Experiment: dtufc_elic-featurecoding_falconmamba_individual Log file: output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/dtufc_elic-featurecoding_falconmamba_individual.log DTUFCCodecConfig: arch: elic-featurecoding handler: falconmamba checkpoint: codec_weights/elic_hybrid/elic2022-official_lambda0.01_epochs600_lr0.0001_bs60_patch256-256_checkpoint_best.pth.tar transform_type: kmeans transform_mapping:featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json bit_depth: 8 device: cuda:0 Loading checkpoint: codec_weights/elic_hybrid/elic2022-official_lambda0.01_epochs600_lr0.0001_bs60_patch256-256_checkpoint_best.pth.tar Checkpoint epoch: 506 Loaded elic-featurecoding (1-channel) on cuda:0 Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_0_conv.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.0.conv_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_0_conv.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_0_ssm.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.0.ssm_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_0_ssm.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_1_conv.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.1.conv_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_1_conv.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_1_ssm.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.1.ssm_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_1_ssm.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_2_conv.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.2.conv_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_2_conv.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_2_ssm.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.2.ssm_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_2_ssm.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_3_conv.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.3.conv_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_3_conv.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_3_ssm.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.3.ssm_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_3_ssm.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_4_conv.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.4.conv_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_4_conv.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_4_ssm.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.4.ssm_state' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_layer_4_ssm.json Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_output.json: torch.Size([256]) Loaded per-key quantization points for key 'layer.4.output' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/falconmamba/arc_fewshot-8bit_output.json Loaded per-key mappings: model=falconmamba Keys: ['layer.0.conv_state', 'layer.0.ssm_state', 'layer.1.conv_state', 'layer.1.ssm_state', 'layer.2.conv_state', 'layer.2.ssm_state', 'layer.3.conv_state', 'layer.3.ssm_state', 'layer.4.conv_state', 'layer.4.ssm_state', 'layer.4.output'] ---------------- ------------------------------------------------------------------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding Checkpoint codec_weights/elic_hybrid/elic2022-official_lambda0.01_epochs600_lr0.0001_bs60_patch256-256_checkpoint_best.pth.tar Transform type kmeans Transform config featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json Input ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande Output output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande ---------------- ------------------------------------------------------------------------------------------------------------------- Files found: 100 ---------------------------------------------------------------------- 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample1-layer4-item1.zst (1/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample1-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 109, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 109, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) -> torch.Size([1, 1, 109, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,196B, BPFP=2.1482 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,136B, BPFP=1.2900 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,576B, BPFP=2.2324 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,136B, BPFP=6.1367 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,344B, BPFP=1.8521 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,692B, BPFP=6.0283 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,592B, BPFP=1.6230 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,824B, BPFP=5.5723 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,988B, BPFP=2.0783 ⌛️ [2/4] FRONTEND: Frontend time: 3.160s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.490s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010498 layer.1.conv_state 0.00050886 0.47421008 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00013788 0.10864612 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00007003 0.10008989 layer.4.ssm_state 0.00000003 0.00001171 layer.4.conv_state 0.00023930 0.19914860 layer.4.output 0.00000273 0.00069927 ------------------------------------------------------------------------------------- TOTAL 0.00002952 0.02784782 (elements=1,265,664) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1265664 Total Bytes 383368 BPFP 2.4232 bits/point EBPFP 4.1132 equivalent bits/point MSE 0.027848 ---------------------- -------------------------------------------------------- Time: 5.658s Load: 0.007s, Pack+Encode: 3.160s, Decode+Unpack: 2.490s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 109, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0278 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample1-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample1-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample10-layer4-item1.zst (2/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample10-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,072B, BPFP=2.1406 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,036B, BPFP=1.2839 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,644B, BPFP=2.2366 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,224B, BPFP=6.1582 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,256B, BPFP=1.8467 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,692B, BPFP=6.0283 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,716B, BPFP=1.6306 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,776B, BPFP=5.5605 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,656B, BPFP=2.1699 ⌛️ [2/4] FRONTEND: Frontend time: 2.586s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.508s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010497 layer.1.conv_state 0.00050781 0.48026085 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013544 0.10906950 layer.3.ssm_state 0.00000001 0.00000683 layer.3.conv_state 0.00006839 0.09975932 layer.4.ssm_state 0.00000001 0.00001145 layer.4.conv_state 0.00024756 0.19685073 layer.4.output 0.00000285 0.00075827 ------------------------------------------------------------------------------------- TOTAL 0.00002999 0.02832502 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 383928 BPFP 2.4586 bits/point EBPFP 4.1701 equivalent bits/point MSE 0.028325 ---------------------- -------------------------------------------------------- Time: 5.104s Load: 0.009s, Pack+Encode: 2.586s, Decode+Unpack: 2.508s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample10-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample10-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample100-layer4-item1.zst (3/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample100-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,016B, BPFP=2.1372 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,628B, BPFP=5.5244 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,552B, BPFP=2.2310 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,208B, BPFP=6.1543 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,244B, BPFP=1.8459 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,692B, BPFP=6.0283 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,768B, BPFP=1.6338 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,296B, BPFP=2.2181 ⌛️ [2/4] FRONTEND: Frontend time: 2.604s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.498s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000014 0.00010544 layer.1.conv_state 0.00050623 0.47619668 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013817 0.10961574 layer.3.ssm_state 0.00000001 0.00000705 layer.3.conv_state 0.00007075 0.10047662 layer.4.ssm_state 0.00000001 0.00001115 layer.4.conv_state 0.00027926 0.19506265 layer.4.output 0.00000298 0.00070646 ------------------------------------------------------------------------------------- TOTAL 0.00003162 0.02883237 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 378532 BPFP 2.4809 bits/point EBPFP 4.2324 equivalent bits/point MSE 0.028832 ---------------------- -------------------------------------------------------- Time: 5.111s Load: 0.009s, Pack+Encode: 2.604s, Decode+Unpack: 2.498s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample100-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample100-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample103-layer4-item1.zst (4/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample103-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,188B, BPFP=2.1477 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,072B, BPFP=1.2861 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,636B, BPFP=5.5264 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,712B, BPFP=2.2407 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,216B, BPFP=6.1562 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,292B, BPFP=1.8489 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,704B, BPFP=6.0312 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,684B, BPFP=1.6287 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,792B, BPFP=5.5645 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,832B, BPFP=2.1988 ⌛️ [2/4] FRONTEND: Frontend time: 2.605s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.413s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010582 layer.1.conv_state 0.00051199 0.47832906 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00013382 0.10927589 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00007055 0.10020711 layer.4.ssm_state 0.00000005 0.00001139 layer.4.conv_state 0.00025245 0.19595283 layer.4.output 0.00000298 0.00074087 ------------------------------------------------------------------------------------- TOTAL 0.00003058 0.02853544 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 382400 BPFP 2.4731 bits/point EBPFP 4.2035 equivalent bits/point MSE 0.028535 ---------------------- -------------------------------------------------------- Time: 5.027s Load: 0.009s, Pack+Encode: 2.605s, Decode+Unpack: 2.413s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample103-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample103-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample104-layer4-item1.zst (5/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample104-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,044B, BPFP=2.1389 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,088B, BPFP=1.2871 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,560B, BPFP=5.5078 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,524B, BPFP=2.2292 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,140B, BPFP=6.1377 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,048B, BPFP=1.8340 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,724B, BPFP=6.0361 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,672B, BPFP=1.6279 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,168B, BPFP=2.1649 ⌛️ [2/4] FRONTEND: Frontend time: 2.575s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.488s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010489 layer.1.conv_state 0.00050971 0.47158551 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00014652 0.10880795 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00006897 0.10035864 layer.4.ssm_state 0.00000002 0.00001166 layer.4.conv_state 0.00023369 0.19485638 layer.4.output 0.00000300 0.00072823 ------------------------------------------------------------------------------------- TOTAL 0.00003021 0.02822399 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 381012 BPFP 2.4560 bits/point EBPFP 4.1761 equivalent bits/point MSE 0.028224 ---------------------- -------------------------------------------------------- Time: 5.071s Load: 0.008s, Pack+Encode: 2.575s, Decode+Unpack: 2.488s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample104-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample104-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample105-layer4-item1.zst (6/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample105-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,220B, BPFP=2.1497 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,096B, BPFP=1.2876 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,580B, BPFP=5.5127 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,548B, BPFP=2.2307 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,120B, BPFP=1.8384 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,600B, BPFP=6.0059 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,560B, BPFP=1.6211 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,748B, BPFP=5.5537 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,332B, BPFP=2.1723 ⌛️ [2/4] FRONTEND: Frontend time: 2.618s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.540s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010485 layer.1.conv_state 0.00050648 0.47440299 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00011943 0.10827046 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00007205 0.09958834 layer.4.ssm_state 0.00000003 0.00001168 layer.4.conv_state 0.00022891 0.19622156 layer.4.output 0.00000312 0.00071560 ------------------------------------------------------------------------------------- TOTAL 0.00002958 0.02847886 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 379232 BPFP 2.4608 bits/point EBPFP 4.1926 equivalent bits/point MSE 0.028479 ---------------------- -------------------------------------------------------- Time: 5.167s Load: 0.008s, Pack+Encode: 2.618s, Decode+Unpack: 2.540s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample105-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample105-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample108-layer4-item1.zst (7/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample108-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,208B, BPFP=2.1489 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,064B, BPFP=1.2856 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,652B, BPFP=5.5303 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,176B, BPFP=6.1465 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,168B, BPFP=1.8413 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,668B, BPFP=6.0225 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,768B, BPFP=1.6338 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,756B, BPFP=5.5557 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,280B, BPFP=2.1757 ⌛️ [2/4] FRONTEND: Frontend time: 2.574s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.473s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010511 layer.1.conv_state 0.00050697 0.47510970 layer.2.ssm_state 0.00000001 0.00000664 layer.2.conv_state 0.00013442 0.10915839 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00006950 0.10012201 layer.4.ssm_state 0.00000001 0.00001141 layer.4.conv_state 0.00023016 0.19471332 layer.4.output 0.00000331 0.00070265 ------------------------------------------------------------------------------------- TOTAL 0.00003020 0.02867687 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 377612 BPFP 2.4666 bits/point EBPFP 4.2129 equivalent bits/point MSE 0.028677 ---------------------- -------------------------------------------------------- Time: 5.055s Load: 0.008s, Pack+Encode: 2.574s, Decode+Unpack: 2.473s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample108-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample108-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample110-layer4-item1.zst (8/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample110-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,092B, BPFP=2.1418 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,016B, BPFP=1.2827 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,596B, BPFP=5.5166 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,568B, BPFP=2.2319 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,244B, BPFP=6.1631 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,308B, BPFP=1.8499 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,728B, BPFP=6.0371 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,732B, BPFP=1.6316 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,420B, BPFP=2.1697 ⌛️ [2/4] FRONTEND: Frontend time: 2.602s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.463s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010579 layer.1.conv_state 0.00051140 0.47660035 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00014293 0.10998994 layer.3.ssm_state 0.00000001 0.00000699 layer.3.conv_state 0.00006973 0.10156721 layer.4.ssm_state 0.00000001 0.00001134 layer.4.conv_state 0.00023680 0.19635002 layer.4.output 0.00000289 0.00072995 ------------------------------------------------------------------------------------- TOTAL 0.00003023 0.02845961 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 381784 BPFP 2.4610 bits/point EBPFP 4.1844 equivalent bits/point MSE 0.028460 ---------------------- -------------------------------------------------------- Time: 5.074s Load: 0.008s, Pack+Encode: 2.602s, Decode+Unpack: 2.463s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample110-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample110-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample111-layer4-item1.zst (9/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample111-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,040B, BPFP=2.1387 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,036B, BPFP=1.2839 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,556B, BPFP=2.2312 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,244B, BPFP=6.1631 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,344B, BPFP=1.8521 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,768B, BPFP=6.0469 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,712B, BPFP=1.6304 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,832B, BPFP=5.5742 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,316B, BPFP=2.1677 ⌛️ [2/4] FRONTEND: Frontend time: 2.556s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.438s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000014 0.00010588 layer.1.conv_state 0.00051350 0.47660309 layer.2.ssm_state 0.00000001 0.00000664 layer.2.conv_state 0.00014131 0.11005110 layer.3.ssm_state 0.00000001 0.00000699 layer.3.conv_state 0.00006915 0.10100612 layer.4.ssm_state 0.00000002 0.00001166 layer.4.conv_state 0.00025593 0.19763675 layer.4.output 0.00000302 0.00073759 ------------------------------------------------------------------------------------- TOTAL 0.00003077 0.02848310 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 381752 BPFP 2.4608 bits/point EBPFP 4.1846 equivalent bits/point MSE 0.028483 ---------------------- -------------------------------------------------------- Time: 5.002s Load: 0.009s, Pack+Encode: 2.556s, Decode+Unpack: 2.438s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample111-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample111-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample112-layer4-item1.zst (10/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample112-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,012B, BPFP=2.1370 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,100B, BPFP=1.2878 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,508B, BPFP=2.2283 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,180B, BPFP=6.1475 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,104B, BPFP=1.8374 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,868B, BPFP=1.6399 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,696B, BPFP=5.5410 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,688B, BPFP=2.2287 ⌛️ [2/4] FRONTEND: Frontend time: 2.588s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.492s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000015 0.00010514 layer.1.conv_state 0.00050852 0.47609317 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013698 0.10959301 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00006971 0.10048077 layer.4.ssm_state 0.00000001 0.00001131 layer.4.conv_state 0.00027123 0.19464941 layer.4.output 0.00000318 0.00079120 ------------------------------------------------------------------------------------- TOTAL 0.00003157 0.02894029 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 377696 BPFP 2.4838 bits/point EBPFP 4.2397 equivalent bits/point MSE 0.028940 ---------------------- -------------------------------------------------------- Time: 5.089s Load: 0.009s, Pack+Encode: 2.588s, Decode+Unpack: 2.492s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0289 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample112-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample112-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample113-layer4-item1.zst (11/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample113-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,088B, BPFP=2.1416 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,132B, BPFP=1.2898 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,556B, BPFP=2.2312 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,284B, BPFP=1.8484 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,712B, BPFP=6.0332 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,428B, BPFP=1.6130 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 107,572B, BPFP=2.1660 ⌛️ [2/4] FRONTEND: Frontend time: 2.617s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.484s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010433 layer.1.conv_state 0.00049980 0.47873104 layer.2.ssm_state 0.00000001 0.00000667 layer.2.conv_state 0.00011977 0.10996350 layer.3.ssm_state 0.00000001 0.00000703 layer.3.conv_state 0.00007203 0.10065315 layer.4.ssm_state 0.00000001 0.00001176 layer.4.conv_state 0.00024562 0.19901745 layer.4.output 0.00000308 0.00079152 ------------------------------------------------------------------------------------- TOTAL 0.00003020 0.02914372 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 374616 BPFP 2.4635 bits/point EBPFP 4.2197 equivalent bits/point MSE 0.029144 ---------------------- -------------------------------------------------------- Time: 5.108s Load: 0.007s, Pack+Encode: 2.617s, Decode+Unpack: 2.484s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0291 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample113-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample113-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample117-layer4-item1.zst (12/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample117-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 95, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 95, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) -> torch.Size([1, 1, 95, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,032B, BPFP=2.1382 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,124B, BPFP=1.2893 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,552B, BPFP=2.2310 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,072B, BPFP=1.8354 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,688B, BPFP=6.0273 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,808B, BPFP=1.6362 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,768B, BPFP=5.5586 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 104,616B, BPFP=2.1508 ⌛️ [2/4] FRONTEND: Frontend time: 2.579s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.480s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010572 layer.1.conv_state 0.00051224 0.47732151 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00013107 0.10930482 layer.3.ssm_state 0.00000001 0.00000682 layer.3.conv_state 0.00006945 0.10069373 layer.4.ssm_state 0.00000001 0.00001135 layer.4.conv_state 0.00027531 0.19541477 layer.4.output 0.00000313 0.00070136 ------------------------------------------------------------------------------------- TOTAL 0.00003178 0.02915429 (elements=1,208,320) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1208320 Total Bytes 371728 BPFP 2.4611 bits/point EBPFP 4.2296 equivalent bits/point MSE 0.029154 ---------------------- -------------------------------------------------------- Time: 5.068s Load: 0.009s, Pack+Encode: 2.579s, Decode+Unpack: 2.480s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 95, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0292 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample117-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample117-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample121-layer4-item1.zst (13/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample121-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 94, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 94, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 94, 4096]) -> torch.Size([1, 1, 94, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,972B, BPFP=2.1345 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,020B, BPFP=1.2830 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,500B, BPFP=2.2278 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,248B, BPFP=6.1641 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,832B, BPFP=1.8208 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,708B, BPFP=6.0322 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,516B, BPFP=1.6184 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,748B, BPFP=5.5537 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 102,116B, BPFP=2.1218 ⌛️ [2/4] FRONTEND: Frontend time: 2.578s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 94, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.532s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 94, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000009 0.00010380 layer.1.conv_state 0.00049706 0.47508165 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00015153 0.10871513 layer.3.ssm_state 0.00000001 0.00000675 layer.3.conv_state 0.00007501 0.09983844 layer.4.ssm_state 0.00000002 0.00001142 layer.4.conv_state 0.00026177 0.19757409 layer.4.output 0.00000322 0.00069660 ------------------------------------------------------------------------------------- TOTAL 0.00003183 0.02920784 (elements=1,204,224) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1204224 Total Bytes 368536 BPFP 2.4483 bits/point EBPFP 4.2182 equivalent bits/point MSE 0.029208 ---------------------- -------------------------------------------------------- Time: 5.119s Load: 0.009s, Pack+Encode: 2.578s, Decode+Unpack: 2.532s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 94, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0292 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample121-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample121-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample123-layer4-item1.zst (14/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample123-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,976B, BPFP=2.1348 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,020B, BPFP=1.2830 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,636B, BPFP=5.5264 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,436B, BPFP=2.2239 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,204B, BPFP=6.1533 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,728B, BPFP=1.8145 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,688B, BPFP=6.0273 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,704B, BPFP=1.6299 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,780B, BPFP=5.5615 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,012B, BPFP=2.1682 ⌛️ [2/4] FRONTEND: Frontend time: 2.600s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.488s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010635 layer.1.conv_state 0.00051295 0.47643682 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00015130 0.11004719 layer.3.ssm_state 0.00000001 0.00000668 layer.3.conv_state 0.00007484 0.10083423 layer.4.ssm_state 0.00000001 0.00001137 layer.4.conv_state 0.00021889 0.19621363 layer.4.output 0.00000110 0.00073097 ------------------------------------------------------------------------------------- TOTAL 0.00002983 0.02871125 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 377456 BPFP 2.4574 bits/point EBPFP 4.1921 equivalent bits/point MSE 0.028711 ---------------------- -------------------------------------------------------- Time: 5.095s Load: 0.007s, Pack+Encode: 2.600s, Decode+Unpack: 2.488s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample123-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample123-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample124-layer4-item1.zst (15/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample124-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,048B, BPFP=2.1392 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,048B, BPFP=1.2847 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,596B, BPFP=5.5166 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,504B, BPFP=2.2280 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,200B, BPFP=6.1523 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,240B, BPFP=1.8457 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,668B, BPFP=6.0225 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,796B, BPFP=1.6355 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,736B, BPFP=5.5508 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,244B, BPFP=2.2171 ⌛️ [2/4] FRONTEND: Frontend time: 2.581s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.457s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010485 layer.1.conv_state 0.00051116 0.47742623 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00011994 0.10937984 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00007252 0.09985305 layer.4.ssm_state 0.00000001 0.00001136 layer.4.conv_state 0.00026041 0.19366655 layer.4.output 0.00000302 0.00070810 ------------------------------------------------------------------------------------- TOTAL 0.00003081 0.02880532 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 378352 BPFP 2.4798 bits/point EBPFP 4.2304 equivalent bits/point MSE 0.028805 ---------------------- -------------------------------------------------------- Time: 5.046s Load: 0.008s, Pack+Encode: 2.581s, Decode+Unpack: 2.457s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample124-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample124-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample125-layer4-item1.zst (16/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample125-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,052B, BPFP=2.1394 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,044B, BPFP=1.2844 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,576B, BPFP=5.5117 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,460B, BPFP=2.2253 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,212B, BPFP=6.1553 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,068B, BPFP=1.8352 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,620B, BPFP=6.0107 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,784B, BPFP=1.6348 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,048B, BPFP=2.1689 ⌛️ [2/4] FRONTEND: Frontend time: 2.594s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.515s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010479 layer.1.conv_state 0.00051289 0.47621721 layer.2.ssm_state 0.00000001 0.00000652 layer.2.conv_state 0.00013364 0.10892856 layer.3.ssm_state 0.00000001 0.00000680 layer.3.conv_state 0.00007097 0.10030775 layer.4.ssm_state 0.00000004 0.00001126 layer.4.conv_state 0.00024017 0.19821671 layer.4.output 0.00000292 0.00071641 ------------------------------------------------------------------------------------- TOTAL 0.00003044 0.02870992 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 377908 BPFP 2.4603 bits/point EBPFP 4.1977 equivalent bits/point MSE 0.028710 ---------------------- -------------------------------------------------------- Time: 5.116s Load: 0.007s, Pack+Encode: 2.594s, Decode+Unpack: 2.515s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample125-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample125-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample126-layer4-item1.zst (17/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample126-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,944B, BPFP=2.1328 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,976B, BPFP=1.2803 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,620B, BPFP=5.5225 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,544B, BPFP=2.2305 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,172B, BPFP=6.1455 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,000B, BPFP=1.8311 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,728B, BPFP=1.6313 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,836B, BPFP=5.5752 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,096B, BPFP=2.2341 ⌛️ [2/4] FRONTEND: Frontend time: 2.583s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.434s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010580 layer.1.conv_state 0.00050368 0.47878191 layer.2.ssm_state 0.00000001 0.00000655 layer.2.conv_state 0.00013069 0.11008809 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00007150 0.10086694 layer.4.ssm_state 0.00000001 0.00001128 layer.4.conv_state 0.00023706 0.19627701 layer.4.output 0.00000305 0.00070868 ------------------------------------------------------------------------------------- TOTAL 0.00003027 0.02895829 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 378872 BPFP 2.4832 bits/point EBPFP 4.2316 equivalent bits/point MSE 0.028958 ---------------------- -------------------------------------------------------- Time: 5.025s Load: 0.008s, Pack+Encode: 2.583s, Decode+Unpack: 2.434s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0290 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample126-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample126-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample13-layer4-item1.zst (18/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample13-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,048B, BPFP=2.1392 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,076B, BPFP=1.2864 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,556B, BPFP=2.2312 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,136B, BPFP=6.1367 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,892B, BPFP=1.8245 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,680B, BPFP=6.0254 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,636B, BPFP=1.6257 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,776B, BPFP=5.5605 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,800B, BPFP=2.1579 ⌛️ [2/4] FRONTEND: Frontend time: 2.610s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.527s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010393 layer.1.conv_state 0.00051308 0.47592345 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00013634 0.10888493 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00007240 0.10043973 layer.4.ssm_state 0.00000002 0.00001134 layer.4.conv_state 0.00022022 0.19409581 layer.4.output 0.00000307 0.00072724 ------------------------------------------------------------------------------------- TOTAL 0.00002979 0.02832214 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 380456 BPFP 2.4524 bits/point EBPFP 4.1713 equivalent bits/point MSE 0.028322 ---------------------- -------------------------------------------------------- Time: 5.144s Load: 0.007s, Pack+Encode: 2.610s, Decode+Unpack: 2.527s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample13-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample13-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample131-layer4-item1.zst (19/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample131-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,096B, BPFP=2.1421 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,064B, BPFP=1.2856 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,548B, BPFP=2.2307 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,288B, BPFP=1.8486 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,732B, BPFP=6.0381 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,368B, BPFP=1.6094 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,788B, BPFP=5.5635 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 108,168B, BPFP=2.1780 ⌛️ [2/4] FRONTEND: Frontend time: 2.576s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.486s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010527 layer.1.conv_state 0.00049676 0.47648057 layer.2.ssm_state 0.00000001 0.00000666 layer.2.conv_state 0.00011509 0.10982990 layer.3.ssm_state 0.00000001 0.00000699 layer.3.conv_state 0.00006928 0.10059142 layer.4.ssm_state 0.00000002 0.00001173 layer.4.conv_state 0.00024590 0.20165166 layer.4.output 0.00000308 0.00079510 ------------------------------------------------------------------------------------- TOTAL 0.00002993 0.02915006 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 375076 BPFP 2.4666 bits/point EBPFP 4.2218 equivalent bits/point MSE 0.029150 ---------------------- -------------------------------------------------------- Time: 5.070s Load: 0.008s, Pack+Encode: 2.576s, Decode+Unpack: 2.486s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0292 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample131-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample131-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample137-layer4-item1.zst (20/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample137-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,068B, BPFP=2.1404 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,592B, BPFP=5.5156 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,436B, BPFP=2.2239 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,180B, BPFP=6.1475 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,876B, BPFP=1.8235 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,712B, BPFP=6.0332 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,428B, BPFP=1.6130 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,792B, BPFP=5.5645 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,228B, BPFP=2.2367 ⌛️ [2/4] FRONTEND: Frontend time: 2.570s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.495s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000018 0.00010864 layer.1.conv_state 0.00050366 0.47828144 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00012362 0.10978800 layer.3.ssm_state 0.00000001 0.00000679 layer.3.conv_state 0.00007017 0.10030152 layer.4.ssm_state 0.00000002 0.00001165 layer.4.conv_state 0.00024945 0.19893941 layer.4.output 0.00000327 0.00072139 ------------------------------------------------------------------------------------- TOTAL 0.00003044 0.02899762 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 378636 BPFP 2.4816 bits/point EBPFP 4.2277 equivalent bits/point MSE 0.028998 ---------------------- -------------------------------------------------------- Time: 5.072s Load: 0.008s, Pack+Encode: 2.570s, Decode+Unpack: 2.495s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0290 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample137-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample137-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample139-layer4-item1.zst (21/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample139-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,100B, BPFP=2.1423 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,092B, BPFP=1.2874 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,684B, BPFP=2.2390 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,180B, BPFP=6.1475 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,172B, BPFP=1.8416 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,580B, BPFP=1.6223 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,776B, BPFP=5.5605 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,536B, BPFP=2.2229 ⌛️ [2/4] FRONTEND: Frontend time: 2.605s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.516s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000028 0.00010562 layer.1.conv_state 0.00050590 0.47464001 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00013813 0.10917608 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00006868 0.09998482 layer.4.ssm_state 0.00000003 0.00001154 layer.4.conv_state 0.00022950 0.19771069 layer.4.output 0.00000311 0.00071422 ------------------------------------------------------------------------------------- TOTAL 0.00003028 0.02883926 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 378660 BPFP 2.4818 bits/point EBPFP 4.2325 equivalent bits/point MSE 0.028839 ---------------------- -------------------------------------------------------- Time: 5.129s Load: 0.008s, Pack+Encode: 2.605s, Decode+Unpack: 2.516s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample139-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample139-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample14-layer4-item1.zst (22/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample14-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 106, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) -> torch.Size([1, 1, 106, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,144B, BPFP=2.1450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,056B, BPFP=1.2852 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,596B, BPFP=5.5166 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,564B, BPFP=2.2317 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,116B, BPFP=6.1318 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,020B, BPFP=1.8323 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,688B, BPFP=6.0273 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,772B, BPFP=1.6340 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,704B, BPFP=5.5430 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,344B, BPFP=2.1069 ⌛️ [2/4] FRONTEND: Frontend time: 2.594s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.458s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010355 layer.1.conv_state 0.00051424 0.47362864 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00013151 0.10862603 layer.3.ssm_state 0.00000001 0.00000670 layer.3.conv_state 0.00006895 0.09983163 layer.4.ssm_state 0.00000001 0.00001148 layer.4.conv_state 0.00023185 0.19488294 layer.4.output 0.00000293 0.00066721 ------------------------------------------------------------------------------------- TOTAL 0.00002960 0.02796869 (elements=1,253,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1253376 Total Bytes 381276 BPFP 2.4336 bits/point EBPFP 4.1374 equivalent bits/point MSE 0.027969 ---------------------- -------------------------------------------------------- Time: 5.060s Load: 0.008s, Pack+Encode: 2.594s, Decode+Unpack: 2.458s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0280 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample14-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample14-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample149-layer4-item1.zst (23/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample149-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,032B, BPFP=2.1382 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,012B, BPFP=1.2825 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,476B, BPFP=2.2263 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,176B, BPFP=6.1465 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,984B, BPFP=1.8301 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,760B, BPFP=6.0449 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,764B, BPFP=1.6335 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,816B, BPFP=5.5703 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,924B, BPFP=2.1603 ⌛️ [2/4] FRONTEND: Frontend time: 2.596s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.509s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010527 layer.1.conv_state 0.00051274 0.47724241 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00014729 0.10938990 layer.3.ssm_state 0.00000001 0.00000698 layer.3.conv_state 0.00006977 0.10078689 layer.4.ssm_state 0.00000001 0.00001133 layer.4.conv_state 0.00021673 0.19723809 layer.4.output 0.00000307 0.00074263 ------------------------------------------------------------------------------------- TOTAL 0.00002991 0.02846782 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 380816 BPFP 2.4547 bits/point EBPFP 4.1751 equivalent bits/point MSE 0.028468 ---------------------- -------------------------------------------------------- Time: 5.113s Load: 0.009s, Pack+Encode: 2.596s, Decode+Unpack: 2.509s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample149-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample149-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample15-layer4-item1.zst (24/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample15-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,956B, BPFP=2.1335 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,028B, BPFP=1.2834 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,724B, BPFP=2.2415 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,200B, BPFP=6.1523 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,212B, BPFP=1.8440 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,708B, BPFP=6.0322 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,388B, BPFP=1.6106 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,764B, BPFP=5.5576 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,592B, BPFP=2.1942 ⌛️ [2/4] FRONTEND: Frontend time: 2.620s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.472s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010501 layer.1.conv_state 0.00050701 0.47647634 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00014784 0.10989565 layer.3.ssm_state 0.00000001 0.00000696 layer.3.conv_state 0.00006861 0.10039608 layer.4.ssm_state 0.00000001 0.00001168 layer.4.conv_state 0.00022948 0.20268889 layer.4.output 0.00000294 0.00072128 ------------------------------------------------------------------------------------- TOTAL 0.00003012 0.02867956 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 381476 BPFP 2.4671 bits/point EBPFP 4.1931 equivalent bits/point MSE 0.028680 ---------------------- -------------------------------------------------------- Time: 5.101s Load: 0.008s, Pack+Encode: 2.620s, Decode+Unpack: 2.472s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample15-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample15-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample153-layer4-item1.zst (25/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample153-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,896B, BPFP=2.1299 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,012B, BPFP=1.2825 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,592B, BPFP=5.5156 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,448B, BPFP=2.2246 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,924B, BPFP=1.8264 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,616B, BPFP=6.0098 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,792B, BPFP=1.6353 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,668B, BPFP=2.0971 ⌛️ [2/4] FRONTEND: Frontend time: 2.601s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.436s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010441 layer.1.conv_state 0.00050709 0.47482973 layer.2.ssm_state 0.00000001 0.00000656 layer.2.conv_state 0.00011582 0.10957312 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00007623 0.10022789 layer.4.ssm_state 0.00000002 0.00001113 layer.4.conv_state 0.00025645 0.19405729 layer.4.output 0.00000303 0.00074740 ------------------------------------------------------------------------------------- TOTAL 0.00003006 0.02822103 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 378184 BPFP 2.4297 bits/point EBPFP 4.1420 equivalent bits/point MSE 0.028221 ---------------------- -------------------------------------------------------- Time: 5.044s Load: 0.007s, Pack+Encode: 2.601s, Decode+Unpack: 2.436s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample153-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample153-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample16-layer4-item1.zst (26/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample16-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,996B, BPFP=2.1360 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,064B, BPFP=1.2856 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,596B, BPFP=5.5166 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,672B, BPFP=2.2383 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,156B, BPFP=1.8406 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,708B, BPFP=6.0322 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,256B, BPFP=1.6025 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,788B, BPFP=5.5635 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,664B, BPFP=2.1809 ⌛️ [2/4] FRONTEND: Frontend time: 2.601s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.517s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010531 layer.1.conv_state 0.00050979 0.47544926 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00012394 0.10991919 layer.3.ssm_state 0.00000001 0.00000678 layer.3.conv_state 0.00006896 0.10075259 layer.4.ssm_state 0.00000002 0.00001203 layer.4.conv_state 0.00026177 0.20207569 layer.4.output 0.00000303 0.00073837 ------------------------------------------------------------------------------------- TOTAL 0.00003064 0.02883808 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 378328 BPFP 2.4631 bits/point EBPFP 4.1992 equivalent bits/point MSE 0.028838 ---------------------- -------------------------------------------------------- Time: 5.125s Load: 0.007s, Pack+Encode: 2.601s, Decode+Unpack: 2.517s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample16-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample16-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample168-layer4-item1.zst (27/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample168-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,020B, BPFP=2.1375 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,076B, BPFP=1.2864 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,576B, BPFP=2.2324 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,140B, BPFP=6.1377 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,200B, BPFP=1.8433 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,680B, BPFP=6.0254 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,736B, BPFP=1.6318 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,016B, BPFP=2.2073 ⌛️ [2/4] FRONTEND: Frontend time: 2.599s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.470s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010549 layer.1.conv_state 0.00051051 0.47573584 layer.2.ssm_state 0.00000001 0.00000666 layer.2.conv_state 0.00012244 0.10944516 layer.3.ssm_state 0.00000001 0.00000691 layer.3.conv_state 0.00006890 0.10003639 layer.4.ssm_state 0.00000001 0.00001125 layer.4.conv_state 0.00024994 0.19101651 layer.4.output 0.00000315 0.00070925 ------------------------------------------------------------------------------------- TOTAL 0.00003034 0.02850933 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 380072 BPFP 2.4744 bits/point EBPFP 4.2131 equivalent bits/point MSE 0.028509 ---------------------- -------------------------------------------------------- Time: 5.077s Load: 0.008s, Pack+Encode: 2.599s, Decode+Unpack: 2.470s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample168-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample168-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample179-layer4-item1.zst (28/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample179-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,144B, BPFP=2.1450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,168B, BPFP=1.2920 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,604B, BPFP=2.2341 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,220B, BPFP=6.1572 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,240B, BPFP=1.8457 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,744B, BPFP=6.0410 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,600B, BPFP=1.6235 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,832B, BPFP=5.5742 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,688B, BPFP=2.1147 ⌛️ [2/4] FRONTEND: Frontend time: 2.578s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.490s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010671 layer.1.conv_state 0.00051452 0.47720230 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013443 0.10873527 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00007113 0.09999829 layer.4.ssm_state 0.00000002 0.00001150 layer.4.conv_state 0.00021975 0.19444411 layer.4.output 0.00000261 0.00074670 ------------------------------------------------------------------------------------- TOTAL 0.00002940 0.02817537 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 381112 BPFP 2.4405 bits/point EBPFP 4.1530 equivalent bits/point MSE 0.028175 ---------------------- -------------------------------------------------------- Time: 5.075s Load: 0.007s, Pack+Encode: 2.578s, Decode+Unpack: 2.490s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample179-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample179-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample180-layer4-item1.zst (29/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample180-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,172B, BPFP=2.1467 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,148B, BPFP=1.2908 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,660B, BPFP=2.2375 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,268B, BPFP=1.8474 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,720B, BPFP=6.0352 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,480B, BPFP=1.6162 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,792B, BPFP=2.1167 ⌛️ [2/4] FRONTEND: Frontend time: 2.594s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.531s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010655 layer.1.conv_state 0.00051408 0.47732240 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013299 0.10837606 layer.3.ssm_state 0.00000001 0.00000686 layer.3.conv_state 0.00006990 0.09982824 layer.4.ssm_state 0.00000003 0.00001149 layer.4.conv_state 0.00024394 0.19945559 layer.4.output 0.00000276 0.00075385 ------------------------------------------------------------------------------------- TOTAL 0.00003002 0.02829854 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 381100 BPFP 2.4404 bits/point EBPFP 4.1522 equivalent bits/point MSE 0.028299 ---------------------- -------------------------------------------------------- Time: 5.133s Load: 0.007s, Pack+Encode: 2.594s, Decode+Unpack: 2.531s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample180-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample180-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample183-layer4-item1.zst (30/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample183-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 98, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) -> torch.Size([1, 1, 98, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,944B, BPFP=2.1328 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,040B, BPFP=1.2842 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,620B, BPFP=5.5225 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,588B, BPFP=2.2332 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,112B, BPFP=1.8379 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,732B, BPFP=6.0381 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,940B, BPFP=1.6443 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,752B, BPFP=5.5547 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,496B, BPFP=2.2620 ⌛️ [2/4] FRONTEND: Frontend time: 2.616s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.459s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 98, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000029 0.00010535 layer.1.conv_state 0.00052074 0.47718480 layer.2.ssm_state 0.00000001 0.00000664 layer.2.conv_state 0.00013395 0.11044904 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00007102 0.10084268 layer.4.ssm_state 0.00000002 0.00001110 layer.4.conv_state 0.00024289 0.19706319 layer.4.output 0.00000300 0.00071059 ------------------------------------------------------------------------------------- TOTAL 0.00003095 0.02894613 (elements=1,220,608) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1220608 Total Bytes 380688 BPFP 2.4951 bits/point EBPFP 4.2463 equivalent bits/point MSE 0.028946 ---------------------- -------------------------------------------------------- Time: 5.083s Load: 0.007s, Pack+Encode: 2.616s, Decode+Unpack: 2.459s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 98, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0289 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample183-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample183-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample185-layer4-item1.zst (31/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample185-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,140B, BPFP=2.1448 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,140B, BPFP=1.2903 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,660B, BPFP=5.5322 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,232B, BPFP=6.1602 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,200B, BPFP=1.8433 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,652B, BPFP=6.0186 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,880B, BPFP=1.6406 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,812B, BPFP=5.5693 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 109,880B, BPFP=2.1678 ⌛️ [2/4] FRONTEND: Frontend time: 2.570s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.495s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000416 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010601 layer.1.conv_state 0.00050627 0.47719094 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00013355 0.10959795 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00006878 0.09994800 layer.4.ssm_state 0.00000005 0.00001125 layer.4.conv_state 0.00026388 0.19137442 layer.4.output 0.00000306 0.00069920 ------------------------------------------------------------------------------------- TOTAL 0.00003096 0.02864925 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 377448 BPFP 2.4656 bits/point EBPFP 4.2134 equivalent bits/point MSE 0.028649 ---------------------- -------------------------------------------------------- Time: 5.073s Load: 0.009s, Pack+Encode: 2.570s, Decode+Unpack: 2.495s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0286 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample185-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample185-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample187-layer4-item1.zst (32/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample187-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,060B, BPFP=2.1399 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,116B, BPFP=1.2888 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,628B, BPFP=5.5244 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,968B, BPFP=1.8291 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,720B, BPFP=6.0352 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,488B, BPFP=1.6167 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,768B, BPFP=5.5586 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 109,980B, BPFP=2.1480 ⌛️ [2/4] FRONTEND: Frontend time: 2.605s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.515s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010503 layer.1.conv_state 0.00050939 0.47526190 layer.2.ssm_state 0.00000001 0.00000665 layer.2.conv_state 0.00012181 0.10929237 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00007190 0.10050250 layer.4.ssm_state 0.00000003 0.00001165 layer.4.conv_state 0.00024002 0.19905894 layer.4.output 0.00000295 0.00073652 ------------------------------------------------------------------------------------- TOTAL 0.00003005 0.02872859 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 376748 BPFP 2.4528 bits/point EBPFP 4.1896 equivalent bits/point MSE 0.028729 ---------------------- -------------------------------------------------------- Time: 5.127s Load: 0.007s, Pack+Encode: 2.605s, Decode+Unpack: 2.515s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample187-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample187-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample19-layer4-item1.zst (33/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample19-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,024B, BPFP=2.1377 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,036B, BPFP=1.2839 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,572B, BPFP=5.5107 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,652B, BPFP=2.2371 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,172B, BPFP=6.1455 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,160B, BPFP=1.8408 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,732B, BPFP=6.0381 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,364B, BPFP=1.6091 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,816B, BPFP=5.5703 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,648B, BPFP=2.1806 ⌛️ [2/4] FRONTEND: Frontend time: 2.594s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.425s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010544 layer.1.conv_state 0.00050793 0.47754982 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00012495 0.10967749 layer.3.ssm_state 0.00000001 0.00000678 layer.3.conv_state 0.00006861 0.10096878 layer.4.ssm_state 0.00000002 0.00001187 layer.4.conv_state 0.00029770 0.20017888 layer.4.output 0.00000302 0.00073324 ------------------------------------------------------------------------------------- TOTAL 0.00003157 0.02884113 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 378448 BPFP 2.4639 bits/point EBPFP 4.2008 equivalent bits/point MSE 0.028841 ---------------------- -------------------------------------------------------- Time: 5.029s Load: 0.009s, Pack+Encode: 2.594s, Decode+Unpack: 2.425s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample19-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample19-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample196-layer4-item1.zst (34/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample196-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,228B, BPFP=2.1501 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,092B, BPFP=1.2874 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,636B, BPFP=5.5264 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,456B, BPFP=2.2251 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,088B, BPFP=1.8364 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,628B, BPFP=6.0127 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,884B, BPFP=1.6409 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,052B, BPFP=2.1885 ⌛️ [2/4] FRONTEND: Frontend time: 2.577s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.486s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000029 0.00010566 layer.1.conv_state 0.00050905 0.47343153 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00011183 0.10948650 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00006817 0.10030472 layer.4.ssm_state 0.00000005 0.00001128 layer.4.conv_state 0.00022612 0.19302240 layer.4.output 0.00000307 0.00072677 ------------------------------------------------------------------------------------- TOTAL 0.00002936 0.02851549 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 379300 BPFP 2.4694 bits/point EBPFP 4.2093 equivalent bits/point MSE 0.028515 ---------------------- -------------------------------------------------------- Time: 5.072s Load: 0.008s, Pack+Encode: 2.577s, Decode+Unpack: 2.486s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample196-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample196-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample2-layer4-item1.zst (35/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample2-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,976B, BPFP=2.1348 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,996B, BPFP=1.2815 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,616B, BPFP=5.5215 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,592B, BPFP=2.2334 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,072B, BPFP=1.8354 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,732B, BPFP=6.0381 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,616B, BPFP=1.6245 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,756B, BPFP=5.5557 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,572B, BPFP=2.1684 ⌛️ [2/4] FRONTEND: Frontend time: 2.589s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.455s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010479 layer.1.conv_state 0.00049482 0.47733983 layer.2.ssm_state 0.00000001 0.00000665 layer.2.conv_state 0.00013819 0.10918721 layer.3.ssm_state 0.00000001 0.00000672 layer.3.conv_state 0.00007115 0.10040709 layer.4.ssm_state 0.00000002 0.00001148 layer.4.conv_state 0.00022532 0.19294685 layer.4.output 0.00000291 0.00073627 ------------------------------------------------------------------------------------- TOTAL 0.00002923 0.02815848 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 383360 BPFP 2.4549 bits/point EBPFP 4.1633 equivalent bits/point MSE 0.028158 ---------------------- -------------------------------------------------------- Time: 5.053s Load: 0.009s, Pack+Encode: 2.589s, Decode+Unpack: 2.455s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample2-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample2-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample21-layer4-item1.zst (36/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample21-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,024B, BPFP=2.1377 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,128B, BPFP=1.2896 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,568B, BPFP=2.2319 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,904B, BPFP=1.8252 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,624B, BPFP=6.0117 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,652B, BPFP=1.6267 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,784B, BPFP=5.5625 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,724B, BPFP=2.1170 ⌛️ [2/4] FRONTEND: Frontend time: 2.619s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.518s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010290 layer.1.conv_state 0.00050941 0.47661451 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00014139 0.10898156 layer.3.ssm_state 0.00000001 0.00000674 layer.3.conv_state 0.00007401 0.10016844 layer.4.ssm_state 0.00000002 0.00001145 layer.4.conv_state 0.00025094 0.19647555 layer.4.output 0.00000282 0.00075379 ------------------------------------------------------------------------------------- TOTAL 0.00003050 0.02831656 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 379440 BPFP 2.4378 bits/point EBPFP 4.1514 equivalent bits/point MSE 0.028317 ---------------------- -------------------------------------------------------- Time: 5.145s Load: 0.007s, Pack+Encode: 2.619s, Decode+Unpack: 2.518s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample21-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample21-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample22-layer4-item1.zst (37/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample22-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,216B, BPFP=2.1494 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,076B, BPFP=1.2864 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,640B, BPFP=2.2363 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,124B, BPFP=6.1338 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,932B, BPFP=1.8269 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,728B, BPFP=6.0371 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,468B, BPFP=1.6155 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,720B, BPFP=5.5469 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,676B, BPFP=2.1536 ⌛️ [2/4] FRONTEND: Frontend time: 2.606s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.466s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010467 layer.1.conv_state 0.00050400 0.47456118 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013267 0.10878957 layer.3.ssm_state 0.00000001 0.00000669 layer.3.conv_state 0.00007191 0.09975532 layer.4.ssm_state 0.00000002 0.00001147 layer.4.conv_state 0.00022821 0.19538389 layer.4.output 0.00000291 0.00075272 ------------------------------------------------------------------------------------- TOTAL 0.00002951 0.02821769 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 381452 BPFP 2.4507 bits/point EBPFP 4.1647 equivalent bits/point MSE 0.028218 ---------------------- -------------------------------------------------------- Time: 5.079s Load: 0.007s, Pack+Encode: 2.606s, Decode+Unpack: 2.466s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample22-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample22-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample220-layer4-item1.zst (38/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample220-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,000B, BPFP=2.1362 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,044B, BPFP=1.2844 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,572B, BPFP=2.2322 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,116B, BPFP=1.8381 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,692B, BPFP=6.0283 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,560B, BPFP=1.6211 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,796B, BPFP=5.5654 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,244B, BPFP=2.1512 ⌛️ [2/4] FRONTEND: Frontend time: 2.592s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.494s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010404 layer.1.conv_state 0.00051497 0.47480440 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00014786 0.10883500 layer.3.ssm_state 0.00000001 0.00000694 layer.3.conv_state 0.00006809 0.10072751 layer.4.ssm_state 0.00000002 0.00001146 layer.4.conv_state 0.00024536 0.19549106 layer.4.output 0.00000298 0.00072015 ------------------------------------------------------------------------------------- TOTAL 0.00003085 0.02851682 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 378064 BPFP 2.4532 bits/point EBPFP 4.1845 equivalent bits/point MSE 0.028517 ---------------------- -------------------------------------------------------- Time: 5.094s Load: 0.008s, Pack+Encode: 2.592s, Decode+Unpack: 2.494s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample220-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample220-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample23-layer4-item1.zst (39/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample23-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 109, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 109, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) -> torch.Size([1, 1, 109, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,212B, BPFP=2.1492 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,084B, BPFP=1.2869 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,172B, BPFP=1.8416 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,744B, BPFP=6.0410 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,476B, BPFP=1.6160 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,812B, BPFP=5.5693 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,572B, BPFP=2.0888 ⌛️ [2/4] FRONTEND: Frontend time: 2.600s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.522s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 109, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010658 layer.1.conv_state 0.00052078 0.47763243 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00012449 0.11043929 layer.3.ssm_state 0.00000001 0.00000689 layer.3.conv_state 0.00006985 0.10055713 layer.4.ssm_state 0.00000002 0.00001139 layer.4.conv_state 0.00025944 0.19610077 layer.4.output 0.00000291 0.00070234 ------------------------------------------------------------------------------------- TOTAL 0.00003006 0.02791726 (elements=1,265,664) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1265664 Total Bytes 383740 BPFP 2.4255 bits/point EBPFP 4.1143 equivalent bits/point MSE 0.027917 ---------------------- -------------------------------------------------------- Time: 5.129s Load: 0.007s, Pack+Encode: 2.600s, Decode+Unpack: 2.522s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 109, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0279 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample23-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample23-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample236-layer4-item1.zst (40/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample236-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 96, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 96, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 96, 4096]) -> torch.Size([1, 1, 96, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,108B, BPFP=2.1428 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,028B, BPFP=1.2834 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,604B, BPFP=2.2341 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,172B, BPFP=6.1455 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,192B, BPFP=1.8428 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,668B, BPFP=6.0225 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,496B, BPFP=1.6172 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,800B, BPFP=5.5664 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 109,236B, BPFP=2.2224 ⌛️ [2/4] FRONTEND: Frontend time: 2.593s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 96, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.482s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 96, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010531 layer.1.conv_state 0.00050897 0.47495803 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00013010 0.10952549 layer.3.ssm_state 0.00000001 0.00000694 layer.3.conv_state 0.00007001 0.10100678 layer.4.ssm_state 0.00000002 0.00001153 layer.4.conv_state 0.00022365 0.19749178 layer.4.output 0.00000328 0.00081421 ------------------------------------------------------------------------------------- TOTAL 0.00003024 0.02910144 (elements=1,212,416) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1212416 Total Bytes 376160 BPFP 2.4821 bits/point EBPFP 4.2433 equivalent bits/point MSE 0.029101 ---------------------- -------------------------------------------------------- Time: 5.083s Load: 0.008s, Pack+Encode: 2.593s, Decode+Unpack: 2.482s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 96, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0291 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample236-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample236-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample24-layer4-item1.zst (41/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample24-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,112B, BPFP=2.1431 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,044B, BPFP=1.2844 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,136B, BPFP=6.1367 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,964B, BPFP=1.8289 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,660B, BPFP=6.0205 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,756B, BPFP=1.6331 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,740B, BPFP=5.5518 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,876B, BPFP=2.1441 ⌛️ [2/4] FRONTEND: Frontend time: 2.573s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.456s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010507 layer.1.conv_state 0.00050621 0.47305590 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00012991 0.10887621 layer.3.ssm_state 0.00000001 0.00000660 layer.3.conv_state 0.00007072 0.09985082 layer.4.ssm_state 0.00000001 0.00001134 layer.4.conv_state 0.00026898 0.19267601 layer.4.output 0.00000300 0.00072101 ------------------------------------------------------------------------------------- TOTAL 0.00003084 0.02837367 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 377764 BPFP 2.4512 bits/point EBPFP 4.1830 equivalent bits/point MSE 0.028374 ---------------------- -------------------------------------------------------- Time: 5.036s Load: 0.007s, Pack+Encode: 2.573s, Decode+Unpack: 2.456s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample24-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample24-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample28-layer4-item1.zst (42/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample28-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 106, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) -> torch.Size([1, 1, 106, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,088B, BPFP=2.1416 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,056B, BPFP=1.2852 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,576B, BPFP=5.5117 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,176B, BPFP=6.1465 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,948B, BPFP=1.8279 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,704B, BPFP=6.0312 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,584B, BPFP=1.6226 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,740B, BPFP=5.5518 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,748B, BPFP=2.1143 ⌛️ [2/4] FRONTEND: Frontend time: 2.624s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.427s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010509 layer.1.conv_state 0.00051084 0.47653329 layer.2.ssm_state 0.00000001 0.00000665 layer.2.conv_state 0.00012173 0.10846632 layer.3.ssm_state 0.00000001 0.00000688 layer.3.conv_state 0.00007090 0.09998526 layer.4.ssm_state 0.00000002 0.00001136 layer.4.conv_state 0.00021893 0.19475916 layer.4.output 0.00000300 0.00069820 ------------------------------------------------------------------------------------- TOTAL 0.00002901 0.02805214 (elements=1,253,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1253376 Total Bytes 381472 BPFP 2.4348 bits/point EBPFP 4.1373 equivalent bits/point MSE 0.028052 ---------------------- -------------------------------------------------------- Time: 5.060s Load: 0.009s, Pack+Encode: 2.624s, Decode+Unpack: 2.427s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0281 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample28-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample28-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample29-layer4-item1.zst (43/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample29-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,112B, BPFP=2.1431 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,036B, BPFP=1.2839 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,564B, BPFP=5.5088 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,540B, BPFP=2.2302 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,228B, BPFP=6.1592 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,232B, BPFP=1.8452 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,752B, BPFP=6.0430 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,792B, BPFP=1.6353 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,828B, BPFP=5.5732 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,948B, BPFP=2.2011 ⌛️ [2/4] FRONTEND: Frontend time: 2.575s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.491s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010506 layer.1.conv_state 0.00051310 0.47809649 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00013079 0.10914785 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00006987 0.10102058 layer.4.ssm_state 0.00000001 0.00001147 layer.4.conv_state 0.00025978 0.19382074 layer.4.output 0.00000300 0.00074293 ------------------------------------------------------------------------------------- TOTAL 0.00003069 0.02849157 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 382304 BPFP 2.4725 bits/point EBPFP 4.2015 equivalent bits/point MSE 0.028492 ---------------------- -------------------------------------------------------- Time: 5.075s Load: 0.008s, Pack+Encode: 2.575s, Decode+Unpack: 2.491s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample29-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample29-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample30-layer4-item1.zst (44/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample30-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 108, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 108, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) -> torch.Size([1, 1, 108, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,088B, BPFP=2.1416 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,072B, BPFP=1.2861 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,524B, BPFP=2.2292 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,224B, BPFP=6.1582 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,172B, BPFP=1.8416 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,752B, BPFP=6.0430 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,536B, BPFP=1.6196 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,784B, BPFP=5.5625 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,248B, BPFP=2.0299 ⌛️ [2/4] FRONTEND: Frontend time: 2.543s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.335s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010455 layer.1.conv_state 0.00051348 0.47201830 layer.2.ssm_state 0.00000001 0.00000664 layer.2.conv_state 0.00012963 0.10984187 layer.3.ssm_state 0.00000001 0.00000696 layer.3.conv_state 0.00007448 0.10054972 layer.4.ssm_state 0.00000002 0.00001145 layer.4.conv_state 0.00025769 0.19883816 layer.4.output 0.00000283 0.00068253 ------------------------------------------------------------------------------------- TOTAL 0.00003014 0.02790804 (elements=1,261,568) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1261568 Total Bytes 379280 BPFP 2.4051 bits/point EBPFP 4.0985 equivalent bits/point MSE 0.027908 ---------------------- -------------------------------------------------------- Time: 4.888s Load: 0.009s, Pack+Encode: 2.543s, Decode+Unpack: 2.335s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 108, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0279 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample30-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample30-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample31-layer4-item1.zst (45/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample31-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.010s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 107, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) -> torch.Size([1, 1, 107, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,140B, BPFP=2.1448 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,088B, BPFP=1.2871 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,588B, BPFP=5.5146 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,176B, BPFP=6.1465 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,916B, BPFP=1.8259 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,668B, BPFP=6.0225 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,944B, BPFP=1.6445 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,760B, BPFP=5.5566 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,688B, BPFP=2.0752 ⌛️ [2/4] FRONTEND: Frontend time: 2.576s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.500s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000422 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000025 0.00010410 layer.1.conv_state 0.00050435 0.47391626 layer.2.ssm_state 0.00000001 0.00000665 layer.2.conv_state 0.00013424 0.10822295 layer.3.ssm_state 0.00000001 0.00000668 layer.3.conv_state 0.00007347 0.09997198 layer.4.ssm_state 0.00000005 0.00001127 layer.4.conv_state 0.00029085 0.19149427 layer.4.output 0.00000286 0.00066210 ------------------------------------------------------------------------------------- TOTAL 0.00003097 0.02779036 (elements=1,257,472) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1257472 Total Bytes 380840 BPFP 2.4229 bits/point EBPFP 4.1225 equivalent bits/point MSE 0.027790 ---------------------- -------------------------------------------------------- Time: 5.086s Load: 0.010s, Pack+Encode: 2.576s, Decode+Unpack: 2.500s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0278 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample31-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample31-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample32-layer4-item1.zst (46/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample32-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,156B, BPFP=2.1458 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,988B, BPFP=1.2810 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,076B, BPFP=1.8357 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,800B, BPFP=1.6357 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,816B, BPFP=5.5703 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,576B, BPFP=2.0940 ⌛️ [2/4] FRONTEND: Frontend time: 2.609s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.511s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010558 layer.1.conv_state 0.00051859 0.47465420 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013121 0.10939139 layer.3.ssm_state 0.00000001 0.00000700 layer.3.conv_state 0.00006790 0.10035543 layer.4.ssm_state 0.00000004 0.00001138 layer.4.conv_state 0.00024976 0.19432688 layer.4.output 0.00000299 0.00075688 ------------------------------------------------------------------------------------- TOTAL 0.00003026 0.02813543 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 379740 BPFP 2.4317 bits/point EBPFP 4.1426 equivalent bits/point MSE 0.028135 ---------------------- -------------------------------------------------------- Time: 5.129s Load: 0.009s, Pack+Encode: 2.609s, Decode+Unpack: 2.511s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0281 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample32-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample32-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample33-layer4-item1.zst (47/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample33-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 107, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) -> torch.Size([1, 1, 107, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,160B, BPFP=2.1460 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,572B, BPFP=2.2322 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,220B, BPFP=6.1572 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,076B, BPFP=1.8357 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,592B, BPFP=6.0039 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 27,128B, BPFP=1.6558 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,784B, BPFP=5.5625 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,232B, BPFP=2.0851 ⌛️ [2/4] FRONTEND: Frontend time: 2.615s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.455s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000422 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010429 layer.1.conv_state 0.00050815 0.47471303 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00013662 0.10813171 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00006925 0.09923253 layer.4.ssm_state 0.00000002 0.00001112 layer.4.conv_state 0.00024406 0.18859623 layer.4.output 0.00000278 0.00066566 ------------------------------------------------------------------------------------- TOTAL 0.00002977 0.02771522 (elements=1,257,472) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1257472 Total Bytes 381696 BPFP 2.4283 bits/point EBPFP 4.1299 equivalent bits/point MSE 0.027715 ---------------------- -------------------------------------------------------- Time: 5.078s Load: 0.008s, Pack+Encode: 2.615s, Decode+Unpack: 2.455s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0277 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample33-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample33-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample36-layer4-item1.zst (48/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample36-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,136B, BPFP=2.1445 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,164B, BPFP=1.2917 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,664B, BPFP=5.5332 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,632B, BPFP=2.2358 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,112B, BPFP=6.1309 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,324B, BPFP=1.8508 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,428B, BPFP=1.6130 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,752B, BPFP=5.5547 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,528B, BPFP=2.2122 ⌛️ [2/4] FRONTEND: Frontend time: 3.659s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.426s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010396 layer.1.conv_state 0.00050924 0.47400466 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00010465 0.10889541 layer.3.ssm_state 0.00000001 0.00000698 layer.3.conv_state 0.00006887 0.09936336 layer.4.ssm_state 0.00000002 0.00001157 layer.4.conv_state 0.00021814 0.19644894 layer.4.output 0.00000296 0.00074382 ------------------------------------------------------------------------------------- TOTAL 0.00002875 0.02840243 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 382696 BPFP 2.4750 bits/point EBPFP 4.2029 equivalent bits/point MSE 0.028402 ---------------------- -------------------------------------------------------- Time: 6.094s Load: 0.009s, Pack+Encode: 3.659s, Decode+Unpack: 2.426s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample36-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample36-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample37-layer4-item1.zst (49/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample37-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.010s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 106, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) -> torch.Size([1, 1, 106, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,996B, BPFP=2.1360 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,912B, BPFP=1.2764 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,328B, BPFP=2.2173 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,784B, BPFP=1.8179 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,640B, BPFP=6.0156 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 27,156B, BPFP=1.6575 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,780B, BPFP=5.5615 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,320B, BPFP=2.0880 ⌛️ [2/4] FRONTEND: Frontend time: 2.627s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.471s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010483 layer.1.conv_state 0.00051009 0.47642076 layer.2.ssm_state 0.00000001 0.00000655 layer.2.conv_state 0.00012190 0.10885895 layer.3.ssm_state 0.00000001 0.00000700 layer.3.conv_state 0.00011307 0.09976973 layer.4.ssm_state 0.00000002 0.00001094 layer.4.conv_state 0.00021713 0.18835032 layer.4.output 0.00000285 0.00065496 ------------------------------------------------------------------------------------- TOTAL 0.00003000 0.02787123 (elements=1,253,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1253376 Total Bytes 379964 BPFP 2.4252 bits/point EBPFP 4.1271 equivalent bits/point MSE 0.027871 ---------------------- -------------------------------------------------------- Time: 5.108s Load: 0.010s, Pack+Encode: 2.627s, Decode+Unpack: 2.471s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0279 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample37-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample37-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample38-layer4-item1.zst (50/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample38-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,028B, BPFP=2.1379 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,452B, BPFP=2.2249 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,956B, BPFP=1.8284 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,704B, BPFP=6.0312 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,964B, BPFP=1.6458 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,648B, BPFP=2.1155 ⌛️ [2/4] FRONTEND: Frontend time: 2.627s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.446s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010420 layer.1.conv_state 0.00050753 0.47594124 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00011558 0.10839146 layer.3.ssm_state 0.00000001 0.00000681 layer.3.conv_state 0.00006813 0.10048728 layer.4.ssm_state 0.00000002 0.00001113 layer.4.conv_state 0.00021580 0.19160454 layer.4.output 0.00000303 0.00075472 ------------------------------------------------------------------------------------- TOTAL 0.00002878 0.02816395 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 379672 BPFP 2.4393 bits/point EBPFP 4.1549 equivalent bits/point MSE 0.028164 ---------------------- -------------------------------------------------------- Time: 5.081s Load: 0.008s, Pack+Encode: 2.627s, Decode+Unpack: 2.446s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample38-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample38-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample39-layer4-item1.zst (51/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample39-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,064B, BPFP=2.1401 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,108B, BPFP=1.2883 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,588B, BPFP=5.5146 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,132B, BPFP=6.1357 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,208B, BPFP=1.8438 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,680B, BPFP=6.0254 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,560B, BPFP=1.6211 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,756B, BPFP=5.5557 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,120B, BPFP=2.1450 ⌛️ [2/4] FRONTEND: Frontend time: 2.628s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.507s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010420 layer.1.conv_state 0.00050918 0.47238219 layer.2.ssm_state 0.00000001 0.00000656 layer.2.conv_state 0.00013447 0.10851502 layer.3.ssm_state 0.00000001 0.00000676 layer.3.conv_state 0.00006924 0.09975730 layer.4.ssm_state 0.00000002 0.00001152 layer.4.conv_state 0.00025518 0.19408141 layer.4.output 0.00000276 0.00072456 ------------------------------------------------------------------------------------- TOTAL 0.00003039 0.02819960 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 380068 BPFP 2.4499 bits/point EBPFP 4.1706 equivalent bits/point MSE 0.028200 ---------------------- -------------------------------------------------------- Time: 5.144s Load: 0.008s, Pack+Encode: 2.628s, Decode+Unpack: 2.507s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample39-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample39-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample4-layer4-item1.zst (52/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample4-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 110, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 110, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 110, 4096]) -> torch.Size([1, 1, 110, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,144B, BPFP=2.1450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,112B, BPFP=1.2886 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,624B, BPFP=5.5234 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,612B, BPFP=2.2346 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,216B, BPFP=6.1562 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,108B, BPFP=1.8376 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,716B, BPFP=6.0342 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,508B, BPFP=1.6179 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,792B, BPFP=5.5645 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,156B, BPFP=2.0624 ⌛️ [2/4] FRONTEND: Frontend time: 2.611s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 110, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.556s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 110, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010527 layer.1.conv_state 0.00050842 0.47751656 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00013835 0.10944808 layer.3.ssm_state 0.00000001 0.00000694 layer.3.conv_state 0.00007010 0.10094535 layer.4.ssm_state 0.00000002 0.00001154 layer.4.conv_state 0.00024386 0.19635294 layer.4.output 0.00000300 0.00060399 ------------------------------------------------------------------------------------- TOTAL 0.00002965 0.02778241 (elements=1,269,760) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1269760 Total Bytes 383260 BPFP 2.4147 bits/point EBPFP 4.0976 equivalent bits/point MSE 0.027782 ---------------------- -------------------------------------------------------- Time: 5.176s Load: 0.009s, Pack+Encode: 2.611s, Decode+Unpack: 2.556s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 110, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0278 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample4-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample4-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample40-layer4-item1.zst (53/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample40-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,920B, BPFP=2.1313 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,988B, BPFP=1.2810 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,488B, BPFP=2.2271 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,176B, BPFP=6.1465 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,260B, BPFP=1.8469 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,708B, BPFP=6.0322 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,356B, BPFP=1.6086 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,760B, BPFP=5.5566 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,660B, BPFP=2.2029 ⌛️ [2/4] FRONTEND: Frontend time: 2.574s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.587s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010509 layer.1.conv_state 0.00049945 0.48105139 layer.2.ssm_state 0.00000001 0.00000653 layer.2.conv_state 0.00012969 0.10943842 layer.3.ssm_state 0.00000001 0.00000695 layer.3.conv_state 0.00007002 0.10093351 layer.4.ssm_state 0.00000003 0.00001153 layer.4.conv_state 0.00023573 0.19960023 layer.4.output 0.00000289 0.00071022 ------------------------------------------------------------------------------------- TOTAL 0.00002990 0.02899832 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 378188 BPFP 2.4704 bits/point EBPFP 4.2114 equivalent bits/point MSE 0.028998 ---------------------- -------------------------------------------------------- Time: 5.170s Load: 0.009s, Pack+Encode: 2.574s, Decode+Unpack: 2.587s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0290 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample40-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample40-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample41-layer4-item1.zst (54/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample41-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,112B, BPFP=2.1431 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,592B, BPFP=5.5156 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,156B, BPFP=6.1416 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,364B, BPFP=1.8533 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,712B, BPFP=6.0332 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,316B, BPFP=1.6062 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,160B, BPFP=2.2243 ⌛️ [2/4] FRONTEND: Frontend time: 2.605s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.645s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010584 layer.1.conv_state 0.00051118 0.47673795 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00014206 0.10890390 layer.3.ssm_state 0.00000001 0.00000702 layer.3.conv_state 0.00007124 0.09979978 layer.4.ssm_state 0.00000002 0.00001174 layer.4.conv_state 0.00028525 0.19827779 layer.4.output 0.00000298 0.00075142 ------------------------------------------------------------------------------------- TOTAL 0.00003164 0.02853785 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 383140 BPFP 2.4779 bits/point EBPFP 4.2045 equivalent bits/point MSE 0.028538 ---------------------- -------------------------------------------------------- Time: 5.259s Load: 0.009s, Pack+Encode: 2.605s, Decode+Unpack: 2.645s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample41-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample41-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample43-layer4-item1.zst (55/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample43-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,148B, BPFP=2.1453 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,116B, BPFP=1.2888 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,568B, BPFP=5.5098 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,612B, BPFP=2.2346 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,096B, BPFP=1.8369 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 27,032B, BPFP=1.6499 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,748B, BPFP=5.5537 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,316B, BPFP=2.1937 ⌛️ [2/4] FRONTEND: Frontend time: 2.577s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.624s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000025 0.00010471 layer.1.conv_state 0.00050676 0.47327757 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00012238 0.10871170 layer.3.ssm_state 0.00000001 0.00000691 layer.3.conv_state 0.00007028 0.09994491 layer.4.ssm_state 0.00000002 0.00001123 layer.4.conv_state 0.00023202 0.19213276 layer.4.output 0.00000314 0.00070861 ------------------------------------------------------------------------------------- TOTAL 0.00002981 0.02845124 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 379760 BPFP 2.4724 bits/point EBPFP 4.2136 equivalent bits/point MSE 0.028451 ---------------------- -------------------------------------------------------- Time: 5.209s Load: 0.008s, Pack+Encode: 2.577s, Decode+Unpack: 2.624s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample43-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample43-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample44-layer4-item1.zst (56/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample44-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,004B, BPFP=2.1365 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,060B, BPFP=1.2854 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,500B, BPFP=2.2278 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,088B, BPFP=1.8364 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,612B, BPFP=6.0088 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,728B, BPFP=1.6313 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,748B, BPFP=5.5537 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 109,540B, BPFP=2.2056 ⌛️ [2/4] FRONTEND: Frontend time: 2.565s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.647s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010483 layer.1.conv_state 0.00051299 0.47433922 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00011946 0.10923504 layer.3.ssm_state 0.00000001 0.00000681 layer.3.conv_state 0.00007081 0.09988132 layer.4.ssm_state 0.00000002 0.00001140 layer.4.conv_state 0.00021939 0.19550377 layer.4.output 0.00000316 0.00079568 ------------------------------------------------------------------------------------- TOTAL 0.00002984 0.02889170 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 376352 BPFP 2.4750 bits/point EBPFP 4.2296 equivalent bits/point MSE 0.028892 ---------------------- -------------------------------------------------------- Time: 5.221s Load: 0.009s, Pack+Encode: 2.565s, Decode+Unpack: 2.647s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0289 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample44-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample44-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample45-layer4-item1.zst (57/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample45-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 107, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) -> torch.Size([1, 1, 107, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,124B, BPFP=2.1438 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,032B, BPFP=1.2837 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,756B, BPFP=2.2434 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,128B, BPFP=6.1348 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,392B, BPFP=1.8550 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,740B, BPFP=6.0400 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,264B, BPFP=1.6030 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,840B, BPFP=5.5762 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,920B, BPFP=2.1159 ⌛️ [2/4] FRONTEND: Frontend time: 2.587s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.625s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 107, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000029 0.00010666 layer.1.conv_state 0.00050544 0.47375467 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013757 0.10889392 layer.3.ssm_state 0.00000001 0.00000703 layer.3.conv_state 0.00007295 0.10006671 layer.4.ssm_state 0.00000001 0.00001183 layer.4.conv_state 0.00026139 0.19781226 layer.4.output 0.00000288 0.00070437 ------------------------------------------------------------------------------------- TOTAL 0.00003031 0.02798582 (elements=1,257,472) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1257472 Total Bytes 383052 BPFP 2.4370 bits/point EBPFP 4.1365 equivalent bits/point MSE 0.027986 ---------------------- -------------------------------------------------------- Time: 5.218s Load: 0.007s, Pack+Encode: 2.587s, Decode+Unpack: 2.625s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 107, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0280 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample45-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample45-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample46-layer4-item1.zst (58/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample46-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 106, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) -> torch.Size([1, 1, 106, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,956B, BPFP=2.1335 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,964B, BPFP=1.2795 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,484B, BPFP=2.2268 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,136B, BPFP=6.1367 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,692B, BPFP=1.8123 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,620B, BPFP=6.0107 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,980B, BPFP=1.6467 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,760B, BPFP=5.5566 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,828B, BPFP=2.1158 ⌛️ [2/4] FRONTEND: Frontend time: 2.592s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.505s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010563 layer.1.conv_state 0.00050804 0.47879472 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00011728 0.10894503 layer.3.ssm_state 0.00000001 0.00000674 layer.3.conv_state 0.00007019 0.10046918 layer.4.ssm_state 0.00000002 0.00001112 layer.4.conv_state 0.00023224 0.19353016 layer.4.output 0.00000302 0.00066437 ------------------------------------------------------------------------------------- TOTAL 0.00002915 0.02809259 (elements=1,253,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1253376 Total Bytes 381324 BPFP 2.4339 bits/point EBPFP 4.1349 equivalent bits/point MSE 0.028093 ---------------------- -------------------------------------------------------- Time: 5.105s Load: 0.008s, Pack+Encode: 2.592s, Decode+Unpack: 2.505s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0281 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample46-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample46-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample47-layer4-item1.zst (59/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample47-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,004B, BPFP=2.1365 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,084B, BPFP=1.2869 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,540B, BPFP=2.2302 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,220B, BPFP=6.1572 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,104B, BPFP=1.8374 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,672B, BPFP=6.0234 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,688B, BPFP=1.6289 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,744B, BPFP=5.5527 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 108,800B, BPFP=2.1907 ⌛️ [2/4] FRONTEND: Frontend time: 2.608s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.507s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010495 layer.1.conv_state 0.00051689 0.47727880 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00012124 0.10950226 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00007153 0.10006690 layer.4.ssm_state 0.00000002 0.00001138 layer.4.conv_state 0.00021058 0.19976123 layer.4.output 0.00000315 0.00079187 ------------------------------------------------------------------------------------- TOTAL 0.00002977 0.02909653 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 375740 BPFP 2.4709 bits/point EBPFP 4.2264 equivalent bits/point MSE 0.029097 ---------------------- -------------------------------------------------------- Time: 5.124s Load: 0.008s, Pack+Encode: 2.608s, Decode+Unpack: 2.507s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0291 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample47-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample47-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample49-layer4-item1.zst (60/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample49-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,112B, BPFP=2.1431 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,020B, BPFP=1.2830 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,592B, BPFP=5.5156 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,640B, BPFP=2.2363 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,008B, BPFP=1.8315 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,672B, BPFP=6.0234 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,592B, BPFP=1.6230 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,792B, BPFP=5.5645 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,248B, BPFP=2.1948 ⌛️ [2/4] FRONTEND: Frontend time: 2.637s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.540s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010429 layer.1.conv_state 0.00050257 0.47617993 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00013543 0.10879265 layer.3.ssm_state 0.00000001 0.00000686 layer.3.conv_state 0.00007075 0.10028108 layer.4.ssm_state 0.00000002 0.00001160 layer.4.conv_state 0.00024175 0.19753204 layer.4.output 0.00000314 0.00070399 ------------------------------------------------------------------------------------- TOTAL 0.00003040 0.02877576 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 378108 BPFP 2.4699 bits/point EBPFP 4.2131 equivalent bits/point MSE 0.028776 ---------------------- -------------------------------------------------------- Time: 5.185s Load: 0.008s, Pack+Encode: 2.637s, Decode+Unpack: 2.540s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample49-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample49-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample50-layer4-item1.zst (61/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample50-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,244B, BPFP=2.1511 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,132B, BPFP=1.2898 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,588B, BPFP=5.5146 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,584B, BPFP=2.2329 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,116B, BPFP=6.1318 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,976B, BPFP=1.8296 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,696B, BPFP=6.0293 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,492B, BPFP=1.6169 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,712B, BPFP=2.1774 ⌛️ [2/4] FRONTEND: Frontend time: 2.650s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.525s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010499 layer.1.conv_state 0.00051037 0.47367722 layer.2.ssm_state 0.00000001 0.00000666 layer.2.conv_state 0.00013143 0.10897424 layer.3.ssm_state 0.00000001 0.00000681 layer.3.conv_state 0.00007554 0.09997252 layer.4.ssm_state 0.00000002 0.00001143 layer.4.conv_state 0.00025897 0.19511002 layer.4.output 0.00000311 0.00073851 ------------------------------------------------------------------------------------- TOTAL 0.00003080 0.02837480 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 380620 BPFP 2.4616 bits/point EBPFP 4.1878 equivalent bits/point MSE 0.028375 ---------------------- -------------------------------------------------------- Time: 5.183s Load: 0.007s, Pack+Encode: 2.650s, Decode+Unpack: 2.525s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample50-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample50-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample51-layer4-item1.zst (62/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample51-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,180B, BPFP=2.1472 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,136B, BPFP=1.2900 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,620B, BPFP=2.2351 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,236B, BPFP=6.1611 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,332B, BPFP=1.8513 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,276B, BPFP=1.6038 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,824B, BPFP=5.5723 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,740B, BPFP=2.1343 ⌛️ [2/4] FRONTEND: Frontend time: 2.623s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.535s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010461 layer.1.conv_state 0.00051108 0.47479859 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00013025 0.10983685 layer.3.ssm_state 0.00000001 0.00000700 layer.3.conv_state 0.00006983 0.10077281 layer.4.ssm_state 0.00000002 0.00001191 layer.4.conv_state 0.00023909 0.20398682 layer.4.output 0.00000287 0.00074964 ------------------------------------------------------------------------------------- TOTAL 0.00002976 0.02841269 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 381912 BPFP 2.4456 bits/point EBPFP 4.1565 equivalent bits/point MSE 0.028413 ---------------------- -------------------------------------------------------- Time: 5.165s Load: 0.007s, Pack+Encode: 2.623s, Decode+Unpack: 2.535s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample51-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample51-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample53-layer4-item1.zst (63/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample53-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,116B, BPFP=2.1433 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,164B, BPFP=1.2917 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,560B, BPFP=5.5078 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,444B, BPFP=2.2244 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,100B, BPFP=6.1279 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,988B, BPFP=1.8303 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,644B, BPFP=6.0166 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 27,024B, BPFP=1.6494 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,688B, BPFP=5.5391 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 108,320B, BPFP=2.1811 ⌛️ [2/4] FRONTEND: Frontend time: 2.662s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.362s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010467 layer.1.conv_state 0.00050795 0.47456115 layer.2.ssm_state 0.00000001 0.00000665 layer.2.conv_state 0.00011983 0.10891617 layer.3.ssm_state 0.00000001 0.00000691 layer.3.conv_state 0.00006952 0.10011397 layer.4.ssm_state 0.00000002 0.00001116 layer.4.conv_state 0.00020281 0.19315150 layer.4.output 0.00000316 0.00078637 ------------------------------------------------------------------------------------- TOTAL 0.00002923 0.02882894 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 375320 BPFP 2.4682 bits/point EBPFP 4.2240 equivalent bits/point MSE 0.028829 ---------------------- -------------------------------------------------------- Time: 5.032s Load: 0.008s, Pack+Encode: 2.662s, Decode+Unpack: 2.362s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample53-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample53-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample54-layer4-item1.zst (64/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample54-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,084B, BPFP=2.1414 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,096B, BPFP=1.2876 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,628B, BPFP=5.5244 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,608B, BPFP=2.2344 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,180B, BPFP=1.8420 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,720B, BPFP=1.6309 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,724B, BPFP=5.5479 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,364B, BPFP=2.1087 ⌛️ [2/4] FRONTEND: Frontend time: 2.681s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.413s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010320 layer.1.conv_state 0.00050049 0.47458968 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00013306 0.10871137 layer.3.ssm_state 0.00000001 0.00000689 layer.3.conv_state 0.00007056 0.09960324 layer.4.ssm_state 0.00000001 0.00001134 layer.4.conv_state 0.00023923 0.19725975 layer.4.output 0.00000281 0.00074533 ------------------------------------------------------------------------------------- TOTAL 0.00002958 0.02816885 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 380500 BPFP 2.4366 bits/point EBPFP 4.1473 equivalent bits/point MSE 0.028169 ---------------------- -------------------------------------------------------- Time: 5.101s Load: 0.007s, Pack+Encode: 2.681s, Decode+Unpack: 2.413s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample54-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample54-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample55-layer4-item1.zst (65/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample55-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,104B, BPFP=2.1426 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,144B, BPFP=1.2905 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,636B, BPFP=5.5264 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,688B, BPFP=2.2393 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,144B, BPFP=6.1387 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,340B, BPFP=1.8518 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,716B, BPFP=6.0342 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,524B, BPFP=1.6189 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,800B, BPFP=5.5664 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 109,676B, BPFP=2.2084 ⌛️ [2/4] FRONTEND: Frontend time: 2.643s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.453s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010490 layer.1.conv_state 0.00050183 0.47872916 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00013844 0.10921958 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00006878 0.09961737 layer.4.ssm_state 0.00000002 0.00001162 layer.4.conv_state 0.00025869 0.19863518 layer.4.output 0.00000296 0.00078849 ------------------------------------------------------------------------------------- TOTAL 0.00003099 0.02908447 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 377044 BPFP 2.4795 bits/point EBPFP 4.2378 equivalent bits/point MSE 0.029084 ---------------------- -------------------------------------------------------- Time: 5.103s Load: 0.007s, Pack+Encode: 2.643s, Decode+Unpack: 2.453s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0291 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample55-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample55-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample56-layer4-item1.zst (66/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample56-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,116B, BPFP=2.1433 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,100B, BPFP=1.2878 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,580B, BPFP=5.5127 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,636B, BPFP=2.2361 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,072B, BPFP=1.8354 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,676B, BPFP=6.0244 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,584B, BPFP=1.6226 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,404B, BPFP=2.1297 ⌛️ [2/4] FRONTEND: Frontend time: 2.640s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.456s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010474 layer.1.conv_state 0.00050960 0.47547194 layer.2.ssm_state 0.00000001 0.00000666 layer.2.conv_state 0.00014477 0.10909413 layer.3.ssm_state 0.00000001 0.00000694 layer.3.conv_state 0.00007109 0.10008173 layer.4.ssm_state 0.00000003 0.00001158 layer.4.conv_state 0.00024741 0.19805852 layer.4.output 0.00000297 0.00075597 ------------------------------------------------------------------------------------- TOTAL 0.00003048 0.02832981 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 380440 BPFP 2.4442 bits/point EBPFP 4.1599 equivalent bits/point MSE 0.028330 ---------------------- -------------------------------------------------------- Time: 5.103s Load: 0.007s, Pack+Encode: 2.640s, Decode+Unpack: 2.456s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample56-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample56-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample57-layer4-item1.zst (67/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample57-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 106, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) -> torch.Size([1, 1, 106, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,988B, BPFP=2.1355 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,028B, BPFP=1.2834 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,536B, BPFP=2.2300 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,224B, BPFP=6.1582 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,844B, BPFP=1.8215 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,716B, BPFP=6.0342 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,816B, BPFP=1.6367 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,232B, BPFP=2.1417 ⌛️ [2/4] FRONTEND: Frontend time: 2.609s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.385s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 106, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010466 layer.1.conv_state 0.00051526 0.47554398 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013199 0.10940911 layer.3.ssm_state 0.00000001 0.00000683 layer.3.conv_state 0.00007059 0.09990373 layer.4.ssm_state 0.00000002 0.00001119 layer.4.conv_state 0.00023257 0.19191889 layer.4.output 0.00000278 0.00068921 ------------------------------------------------------------------------------------- TOTAL 0.00002965 0.02797134 (elements=1,253,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1253376 Total Bytes 383068 BPFP 2.4450 bits/point EBPFP 4.1482 equivalent bits/point MSE 0.027971 ---------------------- -------------------------------------------------------- Time: 5.002s Load: 0.008s, Pack+Encode: 2.609s, Decode+Unpack: 2.385s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 106, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0280 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample57-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample57-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample58-layer4-item1.zst (68/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample58-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,132B, BPFP=2.1443 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,060B, BPFP=1.2854 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,592B, BPFP=2.2334 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,200B, BPFP=6.1523 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,244B, BPFP=1.8459 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,704B, BPFP=6.0312 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,656B, BPFP=1.6270 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,832B, BPFP=5.5742 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,288B, BPFP=2.2076 ⌛️ [2/4] FRONTEND: Frontend time: 2.646s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.423s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010539 layer.1.conv_state 0.00051277 0.47757784 layer.2.ssm_state 0.00000001 0.00000656 layer.2.conv_state 0.00013342 0.10871474 layer.3.ssm_state 0.00000001 0.00000699 layer.3.conv_state 0.00006746 0.09995156 layer.4.ssm_state 0.00000002 0.00001134 layer.4.conv_state 0.00024674 0.19555601 layer.4.output 0.00000297 0.00075255 ------------------------------------------------------------------------------------- TOTAL 0.00003035 0.02848730 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 382580 BPFP 2.4743 bits/point EBPFP 4.2029 equivalent bits/point MSE 0.028487 ---------------------- -------------------------------------------------------- Time: 5.077s Load: 0.008s, Pack+Encode: 2.646s, Decode+Unpack: 2.423s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample58-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample58-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample60-layer4-item1.zst (69/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample60-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,988B, BPFP=2.1355 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,100B, BPFP=1.2878 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,644B, BPFP=5.5283 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,524B, BPFP=2.2292 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,184B, BPFP=6.1484 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,012B, BPFP=1.8318 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,628B, BPFP=6.0127 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,908B, BPFP=1.6423 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,724B, BPFP=5.5479 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 107,600B, BPFP=2.1666 ⌛️ [2/4] FRONTEND: Frontend time: 2.588s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.446s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000010 0.00010445 layer.1.conv_state 0.00050499 0.47426671 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013541 0.10861692 layer.3.ssm_state 0.00000001 0.00000682 layer.3.conv_state 0.00007246 0.09987236 layer.4.ssm_state 0.00000002 0.00001134 layer.4.conv_state 0.00022642 0.19165584 layer.4.output 0.00000327 0.00079126 ------------------------------------------------------------------------------------- TOTAL 0.00003032 0.02876772 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 374584 BPFP 2.4633 bits/point EBPFP 4.2191 equivalent bits/point MSE 0.028768 ---------------------- -------------------------------------------------------- Time: 5.041s Load: 0.007s, Pack+Encode: 2.588s, Decode+Unpack: 2.446s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample60-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample60-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample61-layer4-item1.zst (70/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample61-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.010s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,856B, BPFP=2.1274 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,068B, BPFP=1.2859 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,628B, BPFP=5.5244 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,388B, BPFP=2.2209 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,216B, BPFP=6.1562 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,964B, BPFP=1.8289 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,708B, BPFP=6.0322 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,716B, BPFP=1.6306 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,844B, BPFP=5.5771 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,696B, BPFP=2.2011 ⌛️ [2/4] FRONTEND: Frontend time: 2.608s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.586s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000416 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010454 layer.1.conv_state 0.00050447 0.47891253 layer.2.ssm_state 0.00000001 0.00000664 layer.2.conv_state 0.00013123 0.10979327 layer.3.ssm_state 0.00000001 0.00000668 layer.3.conv_state 0.00007682 0.10062510 layer.4.ssm_state 0.00000002 0.00001172 layer.4.conv_state 0.00025075 0.19556929 layer.4.output 0.00000309 0.00070924 ------------------------------------------------------------------------------------- TOTAL 0.00003063 0.02874035 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 379356 BPFP 2.4698 bits/point EBPFP 4.2058 equivalent bits/point MSE 0.028740 ---------------------- -------------------------------------------------------- Time: 5.204s Load: 0.010s, Pack+Encode: 2.608s, Decode+Unpack: 2.586s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample61-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample61-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample62-layer4-item1.zst (71/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample62-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 34,848B, BPFP=2.1270 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,024B, BPFP=1.2832 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,616B, BPFP=5.5215 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,392B, BPFP=2.2212 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,936B, BPFP=1.8271 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,700B, BPFP=6.0303 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,664B, BPFP=1.6274 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,788B, BPFP=5.5635 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,356B, BPFP=2.1945 ⌛️ [2/4] FRONTEND: Frontend time: 2.587s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.579s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010441 layer.1.conv_state 0.00050765 0.47753412 layer.2.ssm_state 0.00000001 0.00000667 layer.2.conv_state 0.00013519 0.11005528 layer.3.ssm_state 0.00000001 0.00000674 layer.3.conv_state 0.00007639 0.10048658 layer.4.ssm_state 0.00000003 0.00001175 layer.4.conv_state 0.00024199 0.19834560 layer.4.output 0.00000307 0.00070953 ------------------------------------------------------------------------------------- TOTAL 0.00003057 0.02878102 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 378784 BPFP 2.4660 bits/point EBPFP 4.2006 equivalent bits/point MSE 0.028781 ---------------------- -------------------------------------------------------- Time: 5.174s Load: 0.008s, Pack+Encode: 2.587s, Decode+Unpack: 2.579s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample62-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample62-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample63-layer4-item1.zst (72/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample63-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,144B, BPFP=2.1450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,024B, BPFP=1.2832 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,524B, BPFP=2.2292 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,196B, BPFP=6.1514 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,048B, BPFP=1.8340 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,624B, BPFP=1.6250 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,912B, BPFP=2.1662 ⌛️ [2/4] FRONTEND: Frontend time: 2.597s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.595s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010431 layer.1.conv_state 0.00051429 0.47724539 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00012088 0.10881591 layer.3.ssm_state 0.00000001 0.00000693 layer.3.conv_state 0.00007035 0.10054588 layer.4.ssm_state 0.00000002 0.00001142 layer.4.conv_state 0.00025261 0.19602683 layer.4.output 0.00000308 0.00072610 ------------------------------------------------------------------------------------- TOTAL 0.00003049 0.02868550 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 377844 BPFP 2.4599 bits/point EBPFP 4.1978 equivalent bits/point MSE 0.028686 ---------------------- -------------------------------------------------------- Time: 5.200s Load: 0.008s, Pack+Encode: 2.597s, Decode+Unpack: 2.595s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample63-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample63-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample65-layer4-item1.zst (73/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample65-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,036B, BPFP=2.1384 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,108B, BPFP=1.2883 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,572B, BPFP=2.2322 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,196B, BPFP=6.1514 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,124B, BPFP=1.8386 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,784B, BPFP=6.0508 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,436B, BPFP=1.6135 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,796B, BPFP=5.5654 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,968B, BPFP=2.2064 ⌛️ [2/4] FRONTEND: Frontend time: 2.573s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.589s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010534 layer.1.conv_state 0.00051142 0.47700849 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00015094 0.10946779 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00006998 0.10090610 layer.4.ssm_state 0.00000002 0.00001171 layer.4.conv_state 0.00021745 0.19783905 layer.4.output 0.00000300 0.00071124 ------------------------------------------------------------------------------------- TOTAL 0.00003024 0.02874968 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 379876 BPFP 2.4732 bits/point EBPFP 4.2108 equivalent bits/point MSE 0.028750 ---------------------- -------------------------------------------------------- Time: 5.170s Load: 0.009s, Pack+Encode: 2.573s, Decode+Unpack: 2.589s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample65-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample65-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample66-layer4-item1.zst (74/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample66-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,076B, BPFP=2.1409 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,996B, BPFP=1.2815 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,688B, BPFP=5.5391 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,736B, BPFP=2.2422 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,228B, BPFP=6.1592 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,204B, BPFP=1.8435 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,676B, BPFP=6.0244 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,624B, BPFP=1.6250 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,748B, BPFP=5.5537 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,364B, BPFP=2.1478 ⌛️ [2/4] FRONTEND: Frontend time: 2.563s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.725s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000026 0.00010438 layer.1.conv_state 0.00050859 0.47535044 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00015031 0.10928194 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00007130 0.10066168 layer.4.ssm_state 0.00000001 0.00001136 layer.4.conv_state 0.00023281 0.19487225 layer.4.output 0.00000304 0.00075698 ------------------------------------------------------------------------------------- TOTAL 0.00003026 0.02826324 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 381612 BPFP 2.4518 bits/point EBPFP 4.1688 equivalent bits/point MSE 0.028263 ---------------------- -------------------------------------------------------- Time: 5.296s Load: 0.008s, Pack+Encode: 2.563s, Decode+Unpack: 2.725s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample66-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample66-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample67-layer4-item1.zst (75/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample67-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,108B, BPFP=2.1428 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,084B, BPFP=1.2869 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,620B, BPFP=5.5225 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,656B, BPFP=2.2373 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,128B, BPFP=1.8389 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,724B, BPFP=6.0361 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,380B, BPFP=1.6101 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,828B, BPFP=5.5732 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,432B, BPFP=2.2295 ⌛️ [2/4] FRONTEND: Frontend time: 2.550s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.611s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000422 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000014 0.00010628 layer.1.conv_state 0.00050679 0.47545412 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013770 0.10871446 layer.3.ssm_state 0.00000001 0.00000680 layer.3.conv_state 0.00006920 0.10059530 layer.4.ssm_state 0.00000002 0.00001164 layer.4.conv_state 0.00024331 0.19790828 layer.4.output 0.00000290 0.00076300 ------------------------------------------------------------------------------------- TOTAL 0.00003022 0.02851404 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 383424 BPFP 2.4797 bits/point EBPFP 4.2064 equivalent bits/point MSE 0.028514 ---------------------- -------------------------------------------------------- Time: 5.170s Load: 0.009s, Pack+Encode: 2.550s, Decode+Unpack: 2.611s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample67-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample67-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample68-layer4-item1.zst (76/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample68-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,088B, BPFP=2.1416 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,000B, BPFP=1.2817 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,488B, BPFP=2.2271 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,196B, BPFP=6.1514 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,956B, BPFP=1.8284 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,656B, BPFP=6.0195 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,520B, BPFP=1.6187 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,796B, BPFP=5.5654 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 116,524B, BPFP=2.2312 ⌛️ [2/4] FRONTEND: Frontend time: 2.557s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.592s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010505 layer.1.conv_state 0.00050838 0.47594306 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00010671 0.10844078 layer.3.ssm_state 0.00000001 0.00000691 layer.3.conv_state 0.00007018 0.09940262 layer.4.ssm_state 0.00000002 0.00001142 layer.4.conv_state 0.00026429 0.19883823 layer.4.output 0.00000311 0.00076172 ------------------------------------------------------------------------------------- TOTAL 0.00003009 0.02851220 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 383104 BPFP 2.4776 bits/point EBPFP 4.2017 equivalent bits/point MSE 0.028512 ---------------------- -------------------------------------------------------- Time: 5.157s Load: 0.008s, Pack+Encode: 2.557s, Decode+Unpack: 2.592s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample68-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample68-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample7-layer4-item1.zst (77/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample7-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.006s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,108B, BPFP=2.1428 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 20,980B, BPFP=1.2805 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,464B, BPFP=2.2256 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,160B, BPFP=6.1426 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,032B, BPFP=1.8330 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,760B, BPFP=6.0449 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,548B, BPFP=1.6204 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,740B, BPFP=5.5518 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,320B, BPFP=2.1094 ⌛️ [2/4] FRONTEND: Frontend time: 2.567s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.496s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010405 layer.1.conv_state 0.00051536 0.47461903 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00012850 0.10932931 layer.3.ssm_state 0.00000001 0.00000672 layer.3.conv_state 0.00007391 0.10014976 layer.4.ssm_state 0.00000003 0.00001168 layer.4.conv_state 0.00023144 0.20093806 layer.4.output 0.00000309 0.00075642 ------------------------------------------------------------------------------------- TOTAL 0.00002990 0.02839118 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 378992 BPFP 2.4349 bits/point EBPFP 4.1482 equivalent bits/point MSE 0.028391 ---------------------- -------------------------------------------------------- Time: 5.070s Load: 0.006s, Pack+Encode: 2.567s, Decode+Unpack: 2.496s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample7-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample7-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample71-layer4-item1.zst (78/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample71-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.012s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,160B, BPFP=2.1460 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,124B, BPFP=1.2893 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,572B, BPFP=5.5107 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,624B, BPFP=2.2354 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,208B, BPFP=6.1543 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,148B, BPFP=1.8401 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,712B, BPFP=6.0332 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,668B, BPFP=1.6277 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,752B, BPFP=5.5547 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,916B, BPFP=2.1601 ⌛️ [2/4] FRONTEND: Frontend time: 2.606s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.538s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010519 layer.1.conv_state 0.00051718 0.47681320 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00013311 0.10888942 layer.3.ssm_state 0.00000001 0.00000675 layer.3.conv_state 0.00006873 0.09957293 layer.4.ssm_state 0.00000004 0.00001148 layer.4.conv_state 0.00022143 0.19673263 layer.4.output 0.00000284 0.00073667 ------------------------------------------------------------------------------------- TOTAL 0.00002967 0.02839583 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 381156 BPFP 2.4569 bits/point EBPFP 4.1795 equivalent bits/point MSE 0.028396 ---------------------- -------------------------------------------------------- Time: 5.156s Load: 0.012s, Pack+Encode: 2.606s, Decode+Unpack: 2.538s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample71-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample71-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample72-layer4-item1.zst (79/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample72-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,124B, BPFP=2.1438 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,100B, BPFP=1.2878 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,472B, BPFP=2.2261 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,212B, BPFP=6.1553 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,968B, BPFP=1.8291 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,740B, BPFP=6.0400 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,732B, BPFP=1.6316 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,104B, BPFP=2.1637 ⌛️ [2/4] FRONTEND: Frontend time: 2.599s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.562s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010507 layer.1.conv_state 0.00050321 0.47510478 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00014768 0.10911473 layer.3.ssm_state 0.00000001 0.00000697 layer.3.conv_state 0.00007730 0.10058928 layer.4.ssm_state 0.00000002 0.00001136 layer.4.conv_state 0.00022845 0.19692069 layer.4.output 0.00000298 0.00072200 ------------------------------------------------------------------------------------- TOTAL 0.00003014 0.02838349 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 381160 BPFP 2.4569 bits/point EBPFP 4.1784 equivalent bits/point MSE 0.028383 ---------------------- -------------------------------------------------------- Time: 5.169s Load: 0.008s, Pack+Encode: 2.599s, Decode+Unpack: 2.562s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample72-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample72-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample74-layer4-item1.zst (80/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample74-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,024B, BPFP=2.1377 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,020B, BPFP=1.2830 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,616B, BPFP=5.5215 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,520B, BPFP=2.2290 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,252B, BPFP=1.8464 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,600B, BPFP=6.0059 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,776B, BPFP=1.6343 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,716B, BPFP=5.5459 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,440B, BPFP=2.2324 ⌛️ [2/4] FRONTEND: Frontend time: 2.596s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.468s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000416 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000011 0.00010553 layer.1.conv_state 0.00050786 0.47449189 layer.2.ssm_state 0.00000001 0.00000653 layer.2.conv_state 0.00013194 0.10931595 layer.3.ssm_state 0.00000001 0.00000688 layer.3.conv_state 0.00006897 0.09914760 layer.4.ssm_state 0.00000002 0.00001110 layer.4.conv_state 0.00028450 0.18944828 layer.4.output 0.00000302 0.00072087 ------------------------------------------------------------------------------------- TOTAL 0.00003131 0.02831904 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 382404 BPFP 2.4813 bits/point EBPFP 4.2136 equivalent bits/point MSE 0.028319 ---------------------- -------------------------------------------------------- Time: 5.073s Load: 0.008s, Pack+Encode: 2.596s, Decode+Unpack: 2.468s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample74-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample74-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample75-layer4-item1.zst (81/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample75-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 104, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) -> torch.Size([1, 1, 104, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,168B, BPFP=2.1465 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,128B, BPFP=1.2896 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,604B, BPFP=5.5186 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,580B, BPFP=2.2327 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,204B, BPFP=6.1533 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,104B, BPFP=1.8374 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,684B, BPFP=6.0264 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,596B, BPFP=1.6233 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,856B, BPFP=2.1007 ⌛️ [2/4] FRONTEND: Frontend time: 2.620s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.361s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 104, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010493 layer.1.conv_state 0.00051907 0.47632533 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013541 0.10882314 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00006970 0.10058044 layer.4.ssm_state 0.00000001 0.00001125 layer.4.conv_state 0.00021955 0.19690746 layer.4.output 0.00000291 0.00077746 ------------------------------------------------------------------------------------- TOTAL 0.00002969 0.02833529 (elements=1,245,184) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1245184 Total Bytes 378968 BPFP 2.4348 bits/point EBPFP 4.1509 equivalent bits/point MSE 0.028335 ---------------------- -------------------------------------------------------- Time: 4.989s Load: 0.008s, Pack+Encode: 2.620s, Decode+Unpack: 2.361s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 104, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample75-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample75-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample78-layer4-item1.zst (82/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample78-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,088B, BPFP=2.1416 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,056B, BPFP=1.2852 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,508B, BPFP=2.2283 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,164B, BPFP=6.1436 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,004B, BPFP=1.8313 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,660B, BPFP=6.0205 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,760B, BPFP=1.6333 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,788B, BPFP=5.5635 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,100B, BPFP=2.1678 ⌛️ [2/4] FRONTEND: Frontend time: 2.654s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.421s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010509 layer.1.conv_state 0.00050645 0.47495565 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00013674 0.10854726 layer.3.ssm_state 0.00000001 0.00000700 layer.3.conv_state 0.00006915 0.10023578 layer.4.ssm_state 0.00000001 0.00001132 layer.4.conv_state 0.00024526 0.19683702 layer.4.output 0.00000306 0.00072743 ------------------------------------------------------------------------------------- TOTAL 0.00003040 0.02853844 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 378984 BPFP 2.4591 bits/point EBPFP 4.1909 equivalent bits/point MSE 0.028538 ---------------------- -------------------------------------------------------- Time: 5.083s Load: 0.007s, Pack+Encode: 2.654s, Decode+Unpack: 2.421s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample78-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample78-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample8-layer4-item1.zst (83/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample8-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,012B, BPFP=2.1370 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,084B, BPFP=1.2869 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,600B, BPFP=5.5176 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,508B, BPFP=2.2283 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,180B, BPFP=6.1475 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,048B, BPFP=1.8340 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,768B, BPFP=6.0469 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,552B, BPFP=1.6206 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,768B, BPFP=5.5586 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,272B, BPFP=2.1442 ⌛️ [2/4] FRONTEND: Frontend time: 2.651s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.421s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010430 layer.1.conv_state 0.00051340 0.47617039 layer.2.ssm_state 0.00000001 0.00000654 layer.2.conv_state 0.00013703 0.10932352 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00007042 0.10070798 layer.4.ssm_state 0.00000002 0.00001137 layer.4.conv_state 0.00024895 0.19659491 layer.4.output 0.00000292 0.00073982 ------------------------------------------------------------------------------------- TOTAL 0.00003030 0.02823612 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 382064 BPFP 2.4466 bits/point EBPFP 4.1551 equivalent bits/point MSE 0.028236 ---------------------- -------------------------------------------------------- Time: 5.081s Load: 0.009s, Pack+Encode: 2.651s, Decode+Unpack: 2.421s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample8-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample8-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample80-layer4-item1.zst (84/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample80-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,068B, BPFP=2.1404 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,072B, BPFP=1.2861 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,552B, BPFP=2.2310 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,164B, BPFP=6.1436 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,108B, BPFP=1.8376 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,636B, BPFP=6.0146 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,820B, BPFP=1.6370 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,764B, BPFP=5.5576 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,856B, BPFP=2.1631 ⌛️ [2/4] FRONTEND: Frontend time: 2.613s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.497s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000010 0.00010389 layer.1.conv_state 0.00050642 0.47474688 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00011949 0.10890221 layer.3.ssm_state 0.00000001 0.00000679 layer.3.conv_state 0.00006817 0.09985096 layer.4.ssm_state 0.00000004 0.00001141 layer.4.conv_state 0.00024581 0.19374759 layer.4.output 0.00000308 0.00072054 ------------------------------------------------------------------------------------- TOTAL 0.00002991 0.02844753 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 378920 BPFP 2.4587 bits/point EBPFP 4.1917 equivalent bits/point MSE 0.028448 ---------------------- -------------------------------------------------------- Time: 5.119s Load: 0.008s, Pack+Encode: 2.613s, Decode+Unpack: 2.497s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample80-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample80-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample81-layer4-item1.zst (85/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample81-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,080B, BPFP=2.1411 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,068B, BPFP=1.2859 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,624B, BPFP=5.5234 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,556B, BPFP=2.2312 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,084B, BPFP=1.8362 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,716B, BPFP=6.0342 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,432B, BPFP=1.6133 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,804B, BPFP=5.5674 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 108,728B, BPFP=2.1450 ⌛️ [2/4] FRONTEND: Frontend time: 2.637s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.527s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010506 layer.1.conv_state 0.00050345 0.47534758 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00011822 0.10958956 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00007003 0.10080340 layer.4.ssm_state 0.00000003 0.00001161 layer.4.conv_state 0.00022051 0.19801868 layer.4.output 0.00000301 0.00069471 ------------------------------------------------------------------------------------- TOTAL 0.00002933 0.02879882 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 375532 BPFP 2.4530 bits/point EBPFP 4.1959 equivalent bits/point MSE 0.028799 ---------------------- -------------------------------------------------------- Time: 5.172s Load: 0.008s, Pack+Encode: 2.637s, Decode+Unpack: 2.527s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0288 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample81-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample81-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample83-layer4-item1.zst (86/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample83-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,136B, BPFP=2.1445 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,016B, BPFP=1.2827 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,580B, BPFP=5.5127 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,488B, BPFP=2.2271 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,092B, BPFP=6.1260 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,888B, BPFP=1.8242 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,704B, BPFP=6.0312 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,812B, BPFP=1.6365 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,984B, BPFP=2.2017 ⌛️ [2/4] FRONTEND: Frontend time: 2.605s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.506s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000421 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010439 layer.1.conv_state 0.00050585 0.47505116 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00012980 0.10916737 layer.3.ssm_state 0.00000001 0.00000670 layer.3.conv_state 0.00006838 0.10015232 layer.4.ssm_state 0.00000001 0.00001127 layer.4.conv_state 0.00027100 0.19456448 layer.4.output 0.00000302 0.00072278 ------------------------------------------------------------------------------------- TOTAL 0.00003074 0.02840122 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 381744 BPFP 2.4689 bits/point EBPFP 4.1941 equivalent bits/point MSE 0.028401 ---------------------- -------------------------------------------------------- Time: 5.119s Load: 0.009s, Pack+Encode: 2.605s, Decode+Unpack: 2.506s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample83-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample83-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample84-layer4-item1.zst (87/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample84-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 102, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) -> torch.Size([1, 1, 102, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,176B, BPFP=2.1470 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,048B, BPFP=1.2847 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,596B, BPFP=5.5166 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,480B, BPFP=2.2266 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,180B, BPFP=6.1475 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 29,872B, BPFP=1.8232 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,616B, BPFP=6.0098 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,744B, BPFP=1.6323 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,768B, BPFP=5.5586 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,004B, BPFP=2.1830 ⌛️ [2/4] FRONTEND: Frontend time: 2.593s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.583s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 102, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000027 0.00010346 layer.1.conv_state 0.00051229 0.47814575 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00012754 0.10866819 layer.3.ssm_state 0.00000001 0.00000678 layer.3.conv_state 0.00007197 0.10037144 layer.4.ssm_state 0.00000004 0.00001138 layer.4.conv_state 0.00024522 0.19313292 layer.4.output 0.00000311 0.00073225 ------------------------------------------------------------------------------------- TOTAL 0.00003031 0.02844096 (elements=1,236,992) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1236992 Total Bytes 380756 BPFP 2.4625 bits/point EBPFP 4.1876 equivalent bits/point MSE 0.028441 ---------------------- -------------------------------------------------------- Time: 5.184s Load: 0.008s, Pack+Encode: 2.593s, Decode+Unpack: 2.583s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 102, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample84-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample84-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample85-layer4-item1.zst (88/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample85-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 108, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 108, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) -> torch.Size([1, 1, 108, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,160B, BPFP=2.1460 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,040B, BPFP=1.2842 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,556B, BPFP=2.2312 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,172B, BPFP=6.1455 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,104B, BPFP=1.8374 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,764B, BPFP=6.0459 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,348B, BPFP=1.6082 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,752B, BPFP=5.5547 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,268B, BPFP=2.0122 ⌛️ [2/4] FRONTEND: Frontend time: 2.619s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.629s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 108, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000422 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010510 layer.1.conv_state 0.00051405 0.47696748 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00011333 0.10889442 layer.3.ssm_state 0.00000001 0.00000694 layer.3.conv_state 0.00007471 0.09974582 layer.4.ssm_state 0.00000002 0.00001161 layer.4.conv_state 0.00029453 0.19616425 layer.4.output 0.00000297 0.00069370 ------------------------------------------------------------------------------------- TOTAL 0.00003074 0.02792564 (elements=1,261,568) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1261568 Total Bytes 378068 BPFP 2.3974 bits/point EBPFP 4.0893 equivalent bits/point MSE 0.027926 ---------------------- -------------------------------------------------------- Time: 5.257s Load: 0.009s, Pack+Encode: 2.619s, Decode+Unpack: 2.629s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 108, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0279 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample85-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample85-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample86-layer4-item1.zst (89/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample86-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,108B, BPFP=2.1428 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,024B, BPFP=1.2832 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,648B, BPFP=5.5293 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,548B, BPFP=2.2307 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,184B, BPFP=6.1484 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,060B, BPFP=1.8347 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,724B, BPFP=6.0361 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,624B, BPFP=1.6250 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,772B, BPFP=5.5596 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 111,128B, BPFP=2.1705 ⌛️ [2/4] FRONTEND: Frontend time: 2.575s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.608s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010419 layer.1.conv_state 0.00051465 0.47382408 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00011702 0.10893988 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00007124 0.10045037 layer.4.ssm_state 0.00000002 0.00001151 layer.4.conv_state 0.00024120 0.19507138 layer.4.output 0.00000314 0.00072783 ------------------------------------------------------------------------------------- TOTAL 0.00003014 0.02857012 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 378092 BPFP 2.4615 bits/point EBPFP 4.1996 equivalent bits/point MSE 0.028570 ---------------------- -------------------------------------------------------- Time: 5.193s Load: 0.009s, Pack+Encode: 2.575s, Decode+Unpack: 2.608s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0286 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample86-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample86-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample87-layer4-item1.zst (90/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample87-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.008s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 101, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) -> torch.Size([1, 1, 101, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,080B, BPFP=2.1411 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,064B, BPFP=1.2856 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,584B, BPFP=5.5137 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,536B, BPFP=2.2300 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,144B, BPFP=6.1387 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,012B, BPFP=1.8318 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,640B, BPFP=6.0156 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,716B, BPFP=1.6306 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 112,312B, BPFP=2.1719 ⌛️ [2/4] FRONTEND: Frontend time: 2.560s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.545s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 101, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000028 0.00010549 layer.1.conv_state 0.00050888 0.47451580 layer.2.ssm_state 0.00000001 0.00000661 layer.2.conv_state 0.00013734 0.10900076 layer.3.ssm_state 0.00000001 0.00000698 layer.3.conv_state 0.00006850 0.10025675 layer.4.ssm_state 0.00000001 0.00001140 layer.4.conv_state 0.00023156 0.19657139 layer.4.output 0.00000306 0.00072721 ------------------------------------------------------------------------------------- TOTAL 0.00003009 0.02853227 (elements=1,232,896) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1232896 Total Bytes 379168 BPFP 2.4603 bits/point EBPFP 4.1919 equivalent bits/point MSE 0.028532 ---------------------- -------------------------------------------------------- Time: 5.113s Load: 0.008s, Pack+Encode: 2.560s, Decode+Unpack: 2.545s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 101, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0285 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample87-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample87-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample88-layer4-item1.zst (91/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample88-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,268B, BPFP=2.1526 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,060B, BPFP=1.2854 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,608B, BPFP=5.5195 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,656B, BPFP=2.2373 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,144B, BPFP=6.1387 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,280B, BPFP=1.8481 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,680B, BPFP=6.0254 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,512B, BPFP=1.6182 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,756B, BPFP=5.5557 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,736B, BPFP=2.1567 ⌛️ [2/4] FRONTEND: Frontend time: 2.586s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.532s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000422 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000010 0.00010430 layer.1.conv_state 0.00050284 0.47410440 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00012269 0.10843477 layer.3.ssm_state 0.00000001 0.00000697 layer.3.conv_state 0.00006872 0.09967317 layer.4.ssm_state 0.00000003 0.00001159 layer.4.conv_state 0.00023958 0.19954029 layer.4.output 0.00000284 0.00072784 ------------------------------------------------------------------------------------- TOTAL 0.00002950 0.02838602 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 380972 BPFP 2.4557 bits/point EBPFP 4.1783 equivalent bits/point MSE 0.028386 ---------------------- -------------------------------------------------------- Time: 5.126s Load: 0.007s, Pack+Encode: 2.586s, Decode+Unpack: 2.532s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0284 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample88-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample88-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample9-layer4-item1.zst (92/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample9-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 105, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) -> torch.Size([1, 1, 105, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,096B, BPFP=2.1421 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,052B, BPFP=1.2849 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,528B, BPFP=2.2295 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,192B, BPFP=6.1504 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,108B, BPFP=1.8376 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,784B, BPFP=6.0508 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,392B, BPFP=1.6108 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,760B, BPFP=5.5566 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 115,444B, BPFP=2.1474 ⌛️ [2/4] FRONTEND: Frontend time: 2.589s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.552s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 105, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010497 layer.1.conv_state 0.00051499 0.47497120 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00014898 0.10962716 layer.3.ssm_state 0.00000001 0.00000690 layer.3.conv_state 0.00006960 0.10083979 layer.4.ssm_state 0.00000003 0.00001150 layer.4.conv_state 0.00022645 0.19779924 layer.4.output 0.00000285 0.00073738 ------------------------------------------------------------------------------------- TOTAL 0.00003001 0.02824693 (elements=1,249,280) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1249280 Total Bytes 382260 BPFP 2.4479 bits/point EBPFP 4.1565 equivalent bits/point MSE 0.028247 ---------------------- -------------------------------------------------------- Time: 5.149s Load: 0.009s, Pack+Encode: 2.589s, Decode+Unpack: 2.552s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 105, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0282 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample9-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample9-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample91-layer4-item1.zst (93/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample91-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,076B, BPFP=2.1409 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,152B, BPFP=1.2910 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,620B, BPFP=5.5225 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,636B, BPFP=2.2361 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,184B, BPFP=6.1484 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,220B, BPFP=1.8445 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,692B, BPFP=6.0283 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,748B, BPFP=1.6326 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,784B, BPFP=5.5625 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,028B, BPFP=2.2154 ⌛️ [2/4] FRONTEND: Frontend time: 2.602s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.521s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010481 layer.1.conv_state 0.00051394 0.47739124 layer.2.ssm_state 0.00000001 0.00000657 layer.2.conv_state 0.00013358 0.10878946 layer.3.ssm_state 0.00000001 0.00000685 layer.3.conv_state 0.00006601 0.10000623 layer.4.ssm_state 0.00000002 0.00001176 layer.4.conv_state 0.00022760 0.19663332 layer.4.output 0.00000310 0.00079369 ------------------------------------------------------------------------------------- TOTAL 0.00003032 0.02899509 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 377412 BPFP 2.4819 bits/point EBPFP 4.2403 equivalent bits/point MSE 0.028995 ---------------------- -------------------------------------------------------- Time: 5.131s Load: 0.007s, Pack+Encode: 2.602s, Decode+Unpack: 2.521s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0290 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample91-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample91-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample92-layer4-item1.zst (94/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample92-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 103, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) -> torch.Size([1, 1, 103, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,056B, BPFP=2.1396 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,068B, BPFP=1.2859 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,548B, BPFP=5.5049 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,512B, BPFP=2.2285 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,092B, BPFP=6.1260 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,052B, BPFP=1.8342 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,700B, BPFP=6.0303 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,672B, BPFP=1.6279 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,752B, BPFP=5.5547 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 114,268B, BPFP=2.1668 ⌛️ [2/4] FRONTEND: Frontend time: 2.660s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.479s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 103, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000416 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010431 layer.1.conv_state 0.00050709 0.47199094 layer.2.ssm_state 0.00000001 0.00000660 layer.2.conv_state 0.00014738 0.10860899 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00006942 0.10045592 layer.4.ssm_state 0.00000002 0.00001166 layer.4.conv_state 0.00022996 0.19612829 layer.4.output 0.00000303 0.00073768 ------------------------------------------------------------------------------------- TOTAL 0.00003009 0.02826874 (elements=1,241,088) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1241088 Total Bytes 380992 BPFP 2.4559 bits/point EBPFP 4.1751 equivalent bits/point MSE 0.028269 ---------------------- -------------------------------------------------------- Time: 5.147s Load: 0.009s, Pack+Encode: 2.660s, Decode+Unpack: 2.479s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 103, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0283 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample92-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample92-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample93-layer4-item1.zst (95/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample93-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.009s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 100, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) -> torch.Size([1, 1, 100, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,168B, BPFP=2.1465 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,076B, BPFP=1.2864 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,592B, BPFP=5.5156 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,668B, BPFP=2.2380 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,196B, BPFP=6.1514 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,120B, BPFP=1.8384 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,648B, BPFP=1.6265 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,808B, BPFP=5.5684 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,528B, BPFP=2.1587 ⌛️ [2/4] FRONTEND: Frontend time: 2.656s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.500s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 100, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000418 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000010 0.00010428 layer.1.conv_state 0.00052177 0.47341949 layer.2.ssm_state 0.00000001 0.00000663 layer.2.conv_state 0.00012015 0.10796776 layer.3.ssm_state 0.00000001 0.00000686 layer.3.conv_state 0.00007075 0.09989545 layer.4.ssm_state 0.00000003 0.00001169 layer.4.conv_state 0.00024699 0.20052463 layer.4.output 0.00000299 0.00072053 ------------------------------------------------------------------------------------- TOTAL 0.00003050 0.02866162 (elements=1,228,800) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1228800 Total Bytes 377740 BPFP 2.4592 bits/point EBPFP 4.1989 equivalent bits/point MSE 0.028662 ---------------------- -------------------------------------------------------- Time: 5.165s Load: 0.009s, Pack+Encode: 2.656s, Decode+Unpack: 2.500s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 100, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample93-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample93-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample94-layer4-item1.zst (96/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample94-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,172B, BPFP=2.1467 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,084B, BPFP=1.2869 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,636B, BPFP=5.5264 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,548B, BPFP=2.2307 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,168B, BPFP=6.1445 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,232B, BPFP=1.8452 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,660B, BPFP=6.0205 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,576B, BPFP=1.6221 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,812B, BPFP=5.5693 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,224B, BPFP=2.1746 ⌛️ [2/4] FRONTEND: Frontend time: 2.652s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.370s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000012 0.00010560 layer.1.conv_state 0.00050618 0.47538364 layer.2.ssm_state 0.00000001 0.00000655 layer.2.conv_state 0.00013466 0.10844054 layer.3.ssm_state 0.00000001 0.00000687 layer.3.conv_state 0.00006821 0.09996234 layer.4.ssm_state 0.00000002 0.00001143 layer.4.conv_state 0.00020718 0.19648841 layer.4.output 0.00000289 0.00069705 ------------------------------------------------------------------------------------- TOTAL 0.00002940 0.02870640 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 377384 BPFP 2.4651 bits/point EBPFP 4.2103 equivalent bits/point MSE 0.028706 ---------------------- -------------------------------------------------------- Time: 5.029s Load: 0.007s, Pack+Encode: 2.652s, Decode+Unpack: 2.370s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample94-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample94-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample95-layer4-item1.zst (97/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample95-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,192B, BPFP=2.1479 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,068B, BPFP=1.2859 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,632B, BPFP=5.5254 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,588B, BPFP=2.2332 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,132B, BPFP=6.1357 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,280B, BPFP=1.8481 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,664B, BPFP=6.0215 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,568B, BPFP=1.6216 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,756B, BPFP=5.5557 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,472B, BPFP=2.1795 ⌛️ [2/4] FRONTEND: Frontend time: 2.652s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.423s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010571 layer.1.conv_state 0.00050742 0.47505838 layer.2.ssm_state 0.00000001 0.00000658 layer.2.conv_state 0.00013289 0.10863405 layer.3.ssm_state 0.00000001 0.00000688 layer.3.conv_state 0.00006852 0.10018249 layer.4.ssm_state 0.00000003 0.00001162 layer.4.conv_state 0.00023682 0.19689573 layer.4.output 0.00000300 0.00070745 ------------------------------------------------------------------------------------- TOTAL 0.00003022 0.02872315 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 377624 BPFP 2.4667 bits/point EBPFP 4.2118 equivalent bits/point MSE 0.028723 ---------------------- -------------------------------------------------------- Time: 5.083s Load: 0.007s, Pack+Encode: 2.652s, Decode+Unpack: 2.423s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample95-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample95-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample96-layer4-item1.zst (98/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample96-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 95, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 95, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) -> torch.Size([1, 1, 95, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,016B, BPFP=2.1372 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,116B, BPFP=1.2888 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,612B, BPFP=5.5205 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,600B, BPFP=2.2339 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,228B, BPFP=6.1592 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,008B, BPFP=1.8315 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,628B, BPFP=6.0127 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,552B, BPFP=1.6206 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,720B, BPFP=5.5469 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 102,452B, BPFP=2.1063 ⌛️ [2/4] FRONTEND: Frontend time: 2.636s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.493s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 95, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000419 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010469 layer.1.conv_state 0.00050036 0.47378621 layer.2.ssm_state 0.00000001 0.00000662 layer.2.conv_state 0.00012267 0.10936940 layer.3.ssm_state 0.00000001 0.00000684 layer.3.conv_state 0.00007107 0.09984116 layer.4.ssm_state 0.00000002 0.00001154 layer.4.conv_state 0.00023296 0.19482628 layer.4.output 0.00000331 0.00070543 ------------------------------------------------------------------------------------- TOTAL 0.00003019 0.02902231 (elements=1,208,320) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1208320 Total Bytes 369204 BPFP 2.4444 bits/point EBPFP 4.2105 equivalent bits/point MSE 0.029022 ---------------------- -------------------------------------------------------- Time: 5.136s Load: 0.007s, Pack+Encode: 2.636s, Decode+Unpack: 2.493s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 95, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0290 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample96-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample96-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample97-layer4-item1.zst (99/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample97-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 99, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) -> torch.Size([1, 1, 99, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,052B, BPFP=2.1394 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,032B, BPFP=1.2837 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,624B, BPFP=5.5234 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,548B, BPFP=2.2307 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,100B, BPFP=1.8372 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,636B, BPFP=6.0146 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,704B, BPFP=1.6299 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,760B, BPFP=5.5566 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 110,056B, BPFP=2.1712 ⌛️ [2/4] FRONTEND: Frontend time: 2.586s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.561s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 99, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000417 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000010 0.00010466 layer.1.conv_state 0.00050553 0.47545546 layer.2.ssm_state 0.00000001 0.00000659 layer.2.conv_state 0.00013994 0.10896035 layer.3.ssm_state 0.00000001 0.00000691 layer.3.conv_state 0.00007011 0.09986722 layer.4.ssm_state 0.00000002 0.00001143 layer.4.conv_state 0.00023247 0.19583732 layer.4.output 0.00000297 0.00069208 ------------------------------------------------------------------------------------- TOTAL 0.00003027 0.02870053 (elements=1,224,704) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1224704 Total Bytes 376972 BPFP 2.4625 bits/point EBPFP 4.2060 equivalent bits/point MSE 0.028701 ---------------------- -------------------------------------------------------- Time: 5.155s Load: 0.007s, Pack+Encode: 2.586s, Decode+Unpack: 2.561s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 99, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0287 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample97-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample97-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample99-layer4-item1.zst (100/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample99-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu ⌛️ [1/4] FRONTEND: Load time: 0.007s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 97, 4096) Data type: torch.float32 ------------------------------------------------------------ [2/4] FRONTEND: Pack + Encode (strategy: individual)... IndividualPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) -> torch.Size([1, 1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) -> torch.Size([1, 1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) -> torch.Size([1, 1, 97, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 35,052B, BPFP=2.1394 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 22,272B, BPFP=5.4375 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 21,076B, BPFP=1.2864 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 22,568B, BPFP=5.5098 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 36,608B, BPFP=2.2344 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 25,188B, BPFP=6.1494 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 30,252B, BPFP=1.8464 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 24,720B, BPFP=6.0352 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 26,252B, BPFP=1.6023 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 22,816B, BPFP=5.5703 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 113,320B, BPFP=2.2817 ⌛️ [2/4] FRONTEND: Frontend time: 2.562s (Pack+Encode) [3/4] BACKEND: Decode + Unpack... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.528s [4/4] METRICS: Computing MSE Breakdown... Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output IndividualUnPacker: layer.0.ssm_state: torch.Size([1, 8192, 16]) layer.0.conv_state: torch.Size([1, 8192, 4]) layer.1.ssm_state: torch.Size([1, 8192, 16]) layer.1.conv_state: torch.Size([1, 8192, 4]) layer.2.ssm_state: torch.Size([1, 8192, 16]) layer.2.conv_state: torch.Size([1, 8192, 4]) layer.3.ssm_state: torch.Size([1, 8192, 16]) layer.3.conv_state: torch.Size([1, 8192, 4]) layer.4.ssm_state: torch.Size([1, 8192, 16]) layer.4.conv_state: torch.Size([1, 8192, 4]) layer.4.output: torch.Size([1, 97, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000420 layer.0.conv_state 0.00014612 0.18346235 layer.1.ssm_state 0.00000013 0.00010499 layer.1.conv_state 0.00050423 0.47560546 layer.2.ssm_state 0.00000001 0.00000654 layer.2.conv_state 0.00013920 0.10949986 layer.3.ssm_state 0.00000001 0.00000692 layer.3.conv_state 0.00007129 0.10036246 layer.4.ssm_state 0.00000002 0.00001170 layer.4.conv_state 0.00024943 0.20237641 layer.4.output 0.00000302 0.00080550 ------------------------------------------------------------------------------------- TOTAL 0.00003091 0.02913429 (elements=1,216,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1216512 Total Bytes 380124 BPFP 2.4998 bits/point EBPFP 4.2543 equivalent bits/point MSE 0.029134 ---------------------- -------------------------------------------------------- Time: 5.097s Load: 0.007s, Pack+Encode: 2.562s, Decode+Unpack: 2.528s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 97, 4096]), dtype=torch.float32, device=cpu key['ssm_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 16]), dtype=torch.bfloat16, device=cpu key['conv_state']: [list] with 5 items item[0]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[1]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[2]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[3]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu item[4]: [Tensor] shape=torch.Size([1, 8192, 4]), dtype=torch.bfloat16, device=cpu 💾 Converting with 0.0291 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_winogrande/sample99-layer4-item1.zst to output-fixed/falconmamba/lambda0.01/elic-featurecoding-8bit-individual/fc_winogrande/sample99-layer4-item1.zst ------------------------ ---------------------------- TOTAL PROCESSING SUMMARY ------------------------ ---------------------------- Total files 100 Avg BPFP 2.4574 bits/point Avg EBPFP 4.1865 equivalent bits/point Avg MSE 0.028508 Avg Time 5.127s ------------------------ ----------------------------