_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token896_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token896_stage1_mla.elf, llima-compile 2026-03-07 05:08:31,289 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1664.sima 2026-03-07 05:08:31,299 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:31,299 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1664" 2026-03-07 05:08:31,300 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:31,300 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:31,302 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:31,302 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:31,302 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:31,302 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:31,403 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:31,552 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:33,225 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:33,424 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:33,454 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:33,638 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 422 2026-03-07 05:08:33,638 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:34,432 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:34,433 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:34,444 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:36,648 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:37,522 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:38,141 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:38,190 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:38,234 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.233 2026-03-07 05:08:38,234 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:38,241 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:38,389 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:38,389 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:39,846 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:40,677 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:41,278 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:41,307 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:41,363 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:41,411 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.23 2026-03-07 05:08:41,411 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:41,417 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:41,456 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:41,576 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:41,582 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 3s. 2026-03-07 05:08:41,590 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:41,590 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:43,315 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:43,506 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:43,537 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:44,575 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:44,576 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:44,606 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:45,117 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:45,323 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:45,325 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:08:45,957 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:46,584 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:46,640 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:46,691 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.225 2026-03-07 05:08:46,691 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:46,707 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:46,880 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:46,880 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:08:47,260 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:47,261 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1024_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token1024_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1024_mpk.json, llima-compile 2026-03-07 05:08:47,454 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1792.sima 2026-03-07 05:08:47,463 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:47,463 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1792" 2026-03-07 05:08:47,464 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:47,464 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:47,465 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:47,466 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:47,466 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:47,466 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:49,779 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 415 2026-03-07 05:08:49,779 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:50,705 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:08:50,707 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 4s. 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:52,114 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:52,115 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1152_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token1152_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1152_stage1_mla.elf, llima-compile 2026-03-07 05:08:52,307 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n128_cache_token1920.sima 2026-03-07 05:08:52,317 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:52,317 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n128_cache_token1920" 2026-03-07 05:08:52,318 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:52,319 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:52,320 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:52,320 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:52,320 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:52,320 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:54,610 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:08:54,673 - mlc.compiler.model_graph.l1_based - INFO - Trigger large tensor support due to: MLA_0/tuple_einsum_1, 573 2026-03-07 05:08:54,673 - mlc.compiler.model_graph.l1_based - INFO - Segmenting layers by spatial/channel dimension 2026-03-07 05:08:54,767 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:08:57,451 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:08:57,816 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1280_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1280_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token1280_stage1_mla.elf, llima-compile 2026-03-07 05:08:58,017 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token127.sima 2026-03-07 05:08:58,025 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:08:58,026 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token127" 2026-03-07 05:08:58,027 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:08:58,027 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:08:58,028 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:08:58,028 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:08:58,028 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:08:58,028 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:08:58,361 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:08:58,587 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:08:58,668 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:58,680 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:58,755 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:08:58,756 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:08:58,761 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:08:58,818 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:08:58,959 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:08:58,990 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:08:59,105 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:08:59,175 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:08:59,230 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.223 2026-03-07 05:08:59,230 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:08:59,236 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:08:59,450 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:08:59,450 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:00,066 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:00,067 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:00,448 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:03,540 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:03,575 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:03,920 - mlc.compiler.model_graph.large_tensor_helper - INFO - Setting tile layouts 2026-03-07 05:09:03,984 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:03,992 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:09:04,056 - mlc.compiler.model_graph.large_tensor_helper - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:04,087 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:05,228 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:05,232 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:05,282 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:06,643 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:06,649 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:06,816 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:07,015 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:07,137 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:07,150 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:07,154 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:07,154 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:07,155 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:07,200 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:07,201 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:07,819 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:07,920 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:08,841 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:09,484 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:09,539 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:09,597 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.219 2026-03-07 05:09:09,597 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:09,603 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:09,768 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:09,768 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:12,166 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:12,167 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1408_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1408_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token1408_mpk.json, llima-compile 2026-03-07 05:09:12,370 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token255.sima 2026-03-07 05:09:12,379 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:12,379 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token255" 2026-03-07 05:09:12,380 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:12,381 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:12,382 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:12,382 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:12,382 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:12,382 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:13,184 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:13,187 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 5s. 2026-03-07 05:09:13,870 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:13,969 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:13,980 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:14,121 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:14,122 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:14,127 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:14,918 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:15,804 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token127_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token127_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token127_mpk.json, llima-compile 2026-03-07 05:09:16,010 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token383.sima 2026-03-07 05:09:16,018 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:16,018 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token383" 2026-03-07 05:09:16,020 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:16,020 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:16,021 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:16,021 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:16,021 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:16,021 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:17,261 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:17,342 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:17,355 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:17,563 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:17,563 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:17,569 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:19,842 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1536_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token1536_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token1536_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:09:20,047 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token511.sima 2026-03-07 05:09:20,056 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:20,056 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token511" 2026-03-07 05:09:20,058 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:20,058 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:20,059 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:20,059 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:20,059 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:20,059 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:20,966 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:21,025 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:21,073 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:21,118 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:21,361 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:21,455 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:21,467 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:21,721 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:21,747 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:21,747 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:21,754 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:22,639 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:23,337 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:23,398 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:23,462 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.218 2026-03-07 05:09:23,462 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:23,468 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:23,648 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:23,648 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:24,556 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:24,600 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:25,039 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:25,493 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:25,677 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:25,695 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:25,698 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:25,698 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:25,698 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:25,755 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:25,755 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:26,909 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:27,174 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:27,342 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:27,423 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:27,430 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 6s. 2026-03-07 05:09:28,713 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:29,197 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:29,214 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:29,270 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:29,361 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:29,377 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:29,379 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:29,380 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:29,380 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:29,433 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:29,433 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:29,915 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:30,542 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:31,845 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token255_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token255_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token255_stage1_mla.elf, llima-compile 2026-03-07 05:09:32,051 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token639.sima 2026-03-07 05:09:32,060 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:32,060 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token639" 2026-03-07 05:09:32,061 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:32,061 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:32,062 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:32,062 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:32,062 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:32,062 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:33,233 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:33,701 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:33,776 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:33,788 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:34,121 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:34,122 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:34,128 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:34,300 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:34,329 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,583 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1664_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token1664_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token1664_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:09:34,788 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token767.sima 2026-03-07 05:09:34,797 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:34,797 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token767" 2026-03-07 05:09:34,798 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:34,798 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:34,799 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:34,800 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:34,800 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:34,800 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:34,844 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:34,905 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token383_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token383_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token383_stage1_mla.elf, llima-compile 2026-03-07 05:09:35,053 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:35,072 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:35,075 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:35,075 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:35,076 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:35,108 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token895.sima 2026-03-07 05:09:35,116 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:35,116 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token895" 2026-03-07 05:09:35,118 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:35,118 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:35,119 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:35,119 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:35,119 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:35,119 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:35,136 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:35,136 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:35,175 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:35,803 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:35,859 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:35,924 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.215 2026-03-07 05:09:35,924 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:35,930 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:36,097 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:36,097 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:36,159 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:36,248 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:36,260 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:36,423 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:36,459 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:36,530 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:36,542 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:36,667 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:36,668 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:36,673 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:37,007 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:37,008 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:37,013 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:37,579 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:39,545 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:39,548 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 6s. 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:39,651 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token511_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token511_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token511_stage1_mla.elf, llima-compile 2026-03-07 05:09:39,855 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1023.sima 2026-03-07 05:09:39,863 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:39,863 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1023" 2026-03-07 05:09:39,864 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:39,865 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:39,866 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:39,866 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:39,866 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:39,866 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:41,307 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:41,355 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:41,658 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:41,737 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:41,749 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:41,897 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:42,307 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:42,315 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:42,321 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:42,838 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:43,534 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:43,595 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:43,664 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.214 2026-03-07 05:09:43,664 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:43,670 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:43,850 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:43,850 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:44,092 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:44,146 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:44,259 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:44,307 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:45,328 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:45,936 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:46,119 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:46,135 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:46,138 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:46,138 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:46,138 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:46,195 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:46,195 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:46,664 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:46,665 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1792_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1792_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n128_cache_token1792_mpk.json, llima-compile 2026-03-07 05:09:46,869 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1151.sima 2026-03-07 05:09:46,878 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:46,878 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1151" 2026-03-07 05:09:46,880 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:46,880 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:46,881 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:46,881 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:46,881 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:46,881 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:47,346 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:47,693 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:47,697 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 7s. 2026-03-07 05:09:48,744 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:48,772 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:48,807 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:48,819 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:49,235 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:49,255 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:49,417 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:49,418 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:49,420 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:49,423 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:49,437 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:49,440 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:49,440 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:49,440 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:49,496 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:49,496 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:49,782 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:49,965 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:50,013 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:50,025 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:50,033 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:50,036 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:50,037 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:50,039 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:50,142 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:50,142 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:50,777 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:50,786 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:51,485 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:53,106 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token639_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token639_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token639_mpk.json, llima-compile 2026-03-07 05:09:53,311 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1279.sima 2026-03-07 05:09:53,320 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:53,320 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1279" 2026-03-07 05:09:53,322 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:53,322 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:53,323 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:53,323 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:53,323 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:53,323 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:53,672 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:53,880 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:09:54,770 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - EV74: 4 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,211 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n128_cache_token1920_mpk.json, models--LiquidAI--LFM2-350M_language_n128_cache_token1920_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n128_cache_token1920_stage1_mla.elf, llima-compile 2026-03-07 05:09:55,348 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:09:55,415 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1407.sima 2026-03-07 05:09:55,424 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:55,424 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1407" 2026-03-07 05:09:55,425 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:55,425 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:55,426 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:55,426 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:55,427 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:55,427 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:55,531 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:55,566 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:09:55,584 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:09:55,587 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:09:55,587 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:09:55,588 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:09:55,606 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:55,618 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:55,650 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:09:55,650 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:55,931 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token767_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token767_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token767_stage1_mla.elf, llima-compile 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:09:56,019 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token895_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token895_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token895_stage1_mla.elf, llima-compile 2026-03-07 05:09:56,135 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1535.sima 2026-03-07 05:09:56,146 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:56,146 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1535" 2026-03-07 05:09:56,148 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:56,148 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:56,149 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:56,149 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:56,149 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:56,149 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:56,220 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1663.sima 2026-03-07 05:09:56,229 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:09:56,229 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1663" 2026-03-07 05:09:56,231 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:09:56,231 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:09:56,232 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:09:56,232 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:09:56,232 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:09:56,232 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:09:56,276 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:56,277 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:56,283 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:56,871 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:56,932 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:56,944 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:56,974 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:09:57,170 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:09:57,219 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:09:57,671 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:57,671 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:57,677 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:57,678 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:57,738 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:57,750 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:57,963 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:09:58,034 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:09:58,047 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:09:58,631 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:58,632 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:58,637 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:58,854 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:09:58,855 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:09:58,861 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:09:59,414 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:01,783 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1023_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token1023_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1023_mpk.json, llima-compile 2026-03-07 05:10:01,820 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:01,992 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1791.sima 2026-03-07 05:10:02,001 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:10:02,001 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1791" 2026-03-07 05:10:02,002 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:10:02,003 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:10:02,003 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:10:02,004 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:10:02,004 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:10:02,004 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:10:02,315 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:02,503 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:02,520 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:02,522 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:02,522 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:02,523 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:02,579 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:02,580 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:03,762 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:03,860 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:10:03,932 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:10:03,944 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:10:04,220 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:04,283 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:04,891 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:10:04,892 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:10:04,897 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:10:05,854 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:05,903 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:06,885 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:06,937 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:07,417 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:07,473 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:09,463 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:10,008 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:10,231 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:10,250 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:10,253 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:10,253 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:10,253 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:10,318 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:10,318 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:10,455 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:10,945 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:11,140 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:11,158 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:11,160 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:11,160 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:11,161 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:11,218 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:11,218 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:11,336 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:11,568 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:11,631 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:12,081 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:12,275 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:12,292 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:12,295 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:12,295 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:12,295 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:12,353 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:12,354 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:12,416 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:12,802 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:13,318 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:13,377 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:13,378 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:13,557 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:13,591 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:13,592 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1151_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token1151_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token1151_stage1_mla.elf, llima-compile 2026-03-07 05:10:13,604 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:13,624 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:13,627 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:13,627 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:13,628 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:13,694 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:13,694 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:13,797 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token1919.sima 2026-03-07 05:10:13,808 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:10:13,808 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token1919" 2026-03-07 05:10:13,810 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:10:13,810 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:10:13,811 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:10:13,811 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:10:13,811 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:10:13,811 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:10:14,254 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:15,047 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:15,257 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:10:15,314 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:10:15,327 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:10:15,900 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:16,308 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:10:16,308 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:10:16,314 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:16,758 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1279_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1279_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token1279_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:10:16,967 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_cache_token2047.sima 2026-03-07 05:10:16,981 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:10:16,981 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_cache_token2047" 2026-03-07 05:10:16,983 - afe.core.compile_networks - INFO - The model is split into 1 segments for MLA and APU 2026-03-07 05:10:16,983 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 1: compiling for MLA 2026-03-07 05:10:16,984 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:10:16,985 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:10:16,985 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:10:16,985 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:10:17,003 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:17,039 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:18,356 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1407_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token1407_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token1407_stage1_mla.elf, llima-compile 2026-03-07 05:10:18,712 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:18,726 - afe.ir.serializer.api - INFO - Loading model from file: CompiledModels/models--LiquidAI--LFM2-350M/sima_files/sdk/models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final.sima 2026-03-07 05:10:18,804 - afe.apis.model - WARNING - A deepcopy is disabled hence model graph will be mutated and no other APIs can be called after the compile step. 2026-03-07 05:10:18,804 - afe.apis.model - INFO - Compiling quantized net "models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final" 2026-03-07 05:10:18,806 - afe.core.compile_networks - INFO - The model is split into 2 segments for MLA and APU 2026-03-07 05:10:18,806 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 1 of 2: compiling for MLA 2026-03-07 05:10:18,836 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties 2026-03-07 05:10:18,837 - mlc.compiler.model_graph.l1_based - INFO - Checking if layers fit into memory without splitting 2026-03-07 05:10:18,837 - mlc.compiler.model_graph.l1_based - INFO - Generating the list of feasible layouts 2026-03-07 05:10:18,837 - mlc.compiler.model_graph.l1_based - INFO - Setting layer parameters 2026-03-07 05:10:18,878 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:10:18,885 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:10:18,899 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:10:18,942 - mlc.compiler.model_graph.l1_based - INFO - Setting tile layouts 2026-03-07 05:10:19,002 - mlc.compiler.model_graph.l1_based - INFO - Allocating memory for IFM/OFM tensors 2026-03-07 05:10:19,015 - mlc.compiler.model_graph.l1_based - INFO - Get model compilation properties done 2026-03-07 05:10:19,117 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:10:19,118 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:10:19,125 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:10:19,247 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:19,488 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:19,508 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:19,511 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:19,511 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:19,511 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:19,575 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:19,575 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,731 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1663_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token1663_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1663_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:19,819 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1535_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1535_mpk.json, models--LiquidAI--LFM2-350M_language_n1_cache_token1535_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:10:20,071 - afe.backends.mla.afe_to_n2a_compiler.n2a_backend_runner - INFO - Start evaluate process to generate check file 2026-03-07 05:10:20,071 - mlc.kernel.layout - INFO - L2 caching mode: L2CachingMode.NONE 2026-03-07 05:10:20,078 - mlc.compiler.model_graph.l1_based - INFO - Generating model code 2026-03-07 05:10:20,941 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:22,673 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:24,672 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:24,721 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:25,553 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:25,553 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:25,553 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:25,554 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1791_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token1791_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1791_mpk.json, llima-compile 2026-03-07 05:10:26,892 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:26,933 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:28,008 - mlc.compiler.model_graph.l1_based - INFO - Generating model code done 2026-03-07 05:10:28,066 - mlc.test_util.test_context - INFO - Scheduling instructions 2026-03-07 05:10:29,300 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:29,795 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:29,988 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:30,006 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:30,008 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:30,008 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:30,009 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:30,068 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:30,068 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:30,457 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:31,264 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:31,440 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:31,654 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:31,692 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:32,947 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:33,247 - mlc.test_util.test_context - INFO - Performing DRAM/L2 synchronization 2026-03-07 05:10:33,839 - mlc.test_util.test_context - INFO - Inserting and merging NOPs 2026-03-07 05:10:34,067 - mlc.test_util.test_context - INFO - Setting IQ sync bits 2026-03-07 05:10:34,087 - mlc.test_util.test_context - INFO - Run compression 2026-03-07 05:10:34,091 - mlc.test_util.test_context - INFO - Compression done in 0s. Compression ratio: 0.404 2026-03-07 05:10:34,091 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:34,091 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:34,157 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:34,157 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:10:35,517 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:35,640 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:35,641 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:35,641 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:35,641 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:35,641 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:35,641 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token1919_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token1919_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token1919_mpk.json, llima-compile 2026-03-07 05:10:37,067 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 0s. 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - EV74: 5 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - A65 : 0 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:10:39,988 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_cache_token2047_stage1_mla_stats.yaml, models--LiquidAI--LFM2-350M_language_n1_cache_token2047_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_cache_token2047_mpk.json, llima-compile 2026-03-07 05:10:40,610 - mlc.test_util.test_context - INFO - Compression done in 8s. Compression ratio: 0.931 2026-03-07 05:10:40,610 - mlc.test_util.test_context - INFO - Re-allocate dram memory 2026-03-07 05:10:40,873 - mlc.test_util.test_context - INFO - Generating metrics 2026-03-07 05:10:40,999 - mlc.test_util.test_context - INFO - Writing report to MLC file 2026-03-07 05:10:41,000 - mlc.test_util.test_context - INFO - Writing instructions to MLC file 2026-03-07 05:11:03,149 - mlc.test_util.test_context - INFO - Code generation done 2026-03-07 05:11:03,149 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Evaluate model and writing chk file done in 1s. 2026-03-07 05:11:04,870 - afe.backends.mla.afe_to_n2a_compiler.n2a_compiler_operations - INFO - Stage 2 of 2: backend is APU 2026-03-07 05:11:04,872 - afe.core.compile_networks - INFO - Stage 2 of 2: compiling for APU 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - ============================== 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - Compilation summary: 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - Desired batch size: 1 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - Achieved batch size: 1 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - Plugin distribution per backend: 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - MLA : 1 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - EV74: 2 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - A65 : 1 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - ------------------------------ 2026-03-07 05:11:11,690 - afe.backends.mpk.interface - INFO - Generated files: models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final_stage1_mla.elf, models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final_mpk.json, models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final_stage2_a65.so, models--LiquidAI--LFM2-350M_language_n1_post_layer15_conv_final_stage1_mla_stats.yaml, llima-compile 2026-03-07 05:11:13,967 - sima_lmm.model.vision_language_model - INFO - FileGenMode.MODEL_SDK_COMPILE files generation completed. Generated all mode=MODEL_SDK_COMPILE files