Experiment: dtufc_elic-featurecoding_falconmamba_individual Log file: output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/dtufc_elic-featurecoding_falconmamba_individual.log DTUFCCodecConfig: arch: elic-featurecoding handler: falconmamba checkpoint: codec_weights/elic_hybrid/elic2022-official_lambda0.02_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_lambda0.02_epochs600_lr0.0001_bs60_patch256-256_checkpoint_best.pth.tar Checkpoint epoch: 520 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_lambda0.02_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_arc_challenge Output output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge ---------------- ------------------------------------------------------------------------------------------------------------------- Files found: 100 ---------------------------------------------------------------------- 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample0-layer4-item1.zst (1/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample0-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 250, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 250, 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, 250, 4096]) -> torch.Size([1, 1, 250, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,044B, BPFP=4.7024 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,140B, BPFP=2.5110 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,540B, BPFP=8.9209 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,168B, BPFP=4.8931 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,308B, BPFP=10.5732 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,032B, BPFP=3.6030 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,744B, BPFP=10.4355 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,452B, BPFP=3.2014 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,384B, BPFP=9.1270 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 294,896B, BPFP=2.3039 ⌛️ [2/4] FRONTEND: Frontend time: 2.959s (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, 250, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.404s [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, 250, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004222 layer.1.conv_state 0.00050946 0.22746158 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00015435 0.04506392 layer.3.ssm_state 0.00000001 0.00000261 layer.3.conv_state 0.00007182 0.03851791 layer.4.ssm_state 0.00000001 0.00000449 layer.4.conv_state 0.00023205 0.08730576 layer.4.output 0.00000150 0.00013562 ------------------------------------------------------------------------------------- TOTAL 0.00002065 0.00857408 (elements=1,843,200) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1843200 Total Bytes 801852 BPFP 3.4803 bits/point EBPFP 5.6806 equivalent bits/point MSE 0.008574 ---------------------- -------------------------------------------------------- Time: 5.374s Load: 0.011s, Pack+Encode: 2.959s, Decode+Unpack: 2.404s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 250, 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 0.0086 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample0-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample0-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample1-layer4-item1.zst (2/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample1-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 268, 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: 0.014s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 268, 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, 268, 4096]) -> torch.Size([1, 1, 268, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,676B, BPFP=4.6189 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,256B, BPFP=2.5181 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,504B, BPFP=8.9121 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,300B, BPFP=4.8401 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,380B, BPFP=10.5908 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,956B, BPFP=3.5984 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,632B, BPFP=10.4082 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,724B, BPFP=3.1570 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,272B, BPFP=9.0996 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 344,308B, BPFP=2.5092 ⌛️ [2/4] FRONTEND: Frontend time: 2.817s (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, 268, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.429s [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, 268, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004250 layer.1.conv_state 0.00051130 0.22677864 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014046 0.04462231 layer.3.ssm_state 0.00000001 0.00000274 layer.3.conv_state 0.00006868 0.03822772 layer.4.ssm_state 0.00000001 0.00000467 layer.4.conv_state 0.00024322 0.08707429 layer.4.output 0.00000146 0.00015167 ------------------------------------------------------------------------------------- TOTAL 0.00001982 0.00823062 (elements=1,916,928) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1916928 Total Bytes 848152 BPFP 3.5396 bits/point EBPFP 5.6423 equivalent bits/point MSE 0.008231 ---------------------- -------------------------------------------------------- Time: 5.260s Load: 0.014s, Pack+Encode: 2.817s, Decode+Unpack: 2.429s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 268, 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 0.0082 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample1-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample1-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample100-layer4-item1.zst (3/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample100-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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, 178, 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, 178, 4096]) -> torch.Size([1, 1, 178, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,316B, BPFP=4.5969 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,100B, BPFP=2.5085 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,612B, BPFP=8.9385 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,356B, BPFP=4.8435 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,364B, BPFP=10.5869 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,660B, BPFP=3.6414 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,636B, BPFP=10.4092 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,604B, BPFP=3.2107 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,408B, BPFP=9.1328 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 232,084B, BPFP=2.5466 ⌛️ [2/4] FRONTEND: Frontend time: 2.709s (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, 178, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.312s [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, 178, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004284 layer.1.conv_state 0.00048794 0.22820418 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00011881 0.04421601 layer.3.ssm_state 0.00000001 0.00000279 layer.3.conv_state 0.00007930 0.03765530 layer.4.ssm_state 0.00000001 0.00000445 layer.4.conv_state 0.00021563 0.08186142 layer.4.output 0.00000107 0.00018235 ------------------------------------------------------------------------------------- TOTAL 0.00002269 0.01006777 (elements=1,548,288) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1548288 Total Bytes 737284 BPFP 3.8095 bits/point EBPFP 6.4199 equivalent bits/point MSE 0.010068 ---------------------- -------------------------------------------------------- Time: 5.033s Load: 0.012s, Pack+Encode: 2.709s, Decode+Unpack: 2.312s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample100-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample100-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample101-layer4-item1.zst (4/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample101-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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, 174, 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, 174, 4096]) -> torch.Size([1, 1, 174, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,776B, BPFP=4.6250 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,112B, BPFP=2.5093 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,544B, BPFP=8.9219 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,032B, BPFP=4.8237 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,184B, BPFP=10.5430 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,556B, BPFP=3.6350 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,520B, BPFP=10.3809 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,028B, BPFP=3.2366 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,160B, BPFP=9.0723 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,908B, BPFP=2.4909 ⌛️ [2/4] FRONTEND: Frontend time: 2.590s (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, 174, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 174, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00004113 layer.1.conv_state 0.00049132 0.22904751 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014966 0.04429349 layer.3.ssm_state 0.00000001 0.00000275 layer.3.conv_state 0.00011183 0.03766301 layer.4.ssm_state 0.00000003 0.00000438 layer.4.conv_state 0.00021232 0.08343842 layer.4.output 0.00000219 0.00017227 ------------------------------------------------------------------------------------- TOTAL 0.00002481 0.01022225 (elements=1,531,904) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1531904 Total Bytes 726964 BPFP 3.7964 bits/point EBPFP 6.4339 equivalent bits/point MSE 0.010222 ---------------------- -------------------------------------------------------- Time: 4.918s Load: 0.009s, Pack+Encode: 2.590s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample101-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample101-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample102-layer4-item1.zst (5/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample102-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 162, 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, 162, 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, 162, 4096]) -> torch.Size([1, 1, 162, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,520B, BPFP=4.6704 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,224B, BPFP=2.5161 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 77,804B, BPFP=4.7488 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,336B, BPFP=10.5801 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,328B, BPFP=3.6211 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,696B, BPFP=10.4238 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,060B, BPFP=3.2385 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,440B, BPFP=9.1406 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 211,880B, BPFP=2.5545 ⌛️ [2/4] FRONTEND: Frontend time: 2.568s (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, 162, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.311s [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, 162, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000043 0.00004189 layer.1.conv_state 0.00051180 0.22623889 layer.2.ssm_state 0.00000001 0.00000210 layer.2.conv_state 0.00013717 0.04413344 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00012036 0.03822751 layer.4.ssm_state 0.00000005 0.00000426 layer.4.conv_state 0.00026003 0.08321592 layer.4.output 0.00000236 0.00017364 ------------------------------------------------------------------------------------- TOTAL 0.00002708 0.01049801 (elements=1,482,752) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1482752 Total Bytes 716912 BPFP 3.8680 bits/point EBPFP 6.5928 equivalent bits/point MSE 0.010498 ---------------------- -------------------------------------------------------- Time: 4.888s Load: 0.009s, Pack+Encode: 2.568s, Decode+Unpack: 2.311s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 162, 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 0.0105 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample102-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample102-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample103-layer4-item1.zst (6/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample103-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 169, 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, 169, 4096]) -> torch.Size([1, 1, 169, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,804B, BPFP=4.6267 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,128B, BPFP=2.5103 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,520B, BPFP=8.9160 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,448B, BPFP=4.9102 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,256B, BPFP=10.5605 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,508B, BPFP=3.6321 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,552B, BPFP=10.3887 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,276B, BPFP=3.1907 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,356B, BPFP=9.1201 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,324B, BPFP=2.5116 ⌛️ [2/4] FRONTEND: Frontend time: 2.639s (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, 169, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 169, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000031 0.00004078 layer.1.conv_state 0.00049103 0.22923972 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012744 0.04349830 layer.3.ssm_state 0.00000001 0.00000275 layer.3.conv_state 0.00011997 0.03784204 layer.4.ssm_state 0.00000003 0.00000464 layer.4.conv_state 0.00026647 0.08690345 layer.4.output 0.00000220 0.00017144 ------------------------------------------------------------------------------------- TOTAL 0.00002599 0.01042397 (elements=1,511,424) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1511424 Total Bytes 723316 BPFP 3.8285 bits/point EBPFP 6.5068 equivalent bits/point MSE 0.010424 ---------------------- -------------------------------------------------------- Time: 4.965s Load: 0.011s, Pack+Encode: 2.639s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample103-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample103-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample104-layer4-item1.zst (7/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample104-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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, 189, 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, 189, 4096]) -> torch.Size([1, 1, 189, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,176B, BPFP=4.7104 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,132B, BPFP=2.5105 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 77,812B, BPFP=4.7493 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,176B, BPFP=10.5410 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,756B, BPFP=3.5862 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,584B, BPFP=10.3965 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,616B, BPFP=3.2114 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,416B, BPFP=9.1348 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,840B, BPFP=2.2512 ⌛️ [2/4] FRONTEND: Frontend time: 2.582s (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, 189, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 189, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00004187 layer.1.conv_state 0.00050086 0.22829886 layer.2.ssm_state 0.00000001 0.00000212 layer.2.conv_state 0.00014194 0.04439721 layer.3.ssm_state 0.00000001 0.00000269 layer.3.conv_state 0.00011857 0.03769404 layer.4.ssm_state 0.00000005 0.00000449 layer.4.conv_state 0.00026192 0.08724105 layer.4.output 0.00000092 0.00019461 ------------------------------------------------------------------------------------- TOTAL 0.00002453 0.00991121 (elements=1,593,344) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1593344 Total Bytes 722136 BPFP 3.6258 bits/point EBPFP 6.1578 equivalent bits/point MSE 0.009911 ---------------------- -------------------------------------------------------- Time: 4.914s Load: 0.012s, Pack+Encode: 2.582s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample104-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample104-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample106-layer4-item1.zst (8/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample106-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 212, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 212, 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, 212, 4096]) -> torch.Size([1, 1, 212, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,788B, BPFP=4.7478 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,352B, BPFP=2.5239 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,756B, BPFP=4.8069 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,104B, BPFP=10.5234 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,632B, BPFP=3.5786 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,468B, BPFP=10.3682 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,300B, BPFP=3.1921 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,316B, BPFP=9.1104 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 264,832B, BPFP=2.4399 ⌛️ [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, 212, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.377s [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, 212, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00004062 layer.1.conv_state 0.00048019 0.22617222 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00013564 0.04376603 layer.3.ssm_state 0.00000001 0.00000284 layer.3.conv_state 0.00011743 0.03735178 layer.4.ssm_state 0.00000004 0.00000442 layer.4.conv_state 0.00027427 0.08706542 layer.4.output 0.00000069 0.00016509 ------------------------------------------------------------------------------------- TOTAL 0.00002277 0.00928990 (elements=1,687,552) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1687552 Total Bytes 770176 BPFP 3.6511 bits/point EBPFP 6.0467 equivalent bits/point MSE 0.009290 ---------------------- -------------------------------------------------------- Time: 5.018s Load: 0.013s, Pack+Encode: 2.628s, Decode+Unpack: 2.377s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 212, 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 0.0093 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample106-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample106-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample107-layer4-item1.zst (9/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample107-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 172, 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, 172, 4096]) -> torch.Size([1, 1, 172, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,348B, BPFP=4.6599 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,140B, BPFP=2.5110 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,148B, BPFP=4.8918 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,092B, BPFP=10.5205 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,280B, BPFP=3.6182 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,576B, BPFP=10.3945 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,072B, BPFP=3.1782 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,344B, BPFP=9.1172 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,156B, BPFP=2.4659 ⌛️ [2/4] FRONTEND: Frontend time: 2.569s (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, 172, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 172, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000013 0.00004089 layer.1.conv_state 0.00051710 0.22559139 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00009645 0.04450054 layer.3.ssm_state 0.00000001 0.00000276 layer.3.conv_state 0.00011583 0.03725050 layer.4.ssm_state 0.00000004 0.00000477 layer.4.conv_state 0.00025227 0.09016059 layer.4.output 0.00000096 0.00018563 ------------------------------------------------------------------------------------- TOTAL 0.00002471 0.01034829 (elements=1,523,712) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1523712 Total Bytes 722780 BPFP 3.7948 bits/point EBPFP 6.4495 equivalent bits/point MSE 0.010348 ---------------------- -------------------------------------------------------- Time: 4.904s Load: 0.011s, Pack+Encode: 2.569s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample107-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample107-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample108-layer4-item1.zst (10/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample108-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 174, 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, 174, 4096]) -> torch.Size([1, 1, 174, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,972B, BPFP=4.6370 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,156B, BPFP=2.5120 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,856B, BPFP=4.8130 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,324B, BPFP=10.5771 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,948B, BPFP=3.5979 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,620B, BPFP=10.4053 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,732B, BPFP=3.1575 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,336B, BPFP=9.1152 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,892B, BPFP=2.4570 ⌛️ [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, 174, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.309s [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, 174, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004053 layer.1.conv_state 0.00048949 0.22937435 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012781 0.04474622 layer.3.ssm_state 0.00000001 0.00000274 layer.3.conv_state 0.00007456 0.03762795 layer.4.ssm_state 0.00000001 0.00000479 layer.4.conv_state 0.00026463 0.08869506 layer.4.output 0.00000097 0.00017236 ------------------------------------------------------------------------------------- TOTAL 0.00002404 0.01035064 (elements=1,531,904) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1531904 Total Bytes 722504 BPFP 3.7731 bits/point EBPFP 6.4031 equivalent bits/point MSE 0.010351 ---------------------- -------------------------------------------------------- Time: 4.917s Load: 0.011s, Pack+Encode: 2.597s, Decode+Unpack: 2.309s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample108-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample108-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample109-layer4-item1.zst (11/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample109-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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, 180, 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, 180, 4096]) -> torch.Size([1, 1, 180, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,108B, BPFP=4.6453 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,360B, BPFP=2.5244 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,516B, BPFP=8.9150 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,576B, BPFP=4.8569 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,268B, BPFP=10.5635 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,784B, BPFP=3.6489 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,536B, BPFP=10.3848 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,492B, BPFP=3.1428 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,212B, BPFP=9.0850 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 224,000B, BPFP=2.4306 ⌛️ [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, 180, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.315s [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, 180, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004136 layer.1.conv_state 0.00051305 0.22870813 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00014565 0.04412624 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00011640 0.03784061 layer.4.ssm_state 0.00000001 0.00000461 layer.4.conv_state 0.00024851 0.08790866 layer.4.output 0.00000215 0.00018212 ------------------------------------------------------------------------------------- TOTAL 0.00002566 0.01015545 (elements=1,556,480) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1556480 Total Bytes 728996 BPFP 3.7469 bits/point EBPFP 6.3425 equivalent bits/point MSE 0.010155 ---------------------- -------------------------------------------------------- Time: 4.905s Load: 0.012s, Pack+Encode: 2.578s, Decode+Unpack: 2.315s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample109-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample109-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample110-layer4-item1.zst (12/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample110-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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, 174, 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, 174, 4096]) -> torch.Size([1, 1, 174, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,184B, BPFP=4.6499 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,220B, BPFP=2.5159 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,512B, BPFP=8.9141 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,176B, BPFP=4.9546 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,336B, BPFP=10.5801 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,088B, BPFP=3.6064 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,548B, BPFP=10.3877 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,556B, BPFP=3.2078 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,412B, BPFP=9.1338 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 220,104B, BPFP=2.4706 ⌛️ [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, 174, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.309s [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, 174, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000138 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00004275 layer.1.conv_state 0.00049883 0.22795279 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00011290 0.04452610 layer.3.ssm_state 0.00000001 0.00000277 layer.3.conv_state 0.00007560 0.03794086 layer.4.ssm_state 0.00000004 0.00000462 layer.4.conv_state 0.00028441 0.08374210 layer.4.output 0.00000099 0.00017610 ------------------------------------------------------------------------------------- TOTAL 0.00002440 0.01021818 (elements=1,531,904) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1531904 Total Bytes 727280 BPFP 3.7980 bits/point EBPFP 6.4466 equivalent bits/point MSE 0.010218 ---------------------- -------------------------------------------------------- Time: 4.899s Load: 0.009s, Pack+Encode: 2.581s, Decode+Unpack: 2.309s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 174, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample110-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample110-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample113-layer4-item1.zst (13/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample113-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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, 167, 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, 167, 4096]) -> torch.Size([1, 1, 167, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,180B, BPFP=4.6497 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,100B, BPFP=2.5085 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,456B, BPFP=8.9004 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,752B, BPFP=4.9287 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,148B, BPFP=10.5342 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,000B, BPFP=3.6011 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,532B, BPFP=10.3838 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,484B, BPFP=3.2034 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,392B, BPFP=9.1289 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,428B, BPFP=2.5312 ⌛️ [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, 167, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.334s [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, 167, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00004049 layer.1.conv_state 0.00051076 0.22424726 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00016296 0.04439956 layer.3.ssm_state 0.00000001 0.00000279 layer.3.conv_state 0.00011872 0.03785230 layer.4.ssm_state 0.00000003 0.00000468 layer.4.conv_state 0.00025742 0.08676006 layer.4.output 0.00000098 0.00016720 ------------------------------------------------------------------------------------- TOTAL 0.00002655 0.01038581 (elements=1,503,232) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1503232 Total Bytes 722616 BPFP 3.8457 bits/point EBPFP 6.5395 equivalent bits/point MSE 0.010386 ---------------------- -------------------------------------------------------- Time: 4.946s Load: 0.012s, Pack+Encode: 2.600s, Decode+Unpack: 2.334s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample113-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample113-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample114-layer4-item1.zst (14/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample114-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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, 165, 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, 165, 4096]) -> torch.Size([1, 1, 165, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,500B, BPFP=4.6692 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,108B, BPFP=2.5090 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,812B, BPFP=4.9934 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,184B, BPFP=10.5430 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,608B, BPFP=3.6382 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,464B, BPFP=10.3672 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,780B, BPFP=3.1604 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,292B, BPFP=9.1045 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 213,928B, BPFP=2.5323 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 165, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.327s [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, 165, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000025 0.00003975 layer.1.conv_state 0.00049071 0.22892812 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00013667 0.04387150 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00011918 0.03749678 layer.4.ssm_state 0.00000005 0.00000479 layer.4.conv_state 0.00026751 0.09183728 layer.4.output 0.00000105 0.00018441 ------------------------------------------------------------------------------------- TOTAL 0.00002593 0.01064403 (elements=1,495,040) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1495040 Total Bytes 721344 BPFP 3.8599 bits/point EBPFP 6.5751 equivalent bits/point MSE 0.010644 ---------------------- -------------------------------------------------------- Time: 4.930s Load: 0.012s, Pack+Encode: 2.591s, Decode+Unpack: 2.327s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample114-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample114-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample115-layer4-item1.zst (15/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample115-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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, 169, 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, 169, 4096]) -> torch.Size([1, 1, 169, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,756B, BPFP=4.6238 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,144B, BPFP=2.5112 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,476B, BPFP=8.9053 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,024B, BPFP=4.8232 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,200B, BPFP=10.5469 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,076B, BPFP=3.6057 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,576B, BPFP=10.3945 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,980B, BPFP=3.2336 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,240B, BPFP=9.0918 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,060B, BPFP=2.4970 ⌛️ [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, 169, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 169, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000026 0.00004038 layer.1.conv_state 0.00050717 0.22786210 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012932 0.04355327 layer.3.ssm_state 0.00000001 0.00000266 layer.3.conv_state 0.00011861 0.03784433 layer.4.ssm_state 0.00000004 0.00000426 layer.4.conv_state 0.00023032 0.08734313 layer.4.output 0.00000228 0.00017066 ------------------------------------------------------------------------------------- TOTAL 0.00002561 0.01040444 (elements=1,511,424) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1511424 Total Bytes 720676 BPFP 3.8146 bits/point EBPFP 6.4855 equivalent bits/point MSE 0.010404 ---------------------- -------------------------------------------------------- Time: 4.926s Load: 0.012s, Pack+Encode: 2.594s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample115-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample115-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample116-layer4-item1.zst (16/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample116-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 191, 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, 191, 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, 191, 4096]) -> torch.Size([1, 1, 191, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,036B, BPFP=4.7019 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,172B, BPFP=2.5129 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,504B, BPFP=8.9121 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,704B, BPFP=4.8037 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,352B, BPFP=10.5840 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,068B, BPFP=3.5442 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,780B, BPFP=10.4443 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,192B, BPFP=3.1855 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,580B, BPFP=9.1748 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 208,796B, BPFP=2.1351 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 191, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.323s [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, 191, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000135 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000020 0.00004185 layer.1.conv_state 0.00048712 0.22626902 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012842 0.04554232 layer.3.ssm_state 0.00000001 0.00000262 layer.3.conv_state 0.00007613 0.03836765 layer.4.ssm_state 0.00000006 0.00000491 layer.4.conv_state 0.00030268 0.08916326 layer.4.output 0.00000202 0.00017428 ------------------------------------------------------------------------------------- TOTAL 0.00002434 0.00988661 (elements=1,601,536) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1601536 Total Bytes 713328 BPFP 3.5632 bits/point EBPFP 6.0835 equivalent bits/point MSE 0.009887 ---------------------- -------------------------------------------------------- Time: 4.924s Load: 0.010s, Pack+Encode: 2.591s, Decode+Unpack: 2.323s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 191, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample116-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample116-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample117-layer4-item1.zst (17/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample117-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 168, 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, 168, 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, 168, 4096]) -> torch.Size([1, 1, 168, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,104B, BPFP=4.6450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,200B, BPFP=2.5146 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,520B, BPFP=8.9160 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,712B, BPFP=4.9263 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,200B, BPFP=10.5469 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,116B, BPFP=3.6082 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,612B, BPFP=10.4033 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,316B, BPFP=3.1931 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,216B, BPFP=9.0859 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,164B, BPFP=2.5363 ⌛️ [2/4] FRONTEND: Frontend time: 2.582s (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, 168, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 168, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000032 0.00004059 layer.1.conv_state 0.00049322 0.22523990 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00013530 0.04333073 layer.3.ssm_state 0.00000001 0.00000277 layer.3.conv_state 0.00007654 0.03747981 layer.4.ssm_state 0.00000004 0.00000468 layer.4.conv_state 0.00024931 0.08821815 layer.4.output 0.00000099 0.00017751 ------------------------------------------------------------------------------------- TOTAL 0.00002441 0.01038470 (elements=1,507,328) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1507328 Total Bytes 724304 BPFP 3.8442 bits/point EBPFP 6.5305 equivalent bits/point MSE 0.010385 ---------------------- -------------------------------------------------------- Time: 4.914s Load: 0.012s, Pack+Encode: 2.582s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 168, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample117-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample117-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample118-layer4-item1.zst (18/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample118-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 166, 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, 166, 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, 166, 4096]) -> torch.Size([1, 1, 166, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,848B, BPFP=4.6294 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,136B, BPFP=2.5107 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,436B, BPFP=4.9094 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,216B, BPFP=10.5508 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,996B, BPFP=3.6008 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,564B, BPFP=10.3916 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,932B, BPFP=3.2307 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,268B, BPFP=9.0986 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,172B, BPFP=2.5670 ⌛️ [2/4] FRONTEND: Frontend time: 2.585s (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, 166, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.327s [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, 166, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00003848 layer.1.conv_state 0.00049022 0.22666533 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00015195 0.04387470 layer.3.ssm_state 0.00000001 0.00000269 layer.3.conv_state 0.00007647 0.03767405 layer.4.ssm_state 0.00000002 0.00000430 layer.4.conv_state 0.00025737 0.08363914 layer.4.output 0.00000101 0.00017074 ------------------------------------------------------------------------------------- TOTAL 0.00002500 0.01038439 (elements=1,499,136) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1499136 Total Bytes 724196 BPFP 3.8646 bits/point EBPFP 6.5650 equivalent bits/point MSE 0.010384 ---------------------- -------------------------------------------------------- Time: 4.921s Load: 0.010s, Pack+Encode: 2.585s, Decode+Unpack: 2.327s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 166, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample118-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample118-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample119-layer4-item1.zst (19/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample119-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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, 173, 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, 173, 4096]) -> torch.Size([1, 1, 173, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,568B, BPFP=4.6733 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,220B, BPFP=2.5159 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,568B, BPFP=8.9277 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,832B, BPFP=4.9336 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,340B, BPFP=10.5811 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,888B, BPFP=3.6553 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,728B, BPFP=10.4316 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,708B, BPFP=3.1560 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,488B, BPFP=9.1523 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 231,780B, BPFP=2.6167 ⌛️ [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, 173, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 173, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004103 layer.1.conv_state 0.00049119 0.22429855 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00012655 0.04499030 layer.3.ssm_state 0.00000001 0.00000296 layer.3.conv_state 0.00006970 0.03757416 layer.4.ssm_state 0.00000005 0.00000481 layer.4.conv_state 0.00034659 0.08867839 layer.4.output 0.00000108 0.00019821 ------------------------------------------------------------------------------------- TOTAL 0.00002583 0.01028484 (elements=1,527,808) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1527808 Total Bytes 739264 BPFP 3.8710 bits/point EBPFP 6.5283 equivalent bits/point MSE 0.010285 ---------------------- -------------------------------------------------------- Time: 4.913s Load: 0.010s, Pack+Encode: 2.589s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample119-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample119-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample12-layer4-item1.zst (20/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample12-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 211, 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, 211, 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, 211, 4096]) -> torch.Size([1, 1, 211, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,444B, BPFP=4.6047 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,232B, BPFP=2.5166 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,500B, BPFP=8.9111 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,240B, BPFP=4.8364 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,268B, BPFP=10.5635 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,320B, BPFP=3.6206 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,700B, BPFP=10.4248 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,096B, BPFP=3.2407 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,488B, BPFP=9.1523 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 268,492B, BPFP=2.4853 ⌛️ [2/4] FRONTEND: Frontend time: 2.647s (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, 211, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.368s [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, 211, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000013 0.00004347 layer.1.conv_state 0.00049046 0.22208822 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012141 0.04477833 layer.3.ssm_state 0.00000001 0.00000283 layer.3.conv_state 0.00007583 0.03797333 layer.4.ssm_state 0.00000003 0.00000456 layer.4.conv_state 0.00022251 0.08862507 layer.4.output 0.00000070 0.00015296 ------------------------------------------------------------------------------------- TOTAL 0.00002094 0.00928877 (elements=1,683,456) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1683456 Total Bytes 773924 BPFP 3.6778 bits/point EBPFP 6.0797 equivalent bits/point MSE 0.009289 ---------------------- -------------------------------------------------------- Time: 5.027s Load: 0.012s, Pack+Encode: 2.647s, Decode+Unpack: 2.368s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 211, 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 0.0093 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample12-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample12-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample120-layer4-item1.zst (21/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample120-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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, 175, 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, 175, 4096]) -> torch.Size([1, 1, 175, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,784B, BPFP=4.6865 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,188B, BPFP=2.5139 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,532B, BPFP=8.9189 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,548B, BPFP=4.7942 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,364B, BPFP=10.5869 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,148B, BPFP=3.6101 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,708B, BPFP=10.4268 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,004B, BPFP=3.1741 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,508B, BPFP=9.1572 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,688B, BPFP=2.4854 ⌛️ [2/4] FRONTEND: Frontend time: 2.582s (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, 175, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.326s [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, 175, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004126 layer.1.conv_state 0.00050827 0.22506750 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012854 0.04405423 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00007639 0.03746553 layer.4.ssm_state 0.00000003 0.00000468 layer.4.conv_state 0.00023882 0.08992925 layer.4.output 0.00000097 0.00017328 ------------------------------------------------------------------------------------- TOTAL 0.00002389 0.01024020 (elements=1,536,000) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1536000 Total Bytes 727616 BPFP 3.7897 bits/point EBPFP 6.4195 equivalent bits/point MSE 0.010240 ---------------------- -------------------------------------------------------- Time: 4.920s Load: 0.012s, Pack+Encode: 2.582s, Decode+Unpack: 2.326s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample120-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample120-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample121-layer4-item1.zst (22/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample121-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 157, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 157, 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, 157, 4096]) -> torch.Size([1, 1, 157, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,952B, BPFP=4.6357 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,228B, BPFP=2.5164 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,456B, BPFP=8.9004 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,400B, BPFP=4.9072 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,296B, BPFP=10.5703 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,016B, BPFP=3.6021 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,496B, BPFP=10.3750 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,376B, BPFP=3.1968 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,204B, BPFP=9.0830 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 207,916B, BPFP=2.5865 ⌛️ [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, 157, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 157, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000002 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00004147 layer.1.conv_state 0.00053269 0.22914065 layer.2.ssm_state 0.00000001 0.00000211 layer.2.conv_state 0.00012877 0.04462170 layer.3.ssm_state 0.00000001 0.00000265 layer.3.conv_state 0.00007005 0.03739958 layer.4.ssm_state 0.00000001 0.00000480 layer.4.conv_state 0.00032121 0.08981387 layer.4.output 0.00000217 0.00019178 ------------------------------------------------------------------------------------- TOTAL 0.00002783 0.01085586 (elements=1,462,272) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1462272 Total Bytes 713484 BPFP 3.9034 bits/point EBPFP 6.6694 equivalent bits/point MSE 0.010856 ---------------------- -------------------------------------------------------- Time: 4.907s Load: 0.011s, Pack+Encode: 2.576s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 157, 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 0.0109 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample121-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample121-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample122-layer4-item1.zst (23/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample122-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 173, 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, 173, 4096]) -> torch.Size([1, 1, 173, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,196B, BPFP=4.6506 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,104B, BPFP=2.5088 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,488B, BPFP=8.9082 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,396B, BPFP=4.9070 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,148B, BPFP=10.5342 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,608B, BPFP=3.6382 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,496B, BPFP=10.3750 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,524B, BPFP=3.2058 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,352B, BPFP=9.1191 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,792B, BPFP=2.4814 ⌛️ [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, 173, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 173, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00003949 layer.1.conv_state 0.00050344 0.22390787 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00017044 0.04418293 layer.3.ssm_state 0.00000001 0.00000290 layer.3.conv_state 0.00006875 0.03743217 layer.4.ssm_state 0.00000001 0.00000427 layer.4.conv_state 0.00023119 0.08742347 layer.4.output 0.00000097 0.00018922 ------------------------------------------------------------------------------------- TOTAL 0.00002450 0.01022483 (elements=1,527,808) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1527808 Total Bytes 726248 BPFP 3.8028 bits/point EBPFP 6.4548 equivalent bits/point MSE 0.010225 ---------------------- -------------------------------------------------------- Time: 4.916s Load: 0.011s, Pack+Encode: 2.586s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample122-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample122-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample123-layer4-item1.zst (24/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample123-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 168, 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, 168, 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, 168, 4096]) -> torch.Size([1, 1, 168, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,664B, BPFP=4.6182 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,256B, BPFP=2.5181 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,476B, BPFP=8.9053 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,428B, BPFP=4.8479 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,088B, BPFP=10.5195 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,384B, BPFP=3.6245 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,420B, BPFP=10.3564 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,028B, BPFP=3.2366 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,204B, BPFP=9.0830 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 213,704B, BPFP=2.4845 ⌛️ [2/4] FRONTEND: Frontend time: 2.571s (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, 168, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 168, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000044 0.00004039 layer.1.conv_state 0.00048555 0.22845073 layer.2.ssm_state 0.00000001 0.00000211 layer.2.conv_state 0.00015447 0.04411953 layer.3.ssm_state 0.00000001 0.00000264 layer.3.conv_state 0.00011376 0.03749028 layer.4.ssm_state 0.00000003 0.00000418 layer.4.conv_state 0.00022332 0.08415904 layer.4.output 0.00000212 0.00017356 ------------------------------------------------------------------------------------- TOTAL 0.00002543 0.01038175 (elements=1,507,328) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1507328 Total Bytes 718796 BPFP 3.8149 bits/point EBPFP 6.4957 equivalent bits/point MSE 0.010382 ---------------------- -------------------------------------------------------- Time: 4.893s Load: 0.009s, Pack+Encode: 2.571s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 168, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample123-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample123-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample124-layer4-item1.zst (25/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample124-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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, 175, 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, 175, 4096]) -> torch.Size([1, 1, 175, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,108B, BPFP=4.7063 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,188B, BPFP=2.5139 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,440B, BPFP=8.8965 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,288B, BPFP=4.9004 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,204B, BPFP=10.5479 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,976B, BPFP=3.6606 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,472B, BPFP=10.3691 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,772B, BPFP=3.1599 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,200B, BPFP=9.0820 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,620B, BPFP=2.4400 ⌛️ [2/4] FRONTEND: Frontend time: 2.584s (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, 175, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.322s [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, 175, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00003820 layer.1.conv_state 0.00049991 0.23051021 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00017222 0.04448456 layer.3.ssm_state 0.00000001 0.00000281 layer.3.conv_state 0.00011700 0.03768902 layer.4.ssm_state 0.00000001 0.00000455 layer.4.conv_state 0.00021726 0.08125558 layer.4.output 0.00000217 0.00017254 ------------------------------------------------------------------------------------- TOTAL 0.00002563 0.01018461 (elements=1,536,000) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1536000 Total Bytes 725412 BPFP 3.7782 bits/point EBPFP 6.4177 equivalent bits/point MSE 0.010185 ---------------------- -------------------------------------------------------- Time: 4.916s Load: 0.009s, Pack+Encode: 2.584s, Decode+Unpack: 2.322s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample124-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample124-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample126-layer4-item1.zst (26/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample126-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 176, 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, 176, 4096]) -> torch.Size([1, 1, 176, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,072B, BPFP=4.6431 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,228B, BPFP=2.5164 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,512B, BPFP=8.9141 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,308B, BPFP=4.9016 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,208B, BPFP=10.5488 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 60,196B, BPFP=3.6741 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,536B, BPFP=10.3848 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,880B, BPFP=3.1665 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,328B, BPFP=9.1133 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,924B, BPFP=2.4295 ⌛️ [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, 176, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.326s [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, 176, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000017 0.00004027 layer.1.conv_state 0.00049536 0.22657387 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012767 0.04347206 layer.3.ssm_state 0.00000001 0.00000279 layer.3.conv_state 0.00007152 0.03815646 layer.4.ssm_state 0.00000005 0.00000490 layer.4.conv_state 0.00031774 0.08866832 layer.4.output 0.00000096 0.00018581 ------------------------------------------------------------------------------------- TOTAL 0.00002512 0.01022677 (elements=1,540,096) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1540096 Total Bytes 725336 BPFP 3.7677 bits/point EBPFP 6.3983 equivalent bits/point MSE 0.010227 ---------------------- -------------------------------------------------------- Time: 4.948s Load: 0.011s, Pack+Encode: 2.611s, Decode+Unpack: 2.326s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample126-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample126-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample127-layer4-item1.zst (27/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample127-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 171, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 171, 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, 171, 4096]) -> torch.Size([1, 1, 171, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,340B, BPFP=4.6594 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,188B, BPFP=2.5139 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,532B, BPFP=8.9189 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,156B, BPFP=4.9534 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,296B, BPFP=10.5703 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,472B, BPFP=3.6299 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,436B, BPFP=10.3604 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,620B, BPFP=3.2117 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,120B, BPFP=9.0625 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 223,008B, BPFP=2.5471 ⌛️ [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, 171, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.334s [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, 171, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004044 layer.1.conv_state 0.00050833 0.22809355 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00017812 0.04372467 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00007045 0.03778579 layer.4.ssm_state 0.00000001 0.00000443 layer.4.conv_state 0.00022870 0.08378316 layer.4.output 0.00000218 0.00016941 ------------------------------------------------------------------------------------- TOTAL 0.00002542 0.01027938 (elements=1,519,616) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1519616 Total Bytes 730312 BPFP 3.8447 bits/point EBPFP 6.5154 equivalent bits/point MSE 0.010279 ---------------------- -------------------------------------------------------- Time: 4.956s Load: 0.011s, Pack+Encode: 2.610s, Decode+Unpack: 2.334s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 171, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample127-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample127-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample128-layer4-item1.zst (28/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample128-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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, 167, 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, 167, 4096]) -> torch.Size([1, 1, 167, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,904B, BPFP=4.6328 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 40,936B, BPFP=2.4985 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,804B, BPFP=4.8098 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,316B, BPFP=10.5752 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,212B, BPFP=3.6140 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,652B, BPFP=10.4131 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,612B, BPFP=3.2112 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,564B, BPFP=9.1709 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 214,144B, BPFP=2.5045 ⌛️ [2/4] FRONTEND: Frontend time: 2.595s (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, 167, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 167, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00004147 layer.1.conv_state 0.00049222 0.22702736 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00011339 0.04501544 layer.3.ssm_state 0.00000001 0.00000288 layer.3.conv_state 0.00011830 0.03784801 layer.4.ssm_state 0.00000004 0.00000438 layer.4.conv_state 0.00023948 0.08671293 layer.4.output 0.00000097 0.00016667 ------------------------------------------------------------------------------------- TOTAL 0.00002464 0.01045854 (elements=1,503,232) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1503232 Total Bytes 718824 BPFP 3.8255 bits/point EBPFP 6.5113 equivalent bits/point MSE 0.010459 ---------------------- -------------------------------------------------------- Time: 4.924s Load: 0.009s, Pack+Encode: 2.595s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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 0.0105 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample128-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample128-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample129-layer4-item1.zst (29/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample129-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 162, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 162, 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, 162, 4096]) -> torch.Size([1, 1, 162, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,044B, BPFP=4.6414 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,280B, BPFP=2.5195 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,476B, BPFP=8.9053 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,184B, BPFP=4.8330 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,144B, BPFP=10.5332 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,552B, BPFP=3.6348 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,460B, BPFP=10.3662 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,812B, BPFP=3.2234 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,132B, BPFP=9.0654 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 211,964B, BPFP=2.5555 ⌛️ [2/4] FRONTEND: Frontend time: 2.584s (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, 162, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.318s [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, 162, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000038 0.00004031 layer.1.conv_state 0.00051438 0.22813937 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00014631 0.04454682 layer.3.ssm_state 0.00000001 0.00000270 layer.3.conv_state 0.00011822 0.03809444 layer.4.ssm_state 0.00000002 0.00000423 layer.4.conv_state 0.00021955 0.08426554 layer.4.output 0.00000102 0.00017438 ------------------------------------------------------------------------------------- TOTAL 0.00002579 0.01056959 (elements=1,482,752) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1482752 Total Bytes 717192 BPFP 3.8695 bits/point EBPFP 6.5954 equivalent bits/point MSE 0.010570 ---------------------- -------------------------------------------------------- Time: 4.913s Load: 0.011s, Pack+Encode: 2.584s, Decode+Unpack: 2.318s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 162, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample129-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample129-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample13-layer4-item1.zst (30/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample13-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 219, 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, 219, 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, 219, 4096]) -> torch.Size([1, 1, 219, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,516B, BPFP=4.6702 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,232B, BPFP=2.5166 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,728B, BPFP=4.9272 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,228B, BPFP=10.5537 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,308B, BPFP=3.6199 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,652B, BPFP=10.4131 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,396B, BPFP=3.1370 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,180B, BPFP=9.0771 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 276,296B, BPFP=2.4641 ⌛️ [2/4] FRONTEND: Frontend time: 2.672s (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, 219, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.377s [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, 219, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000011 0.00004034 layer.1.conv_state 0.00049576 0.22770710 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014806 0.04401023 layer.3.ssm_state 0.00000001 0.00000269 layer.3.conv_state 0.00007436 0.03743833 layer.4.ssm_state 0.00000001 0.00000464 layer.4.conv_state 0.00026664 0.09027387 layer.4.output 0.00000170 0.00014339 ------------------------------------------------------------------------------------- TOTAL 0.00002249 0.00922298 (elements=1,716,224) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1716224 Total Bytes 782216 BPFP 3.6462 bits/point EBPFP 6.0045 equivalent bits/point MSE 0.009223 ---------------------- -------------------------------------------------------- Time: 5.059s Load: 0.010s, Pack+Encode: 2.672s, Decode+Unpack: 2.377s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 219, 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 0.0092 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample13-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample13-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample131-layer4-item1.zst (31/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample131-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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, 169, 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, 169, 4096]) -> torch.Size([1, 1, 169, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,876B, BPFP=4.6311 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,256B, BPFP=2.5181 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,528B, BPFP=8.9180 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,204B, BPFP=4.8342 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,300B, BPFP=10.5713 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,916B, BPFP=3.5959 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,708B, BPFP=10.4268 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,220B, BPFP=3.2483 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,452B, BPFP=9.1436 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,004B, BPFP=2.5195 ⌛️ [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, 169, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.311s [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, 169, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004116 layer.1.conv_state 0.00050947 0.22862828 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00012820 0.04421166 layer.3.ssm_state 0.00000001 0.00000278 layer.3.conv_state 0.00011665 0.03778424 layer.4.ssm_state 0.00000002 0.00000426 layer.4.conv_state 0.00025445 0.08510243 layer.4.output 0.00000228 0.00017221 ------------------------------------------------------------------------------------- TOTAL 0.00002610 0.01038623 (elements=1,511,424) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1511424 Total Bytes 723608 BPFP 3.8301 bits/point EBPFP 6.5062 equivalent bits/point MSE 0.010386 ---------------------- -------------------------------------------------------- Time: 4.907s Load: 0.010s, Pack+Encode: 2.586s, Decode+Unpack: 2.311s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample131-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample131-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample134-layer4-item1.zst (32/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample134-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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, 172, 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, 172, 4096]) -> torch.Size([1, 1, 172, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,228B, BPFP=4.6526 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,180B, BPFP=2.5134 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,516B, BPFP=8.9150 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,092B, BPFP=4.8884 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,220B, BPFP=10.5518 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,084B, BPFP=3.6062 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,488B, BPFP=10.3730 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,196B, BPFP=3.1858 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,176B, BPFP=9.0762 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,556B, BPFP=2.4704 ⌛️ [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, 172, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.315s [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, 172, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004060 layer.1.conv_state 0.00050930 0.22806400 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00013921 0.04393781 layer.3.ssm_state 0.00000001 0.00000280 layer.3.conv_state 0.00007566 0.03723041 layer.4.ssm_state 0.00000001 0.00000471 layer.4.conv_state 0.00023771 0.08912034 layer.4.output 0.00000095 0.00018786 ------------------------------------------------------------------------------------- TOTAL 0.00002428 0.01036757 (elements=1,523,712) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1523712 Total Bytes 722880 BPFP 3.7954 bits/point EBPFP 6.4485 equivalent bits/point MSE 0.010368 ---------------------- -------------------------------------------------------- Time: 4.932s Load: 0.012s, Pack+Encode: 2.605s, Decode+Unpack: 2.315s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample134-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample134-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample142-layer4-item1.zst (33/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample142-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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, 176, 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, 176, 4096]) -> torch.Size([1, 1, 176, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,916B, BPFP=4.6335 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,228B, BPFP=2.5164 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 77,864B, BPFP=4.7524 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,268B, BPFP=10.5635 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,300B, BPFP=3.5583 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,508B, BPFP=10.3779 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,164B, BPFP=3.1838 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,264B, BPFP=9.0977 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 225,772B, BPFP=2.5055 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 176, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 176, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004113 layer.1.conv_state 0.00050048 0.22425660 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00010870 0.04405002 layer.3.ssm_state 0.00000001 0.00000266 layer.3.conv_state 0.00007647 0.03748007 layer.4.ssm_state 0.00000001 0.00000435 layer.4.conv_state 0.00023388 0.08655697 layer.4.output 0.00000101 0.00019272 ------------------------------------------------------------------------------------- TOTAL 0.00002316 0.01013370 (elements=1,540,096) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1540096 Total Bytes 727964 BPFP 3.7814 bits/point EBPFP 6.3900 equivalent bits/point MSE 0.010134 ---------------------- -------------------------------------------------------- Time: 4.922s Load: 0.010s, Pack+Encode: 2.591s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample142-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample142-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample16-layer4-item1.zst (34/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample16-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 211, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 211, 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, 211, 4096]) -> torch.Size([1, 1, 211, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,256B, BPFP=4.6543 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,192B, BPFP=2.5142 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,476B, BPFP=8.9053 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,712B, BPFP=4.9873 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,288B, BPFP=10.5684 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 60,088B, BPFP=3.6675 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,584B, BPFP=10.3965 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,912B, BPFP=3.1685 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,180B, BPFP=9.0771 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 275,952B, BPFP=2.5544 ⌛️ [2/4] FRONTEND: Frontend time: 2.642s (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, 211, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.383s [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, 211, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00003928 layer.1.conv_state 0.00050701 0.22573087 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00017315 0.04378051 layer.3.ssm_state 0.00000001 0.00000279 layer.3.conv_state 0.00007047 0.03765609 layer.4.ssm_state 0.00000001 0.00000467 layer.4.conv_state 0.00027918 0.09039294 layer.4.output 0.00000169 0.00015837 ------------------------------------------------------------------------------------- TOTAL 0.00002377 0.00937095 (elements=1,683,456) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1683456 Total Bytes 783784 BPFP 3.7246 bits/point EBPFP 6.1379 equivalent bits/point MSE 0.009371 ---------------------- -------------------------------------------------------- Time: 5.038s Load: 0.013s, Pack+Encode: 2.642s, Decode+Unpack: 2.383s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 211, 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 0.0094 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample16-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample16-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample17-layer4-item1.zst (35/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample17-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 220, 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, 220, 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, 220, 4096]) -> torch.Size([1, 1, 220, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,404B, BPFP=4.6633 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,080B, BPFP=2.5073 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,504B, BPFP=8.9121 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,264B, BPFP=4.7769 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,276B, BPFP=10.5654 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,048B, BPFP=3.6040 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,652B, BPFP=10.4131 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,676B, BPFP=3.2151 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,584B, BPFP=9.1758 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 277,480B, BPFP=2.4634 ⌛️ [2/4] FRONTEND: Frontend time: 2.648s (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, 220, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.382s [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, 220, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000138 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00004031 layer.1.conv_state 0.00051478 0.22797228 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00013967 0.04504493 layer.3.ssm_state 0.00000001 0.00000280 layer.3.conv_state 0.00011849 0.03816149 layer.4.ssm_state 0.00000002 0.00000430 layer.4.conv_state 0.00022568 0.08687506 layer.4.output 0.00000159 0.00014695 ------------------------------------------------------------------------------------- TOTAL 0.00002266 0.00917699 (elements=1,720,320) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1720320 Total Bytes 782112 BPFP 3.6371 bits/point EBPFP 5.9837 equivalent bits/point MSE 0.009177 ---------------------- -------------------------------------------------------- Time: 5.040s Load: 0.010s, Pack+Encode: 2.648s, Decode+Unpack: 2.382s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 220, 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 0.0092 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample17-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample17-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample176-layer4-item1.zst (36/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample176-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 161, 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, 161, 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, 161, 4096]) -> torch.Size([1, 1, 161, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,080B, BPFP=4.5825 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,196B, BPFP=2.5144 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,072B, BPFP=4.7651 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,232B, BPFP=10.5547 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,704B, BPFP=3.5830 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,628B, BPFP=10.4072 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,936B, BPFP=3.1699 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,308B, BPFP=9.1084 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 214,352B, BPFP=2.6003 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 161, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.313s [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, 161, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000136 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004087 layer.1.conv_state 0.00050394 0.22387618 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00015457 0.04395396 layer.3.ssm_state 0.00000001 0.00000275 layer.3.conv_state 0.00007685 0.03755976 layer.4.ssm_state 0.00000001 0.00000460 layer.4.conv_state 0.00028870 0.08879188 layer.4.output 0.00000108 0.00018685 ------------------------------------------------------------------------------------- TOTAL 0.00002642 0.01058487 (elements=1,478,656) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1478656 Total Bytes 716176 BPFP 3.8747 bits/point EBPFP 6.5898 equivalent bits/point MSE 0.010585 ---------------------- -------------------------------------------------------- Time: 4.914s Load: 0.010s, Pack+Encode: 2.591s, Decode+Unpack: 2.313s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 161, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample176-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample176-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample18-layer4-item1.zst (37/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample18-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 202, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 202, 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, 202, 4096]) -> torch.Size([1, 1, 202, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,208B, BPFP=4.5903 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,304B, BPFP=2.5210 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,508B, BPFP=8.9131 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,252B, BPFP=4.8982 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,088B, BPFP=10.5195 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,384B, BPFP=3.6245 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,392B, BPFP=10.3496 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,212B, BPFP=3.1868 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,132B, BPFP=9.0654 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 254,272B, BPFP=2.4585 ⌛️ [2/4] FRONTEND: Frontend time: 2.645s (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, 202, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.379s [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, 202, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004093 layer.1.conv_state 0.00049625 0.22444603 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00013993 0.04404069 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00012044 0.03754063 layer.4.ssm_state 0.00000001 0.00000472 layer.4.conv_state 0.00032912 0.08409700 layer.4.output 0.00000189 0.00014714 ------------------------------------------------------------------------------------- TOTAL 0.00002547 0.00942370 (elements=1,646,592) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1646592 Total Bytes 758896 BPFP 3.6871 bits/point EBPFP 6.1388 equivalent bits/point MSE 0.009424 ---------------------- -------------------------------------------------------- Time: 5.034s Load: 0.011s, Pack+Encode: 2.645s, Decode+Unpack: 2.379s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 202, 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 0.0094 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample18-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample18-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample19-layer4-item1.zst (38/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample19-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 198, 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, 198, 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, 198, 4096]) -> torch.Size([1, 1, 198, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,056B, BPFP=4.7031 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,416B, BPFP=2.5278 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,452B, BPFP=8.8994 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,460B, BPFP=4.9109 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,172B, BPFP=10.5400 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,688B, BPFP=3.6431 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,536B, BPFP=10.3848 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,680B, BPFP=3.2153 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,168B, BPFP=9.0742 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 251,560B, BPFP=2.4815 ⌛️ [2/4] FRONTEND: Frontend time: 2.635s (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, 198, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.373s [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, 198, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004215 layer.1.conv_state 0.00051006 0.22838213 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00016659 0.04332852 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00011300 0.03753003 layer.4.ssm_state 0.00000001 0.00000437 layer.4.conv_state 0.00024382 0.08818910 layer.4.output 0.00000089 0.00014691 ------------------------------------------------------------------------------------- TOTAL 0.00002416 0.00966373 (elements=1,630,208) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1630208 Total Bytes 759332 BPFP 3.7263 bits/point EBPFP 6.2181 equivalent bits/point MSE 0.009664 ---------------------- -------------------------------------------------------- Time: 5.018s Load: 0.010s, Pack+Encode: 2.635s, Decode+Unpack: 2.373s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 198, 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 0.0097 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample19-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample19-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample2-layer4-item1.zst (39/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample2-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 251, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 251, 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, 251, 4096]) -> torch.Size([1, 1, 251, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,336B, BPFP=4.6592 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,244B, BPFP=2.5173 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,556B, BPFP=8.9248 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,748B, BPFP=4.8064 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,168B, BPFP=10.5391 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,952B, BPFP=3.5981 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,488B, BPFP=10.3730 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,400B, BPFP=3.1982 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,312B, BPFP=9.1094 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 287,996B, BPFP=2.2410 ⌛️ [2/4] FRONTEND: Frontend time: 2.645s (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, 251, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.376s [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, 251, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000011 0.00003955 layer.1.conv_state 0.00050136 0.22741702 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00017187 0.04361937 layer.3.ssm_state 0.00000001 0.00000263 layer.3.conv_state 0.00011363 0.03818852 layer.4.ssm_state 0.00000001 0.00000454 layer.4.conv_state 0.00027333 0.08582950 layer.4.output 0.00000146 0.00012844 ------------------------------------------------------------------------------------- TOTAL 0.00002222 0.00849275 (elements=1,847,296) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1847296 Total Bytes 792344 BPFP 3.4314 bits/point EBPFP 5.6155 equivalent bits/point MSE 0.008493 ---------------------- -------------------------------------------------------- Time: 5.033s Load: 0.011s, Pack+Encode: 2.645s, Decode+Unpack: 2.376s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 251, 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 0.0085 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample2-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample2-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample21-layer4-item1.zst (40/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample21-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 200, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 200, 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, 200, 4096]) -> torch.Size([1, 1, 200, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,720B, BPFP=4.6216 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,128B, BPFP=2.5103 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,488B, BPFP=8.9082 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,544B, BPFP=4.9160 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,332B, BPFP=10.5791 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,268B, BPFP=3.6174 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,500B, BPFP=10.3760 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,632B, BPFP=3.1514 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,256B, BPFP=9.0957 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 258,588B, BPFP=2.5253 ⌛️ [2/4] FRONTEND: Frontend time: 2.630s (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, 200, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.384s [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, 200, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004027 layer.1.conv_state 0.00050167 0.22510716 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00014870 0.04433564 layer.3.ssm_state 0.00000001 0.00000270 layer.3.conv_state 0.00007362 0.03783605 layer.4.ssm_state 0.00000001 0.00000488 layer.4.conv_state 0.00028082 0.08634137 layer.4.output 0.00000199 0.00015663 ------------------------------------------------------------------------------------- TOTAL 0.00002402 0.00954470 (elements=1,638,400) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1638400 Total Bytes 763600 BPFP 3.7285 bits/point EBPFP 6.1944 equivalent bits/point MSE 0.009545 ---------------------- -------------------------------------------------------- Time: 5.027s Load: 0.013s, Pack+Encode: 2.630s, Decode+Unpack: 2.384s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 200, 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 0.0095 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample21-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample21-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample23-layer4-item1.zst (41/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample23-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 197, 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, 197, 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, 197, 4096]) -> torch.Size([1, 1, 197, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,168B, BPFP=4.5879 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,224B, BPFP=2.5161 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,464B, BPFP=8.9023 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,212B, BPFP=4.8347 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,136B, BPFP=10.5312 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,220B, BPFP=3.6145 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,520B, BPFP=10.3809 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,284B, BPFP=3.1912 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,348B, BPFP=9.1182 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 250,976B, BPFP=2.4883 ⌛️ [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, 197, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.375s [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, 197, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000018 0.00004024 layer.1.conv_state 0.00050137 0.22367963 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00011144 0.04449563 layer.3.ssm_state 0.00000001 0.00000259 layer.3.conv_state 0.00011790 0.03739240 layer.4.ssm_state 0.00000003 0.00000474 layer.4.conv_state 0.00023723 0.08776417 layer.4.output 0.00000089 0.00015751 ------------------------------------------------------------------------------------- TOTAL 0.00002291 0.00961025 (elements=1,626,112) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1626112 Total Bytes 754696 BPFP 3.7129 bits/point EBPFP 6.1910 equivalent bits/point MSE 0.009610 ---------------------- -------------------------------------------------------- Time: 5.021s Load: 0.010s, Pack+Encode: 2.636s, Decode+Unpack: 2.375s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 197, 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 0.0096 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample23-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample23-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample24-layer4-item1.zst (42/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample24-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 194, 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, 194, 4096]) -> torch.Size([1, 1, 194, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,464B, BPFP=4.6670 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,172B, BPFP=2.5129 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,568B, BPFP=8.9277 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,680B, BPFP=4.8022 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,332B, BPFP=10.5791 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,688B, BPFP=3.6431 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,644B, BPFP=10.4111 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,092B, BPFP=3.1794 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,460B, BPFP=9.1455 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 251,240B, BPFP=2.5294 ⌛️ [2/4] FRONTEND: Frontend time: 2.633s (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, 194, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.377s [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, 194, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004107 layer.1.conv_state 0.00049004 0.22712000 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00013359 0.04417433 layer.3.ssm_state 0.00000001 0.00000287 layer.3.conv_state 0.00007909 0.03824954 layer.4.ssm_state 0.00000004 0.00000468 layer.4.conv_state 0.00029573 0.08919131 layer.4.output 0.00000091 0.00014894 ------------------------------------------------------------------------------------- TOTAL 0.00002370 0.00978780 (elements=1,613,824) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1613824 Total Bytes 756484 BPFP 3.7500 bits/point EBPFP 6.2546 equivalent bits/point MSE 0.009788 ---------------------- -------------------------------------------------------- Time: 5.022s Load: 0.011s, Pack+Encode: 2.633s, Decode+Unpack: 2.377s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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 0.0098 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample24-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample24-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample25-layer4-item1.zst (43/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample25-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 204, 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, 204, 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, 204, 4096]) -> torch.Size([1, 1, 204, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,072B, BPFP=4.7041 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,024B, BPFP=2.5039 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,548B, BPFP=8.9229 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,608B, BPFP=4.8589 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,332B, BPFP=10.5791 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,408B, BPFP=3.6260 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,612B, BPFP=10.4033 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,608B, BPFP=3.2109 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,420B, BPFP=9.1357 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 266,844B, BPFP=2.5548 ⌛️ [2/4] FRONTEND: Frontend time: 2.629s (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, 204, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.394s [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, 204, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004094 layer.1.conv_state 0.00049055 0.22616565 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00016096 0.04390229 layer.3.ssm_state 0.00000001 0.00000274 layer.3.conv_state 0.00007367 0.03807574 layer.4.ssm_state 0.00000001 0.00000456 layer.4.conv_state 0.00021167 0.08571069 layer.4.output 0.00000088 0.00017402 ------------------------------------------------------------------------------------- TOTAL 0.00002190 0.00946520 (elements=1,654,784) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1654784 Total Bytes 773620 BPFP 3.7400 bits/point EBPFP 6.1900 equivalent bits/point MSE 0.009465 ---------------------- -------------------------------------------------------- Time: 5.033s Load: 0.010s, Pack+Encode: 2.629s, Decode+Unpack: 2.394s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 204, 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 0.0095 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample25-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample25-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample26-layer4-item1.zst (44/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample26-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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, 194, 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, 194, 4096]) -> torch.Size([1, 1, 194, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,872B, BPFP=4.6309 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,292B, BPFP=2.5203 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,772B, BPFP=4.8689 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,376B, BPFP=10.5898 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,488B, BPFP=3.6309 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,740B, BPFP=10.4346 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,300B, BPFP=3.1921 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,412B, BPFP=9.1338 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 255,944B, BPFP=2.5768 ⌛️ [2/4] FRONTEND: Frontend time: 2.634s (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, 194, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.380s [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, 194, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000023 0.00004278 layer.1.conv_state 0.00050992 0.22874403 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014380 0.04526648 layer.3.ssm_state 0.00000001 0.00000266 layer.3.conv_state 0.00008062 0.03823240 layer.4.ssm_state 0.00000004 0.00000447 layer.4.conv_state 0.00031562 0.08847301 layer.4.output 0.00000094 0.00015530 ------------------------------------------------------------------------------------- TOTAL 0.00002477 0.00983126 (elements=1,613,824) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1613824 Total Bytes 761864 BPFP 3.7767 bits/point EBPFP 6.2846 equivalent bits/point MSE 0.009831 ---------------------- -------------------------------------------------------- Time: 5.024s Load: 0.010s, Pack+Encode: 2.634s, Decode+Unpack: 2.380s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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 0.0098 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample26-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample26-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample27-layer4-item1.zst (45/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample27-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 191, 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, 191, 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, 191, 4096]) -> torch.Size([1, 1, 191, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,456B, BPFP=4.6055 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,168B, BPFP=2.5127 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,540B, BPFP=8.9209 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,160B, BPFP=4.8315 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,240B, BPFP=10.5566 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,204B, BPFP=3.6135 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,532B, BPFP=10.3838 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,836B, BPFP=3.2249 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,292B, BPFP=9.1045 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,772B, BPFP=2.2269 ⌛️ [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, 191, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 191, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000011 0.00004022 layer.1.conv_state 0.00049123 0.22795744 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00015211 0.04363123 layer.3.ssm_state 0.00000001 0.00000274 layer.3.conv_state 0.00012051 0.03728300 layer.4.ssm_state 0.00000003 0.00000442 layer.4.conv_state 0.00022142 0.08918255 layer.4.output 0.00000203 0.00017806 ------------------------------------------------------------------------------------- TOTAL 0.00002416 0.00986195 (elements=1,601,536) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1601536 Total Bytes 722344 BPFP 3.6083 bits/point EBPFP 6.1287 equivalent bits/point MSE 0.009862 ---------------------- -------------------------------------------------------- Time: 4.934s Load: 0.010s, Pack+Encode: 2.600s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 191, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample27-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample27-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample29-layer4-item1.zst (46/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample29-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 195, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 195, 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, 195, 4096]) -> torch.Size([1, 1, 195, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,512B, BPFP=4.6699 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,320B, BPFP=2.5220 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,488B, BPFP=8.9082 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,580B, BPFP=4.9182 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,232B, BPFP=10.5547 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,360B, BPFP=3.6230 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,472B, BPFP=10.3691 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,664B, BPFP=3.1533 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,168B, BPFP=9.0742 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 252,596B, BPFP=2.5300 ⌛️ [2/4] FRONTEND: Frontend time: 2.655s (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, 195, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.367s [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, 195, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00004215 layer.1.conv_state 0.00050474 0.22593927 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014103 0.04423062 layer.3.ssm_state 0.00000001 0.00000270 layer.3.conv_state 0.00007151 0.03732691 layer.4.ssm_state 0.00000001 0.00000449 layer.4.conv_state 0.00026369 0.08570183 layer.4.output 0.00000092 0.00014646 ------------------------------------------------------------------------------------- TOTAL 0.00002331 0.00965010 (elements=1,617,920) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1617920 Total Bytes 758536 BPFP 3.7507 bits/point EBPFP 6.2524 equivalent bits/point MSE 0.009650 ---------------------- -------------------------------------------------------- Time: 5.036s Load: 0.013s, Pack+Encode: 2.655s, Decode+Unpack: 2.367s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 195, 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 0.0097 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample29-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample29-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample30-layer4-item1.zst (47/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample30-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 189, 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, 189, 4096]) -> torch.Size([1, 1, 189, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,936B, BPFP=4.6348 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,216B, BPFP=2.5156 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,432B, BPFP=8.8945 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,016B, BPFP=4.8838 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,180B, BPFP=10.5420 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,348B, BPFP=3.6223 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,492B, BPFP=10.3740 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,860B, BPFP=3.2263 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,184B, BPFP=9.0781 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,300B, BPFP=2.2456 ⌛️ [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, 189, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.322s [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, 189, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000021 0.00003993 layer.1.conv_state 0.00049016 0.22833748 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00013810 0.04362129 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00011147 0.03742174 layer.4.ssm_state 0.00000002 0.00000433 layer.4.conv_state 0.00026114 0.08358934 layer.4.output 0.00000090 0.00019448 ------------------------------------------------------------------------------------- TOTAL 0.00002405 0.00981512 (elements=1,593,344) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1593344 Total Bytes 723108 BPFP 3.6306 bits/point EBPFP 6.1702 equivalent bits/point MSE 0.009815 ---------------------- -------------------------------------------------------- Time: 4.919s Load: 0.011s, Pack+Encode: 2.586s, Decode+Unpack: 2.322s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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 0.0098 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample30-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample30-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample31-layer4-item1.zst (48/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample31-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 193, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 193, 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, 193, 4096]) -> torch.Size([1, 1, 193, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,912B, BPFP=4.6333 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,212B, BPFP=2.5154 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,468B, BPFP=8.9033 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,912B, BPFP=4.8164 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,228B, BPFP=10.5537 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,788B, BPFP=3.5881 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,580B, BPFP=10.3955 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,876B, BPFP=3.2273 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,276B, BPFP=9.1006 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 245,832B, BPFP=2.4878 ⌛️ [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, 193, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.371s [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, 193, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000026 0.00004211 layer.1.conv_state 0.00052004 0.22598653 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00014284 0.04333537 layer.3.ssm_state 0.00000001 0.00000265 layer.3.conv_state 0.00011856 0.03730642 layer.4.ssm_state 0.00000002 0.00000430 layer.4.conv_state 0.00031680 0.08373593 layer.4.output 0.00000090 0.00015291 ------------------------------------------------------------------------------------- TOTAL 0.00002580 0.00964391 (elements=1,609,728) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1609728 Total Bytes 750228 BPFP 3.7285 bits/point EBPFP 6.2352 equivalent bits/point MSE 0.009644 ---------------------- -------------------------------------------------------- Time: 5.036s Load: 0.011s, Pack+Encode: 2.654s, Decode+Unpack: 2.371s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 193, 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 0.0096 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample31-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample31-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample33-layer4-item1.zst (49/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample33-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 205, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 205, 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, 205, 4096]) -> torch.Size([1, 1, 205, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,056B, BPFP=4.7031 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,120B, BPFP=2.5098 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,492B, BPFP=8.9092 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,164B, BPFP=4.8928 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,236B, BPFP=10.5557 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,796B, BPFP=3.6497 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,548B, BPFP=10.3877 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,732B, BPFP=3.1575 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,312B, BPFP=9.1094 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 270,332B, BPFP=2.5756 ⌛️ [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, 205, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.374s [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, 205, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00003998 layer.1.conv_state 0.00051517 0.22820874 layer.2.ssm_state 0.00000001 0.00000220 layer.2.conv_state 0.00014686 0.04401007 layer.3.ssm_state 0.00000001 0.00000281 layer.3.conv_state 0.00007506 0.03796390 layer.4.ssm_state 0.00000004 0.00000483 layer.4.conv_state 0.00029211 0.09033328 layer.4.output 0.00000087 0.00017631 ------------------------------------------------------------------------------------- TOTAL 0.00002367 0.00957496 (elements=1,658,880) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1658880 Total Bytes 776932 BPFP 3.7468 bits/point EBPFP 6.1899 equivalent bits/point MSE 0.009575 ---------------------- -------------------------------------------------------- Time: 5.029s Load: 0.013s, Pack+Encode: 2.643s, Decode+Unpack: 2.374s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 205, 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 0.0096 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample33-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample33-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample34-layer4-item1.zst (50/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample34-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 181, 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, 181, 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, 181, 4096]) -> torch.Size([1, 1, 181, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,512B, BPFP=4.6089 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,096B, BPFP=2.5083 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,548B, BPFP=8.9229 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,540B, BPFP=4.7937 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,276B, BPFP=10.5654 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,524B, BPFP=3.6331 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,708B, BPFP=10.4268 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,836B, BPFP=3.2249 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,428B, BPFP=9.1377 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 220,220B, BPFP=2.3763 ⌛️ [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, 181, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.315s [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, 181, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00003948 layer.1.conv_state 0.00050471 0.22688836 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00013702 0.04399802 layer.3.ssm_state 0.00000001 0.00000270 layer.3.conv_state 0.00006847 0.03836158 layer.4.ssm_state 0.00000002 0.00000422 layer.4.conv_state 0.00022775 0.08360817 layer.4.output 0.00000093 0.00017794 ------------------------------------------------------------------------------------- TOTAL 0.00002322 0.01000683 (elements=1,560,576) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1560576 Total Bytes 724832 BPFP 3.7157 bits/point EBPFP 6.3025 equivalent bits/point MSE 0.010007 ---------------------- -------------------------------------------------------- Time: 4.898s Load: 0.009s, Pack+Encode: 2.574s, Decode+Unpack: 2.315s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 181, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample34-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample34-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample35-layer4-item1.zst (51/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample35-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 202, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 202, 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, 202, 4096]) -> torch.Size([1, 1, 202, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,104B, BPFP=4.7061 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,024B, BPFP=2.5039 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,452B, BPFP=8.8994 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,008B, BPFP=4.8223 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,180B, BPFP=10.5420 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,204B, BPFP=3.6135 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,608B, BPFP=10.4023 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,124B, BPFP=3.1814 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,460B, BPFP=9.1455 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 250,616B, BPFP=2.4232 ⌛️ [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, 202, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.375s [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, 202, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000030 0.00004054 layer.1.conv_state 0.00049950 0.22337906 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00017947 0.04337948 layer.3.ssm_state 0.00000001 0.00000280 layer.3.conv_state 0.00011997 0.03809360 layer.4.ssm_state 0.00000004 0.00000477 layer.4.conv_state 0.00025113 0.08912598 layer.4.output 0.00000086 0.00014295 ------------------------------------------------------------------------------------- TOTAL 0.00002427 0.00949826 (elements=1,646,592) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1646592 Total Bytes 755924 BPFP 3.6727 bits/point EBPFP 6.1277 equivalent bits/point MSE 0.009498 ---------------------- -------------------------------------------------------- Time: 5.023s Load: 0.011s, Pack+Encode: 2.637s, Decode+Unpack: 2.375s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 202, 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 0.0095 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample35-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample35-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample37-layer4-item1.zst (52/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample37-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 177, 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, 177, 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, 177, 4096]) -> torch.Size([1, 1, 177, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,608B, BPFP=4.6147 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,100B, BPFP=2.5085 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,608B, BPFP=8.9375 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 77,704B, BPFP=4.7427 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,256B, BPFP=10.5605 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,964B, BPFP=3.5989 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,600B, BPFP=10.4004 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,120B, BPFP=3.1812 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,336B, BPFP=9.1152 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 223,436B, BPFP=2.4655 ⌛️ [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, 177, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.313s [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, 177, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004060 layer.1.conv_state 0.00049719 0.22686005 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00011773 0.04460310 layer.3.ssm_state 0.00000001 0.00000269 layer.3.conv_state 0.00012253 0.03784195 layer.4.ssm_state 0.00000003 0.00000476 layer.4.conv_state 0.00028026 0.08997066 layer.4.output 0.00000097 0.00018199 ------------------------------------------------------------------------------------- TOTAL 0.00002516 0.01024938 (elements=1,544,192) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1544192 Total Bytes 725876 BPFP 3.7605 bits/point EBPFP 6.3635 equivalent bits/point MSE 0.010249 ---------------------- -------------------------------------------------------- Time: 4.898s Load: 0.012s, Pack+Encode: 2.574s, Decode+Unpack: 2.313s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 177, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample37-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample37-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample38-layer4-item1.zst (53/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample38-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 187, 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, 187, 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, 187, 4096]) -> torch.Size([1, 1, 187, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,192B, BPFP=4.6504 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,108B, BPFP=2.5090 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,512B, BPFP=8.9141 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,968B, BPFP=4.9419 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,300B, BPFP=10.5713 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,552B, BPFP=3.6348 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,528B, BPFP=10.3828 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,348B, BPFP=3.1951 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,164B, BPFP=9.0732 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,608B, BPFP=2.2937 ⌛️ [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, 187, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.308s [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, 187, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000139 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00004070 layer.1.conv_state 0.00050091 0.22642910 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00015339 0.04371741 layer.3.ssm_state 0.00000001 0.00000275 layer.3.conv_state 0.00006961 0.03733499 layer.4.ssm_state 0.00000001 0.00000471 layer.4.conv_state 0.00023327 0.08700181 layer.4.output 0.00000206 0.00015763 ------------------------------------------------------------------------------------- TOTAL 0.00002382 0.00987841 (elements=1,585,152) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1585152 Total Bytes 726424 BPFP 3.6661 bits/point EBPFP 6.2240 equivalent bits/point MSE 0.009878 ---------------------- -------------------------------------------------------- Time: 4.885s Load: 0.012s, Pack+Encode: 2.565s, Decode+Unpack: 2.308s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 187, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample38-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample38-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample39-layer4-item1.zst (54/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample39-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 190, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 190, 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, 190, 4096]) -> torch.Size([1, 1, 190, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,264B, BPFP=4.5938 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,164B, BPFP=2.5125 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,532B, BPFP=8.9189 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,704B, BPFP=4.8647 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,164B, BPFP=10.5381 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,684B, BPFP=3.6428 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,512B, BPFP=10.3789 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,688B, BPFP=3.2158 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,264B, BPFP=9.0977 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,496B, BPFP=2.2358 ⌛️ [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, 190, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.325s [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, 190, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000043 0.00003934 layer.1.conv_state 0.00052660 0.22707197 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00015850 0.04371934 layer.3.ssm_state 0.00000001 0.00000273 layer.3.conv_state 0.00011238 0.03770189 layer.4.ssm_state 0.00000001 0.00000457 layer.4.conv_state 0.00023404 0.08534465 layer.4.output 0.00000093 0.00020073 ------------------------------------------------------------------------------------- TOTAL 0.00002465 0.00981127 (elements=1,597,440) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1597440 Total Bytes 722616 BPFP 3.6189 bits/point EBPFP 6.1485 equivalent bits/point MSE 0.009811 ---------------------- -------------------------------------------------------- Time: 4.937s Load: 0.013s, Pack+Encode: 2.599s, Decode+Unpack: 2.325s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 190, 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 0.0098 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample39-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample39-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample40-layer4-item1.zst (55/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample40-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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, 184, 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, 184, 4096]) -> torch.Size([1, 1, 184, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,180B, BPFP=4.6497 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,176B, BPFP=2.5132 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,520B, BPFP=8.9160 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,468B, BPFP=4.9114 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,168B, BPFP=10.5391 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,196B, BPFP=3.6130 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,456B, BPFP=10.3652 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,528B, BPFP=3.2061 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,252B, BPFP=9.0947 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,024B, BPFP=2.3461 ⌛️ [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, 184, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.313s [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, 184, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004079 layer.1.conv_state 0.00050506 0.22810736 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012367 0.04367860 layer.3.ssm_state 0.00000001 0.00000277 layer.3.conv_state 0.00011805 0.03744505 layer.4.ssm_state 0.00000004 0.00000470 layer.4.conv_state 0.00025882 0.08713629 layer.4.output 0.00000217 0.00016770 ------------------------------------------------------------------------------------- TOTAL 0.00002505 0.00999844 (elements=1,572,864) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1572864 Total Bytes 727112 BPFP 3.6983 bits/point EBPFP 6.2724 equivalent bits/point MSE 0.009998 ---------------------- -------------------------------------------------------- Time: 4.910s Load: 0.009s, Pack+Encode: 2.589s, Decode+Unpack: 2.313s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample40-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample40-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample41-layer4-item1.zst (56/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample41-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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, 189, 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, 189, 4096]) -> torch.Size([1, 1, 189, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,816B, BPFP=4.6274 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,256B, BPFP=2.5181 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,512B, BPFP=8.9141 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,144B, BPFP=4.8916 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,188B, BPFP=10.5439 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,400B, BPFP=3.6255 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,532B, BPFP=10.3838 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,932B, BPFP=3.1697 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,232B, BPFP=9.0898 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,148B, BPFP=2.2337 ⌛️ [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, 189, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 189, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000015 0.00004163 layer.1.conv_state 0.00050469 0.22563805 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00017072 0.04450596 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00007287 0.03729682 layer.4.ssm_state 0.00000004 0.00000498 layer.4.conv_state 0.00028764 0.09060942 layer.4.output 0.00000205 0.00018951 ------------------------------------------------------------------------------------- TOTAL 0.00002532 0.00991737 (elements=1,593,344) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1593344 Total Bytes 721304 BPFP 3.6216 bits/point EBPFP 6.1579 equivalent bits/point MSE 0.009917 ---------------------- -------------------------------------------------------- Time: 4.916s Load: 0.012s, Pack+Encode: 2.579s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 189, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample41-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample41-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample42-layer4-item1.zst (57/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample42-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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, 165, 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, 165, 4096]) -> torch.Size([1, 1, 165, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,068B, BPFP=4.6428 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,284B, BPFP=2.5198 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,596B, BPFP=8.9346 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,876B, BPFP=4.8142 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,272B, BPFP=10.5645 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,548B, BPFP=3.6345 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,608B, BPFP=10.4023 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,720B, BPFP=3.1567 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,160B, BPFP=9.0723 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,120B, BPFP=2.6174 ⌛️ [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, 165, 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, 165, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004177 layer.1.conv_state 0.00049940 0.23062289 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00014923 0.04504321 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00007577 0.03807978 layer.4.ssm_state 0.00000001 0.00000459 layer.4.conv_state 0.00024214 0.08608007 layer.4.output 0.00000104 0.00018851 ------------------------------------------------------------------------------------- TOTAL 0.00002487 0.01059546 (elements=1,495,040) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1495040 Total Bytes 725396 BPFP 3.8816 bits/point EBPFP 6.5800 equivalent bits/point MSE 0.010595 ---------------------- -------------------------------------------------------- Time: 4.935s Load: 0.012s, Pack+Encode: 2.589s, Decode+Unpack: 2.335s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample42-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample42-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample44-layer4-item1.zst (58/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample44-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 188, 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, 188, 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, 188, 4096]) -> torch.Size([1, 1, 188, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,016B, BPFP=4.6396 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 40,996B, BPFP=2.5022 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,604B, BPFP=4.8586 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,284B, BPFP=10.5674 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,852B, BPFP=3.5920 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,480B, BPFP=10.3711 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,728B, BPFP=3.2183 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,372B, BPFP=9.1240 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,496B, BPFP=2.2699 ⌛️ [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, 188, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.310s [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, 188, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00004027 layer.1.conv_state 0.00050713 0.23003063 layer.2.ssm_state 0.00000001 0.00000213 layer.2.conv_state 0.00014305 0.04357840 layer.3.ssm_state 0.00000001 0.00000274 layer.3.conv_state 0.00011656 0.03737283 layer.4.ssm_state 0.00000003 0.00000436 layer.4.conv_state 0.00022655 0.08690938 layer.4.output 0.00000092 0.00018194 ------------------------------------------------------------------------------------- TOTAL 0.00002396 0.00993534 (elements=1,589,248) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1589248 Total Bytes 723452 BPFP 3.6417 bits/point EBPFP 6.1836 equivalent bits/point MSE 0.009935 ---------------------- -------------------------------------------------------- Time: 4.907s Load: 0.010s, Pack+Encode: 2.587s, Decode+Unpack: 2.310s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 188, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample44-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample44-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample45-layer4-item1.zst (59/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample45-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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, 176, 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, 176, 4096]) -> torch.Size([1, 1, 176, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,380B, BPFP=4.6619 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 40,708B, BPFP=2.4846 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,572B, BPFP=8.9287 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 77,040B, BPFP=4.7021 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,188B, BPFP=10.5439 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,752B, BPFP=3.5859 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,588B, BPFP=10.3975 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,684B, BPFP=3.2766 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,548B, BPFP=9.1670 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,888B, BPFP=2.4180 ⌛️ [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, 176, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 176, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000060 0.00004128 layer.1.conv_state 0.00049474 0.22853181 layer.2.ssm_state 0.00000001 0.00000210 layer.2.conv_state 0.00013052 0.04446343 layer.3.ssm_state 0.00000001 0.00000282 layer.3.conv_state 0.00011507 0.03849719 layer.4.ssm_state 0.00000007 0.00000418 layer.4.conv_state 0.00021714 0.08452472 layer.4.output 0.00000218 0.00018240 ------------------------------------------------------------------------------------- TOTAL 0.00002456 0.01020703 (elements=1,540,096) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1540096 Total Bytes 721492 BPFP 3.7478 bits/point EBPFP 6.3637 equivalent bits/point MSE 0.010207 ---------------------- -------------------------------------------------------- Time: 4.903s Load: 0.009s, Pack+Encode: 2.581s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample45-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample45-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample47-layer4-item1.zst (60/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample47-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 182, 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, 182, 4096]) -> torch.Size([1, 1, 182, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,096B, BPFP=4.7056 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,284B, BPFP=2.5198 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,928B, BPFP=4.8784 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,168B, BPFP=10.5391 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,992B, BPFP=3.6616 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,576B, BPFP=10.3945 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,756B, BPFP=3.2200 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,256B, BPFP=9.0957 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,872B, BPFP=2.3274 ⌛️ [2/4] FRONTEND: Frontend time: 2.585s (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, 182, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 182, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000043 0.00003989 layer.1.conv_state 0.00049140 0.22821158 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00015338 0.04321465 layer.3.ssm_state 0.00000001 0.00000284 layer.3.conv_state 0.00011757 0.03799587 layer.4.ssm_state 0.00000004 0.00000455 layer.4.conv_state 0.00022214 0.08622988 layer.4.output 0.00000093 0.00015911 ------------------------------------------------------------------------------------- TOTAL 0.00002416 0.01003075 (elements=1,564,672) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1564672 Total Bytes 724552 BPFP 3.7046 bits/point EBPFP 6.3003 equivalent bits/point MSE 0.010031 ---------------------- -------------------------------------------------------- Time: 4.920s Load: 0.011s, Pack+Encode: 2.585s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample47-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample47-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample48-layer4-item1.zst (61/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample48-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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, 184, 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, 184, 4096]) -> torch.Size([1, 1, 184, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,988B, BPFP=4.6379 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,280B, BPFP=2.5195 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,528B, BPFP=8.9180 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,832B, BPFP=4.9336 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,360B, BPFP=10.5859 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,300B, BPFP=3.6194 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,656B, BPFP=10.4141 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,112B, BPFP=3.1807 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,296B, BPFP=9.1055 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,724B, BPFP=2.3642 ⌛️ [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, 184, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 184, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000011 0.00004049 layer.1.conv_state 0.00051222 0.22779804 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00016367 0.04359982 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00007453 0.03769272 layer.4.ssm_state 0.00000001 0.00000479 layer.4.conv_state 0.00030342 0.08919907 layer.4.output 0.00000098 0.00016962 ------------------------------------------------------------------------------------- TOTAL 0.00002548 0.01003939 (elements=1,572,864) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1572864 Total Bytes 729220 BPFP 3.7090 bits/point EBPFP 6.2852 equivalent bits/point MSE 0.010039 ---------------------- -------------------------------------------------------- Time: 4.925s Load: 0.010s, Pack+Encode: 2.597s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample48-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample48-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample49-layer4-item1.zst (62/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample49-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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, 180, 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, 180, 4096]) -> torch.Size([1, 1, 180, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,948B, BPFP=4.6355 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,160B, BPFP=2.5122 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,120B, BPFP=4.7681 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,156B, BPFP=10.5361 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,664B, BPFP=3.6416 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,480B, BPFP=10.3711 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,056B, BPFP=3.1772 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,172B, BPFP=9.0752 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,132B, BPFP=2.3777 ⌛️ [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, 180, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.321s [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, 180, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000026 0.00004142 layer.1.conv_state 0.00049255 0.22726324 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00013914 0.04381075 layer.3.ssm_state 0.00000001 0.00000270 layer.3.conv_state 0.00011440 0.03798703 layer.4.ssm_state 0.00000001 0.00000476 layer.4.conv_state 0.00024451 0.08753417 layer.4.output 0.00000214 0.00017366 ------------------------------------------------------------------------------------- TOTAL 0.00002497 0.01010959 (elements=1,556,480) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1556480 Total Bytes 722516 BPFP 3.7136 bits/point EBPFP 6.3009 equivalent bits/point MSE 0.010110 ---------------------- -------------------------------------------------------- Time: 4.935s Load: 0.009s, Pack+Encode: 2.604s, Decode+Unpack: 2.321s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample49-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample49-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample5-layer4-item1.zst (63/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample5-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 231, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 231, 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, 231, 4096]) -> torch.Size([1, 1, 231, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,324B, BPFP=4.5974 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,136B, BPFP=2.5107 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,464B, BPFP=8.9023 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,992B, BPFP=4.8823 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,096B, BPFP=10.5215 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,108B, BPFP=3.6077 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,544B, BPFP=10.3867 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,100B, BPFP=3.2410 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,312B, BPFP=9.1094 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 285,480B, BPFP=2.4138 ⌛️ [2/4] FRONTEND: Frontend time: 2.630s (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, 231, 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, 231, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00004101 layer.1.conv_state 0.00049501 0.22635341 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012390 0.04367210 layer.3.ssm_state 0.00000001 0.00000273 layer.3.conv_state 0.00007139 0.03709926 layer.4.ssm_state 0.00000005 0.00000431 layer.4.conv_state 0.00023197 0.08613385 layer.4.output 0.00000151 0.00013462 ------------------------------------------------------------------------------------- TOTAL 0.00002066 0.00885097 (elements=1,765,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1765376 Total Bytes 790700 BPFP 3.5831 bits/point EBPFP 5.8726 equivalent bits/point MSE 0.008851 ---------------------- -------------------------------------------------------- Time: 5.027s Load: 0.013s, Pack+Encode: 2.630s, Decode+Unpack: 2.385s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 231, 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 0.0089 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample5-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample5-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample50-layer4-item1.zst (64/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample50-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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, 182, 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, 182, 4096]) -> torch.Size([1, 1, 182, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,192B, BPFP=4.6504 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,276B, BPFP=2.5193 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,496B, BPFP=8.9102 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,152B, BPFP=4.9531 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,220B, BPFP=10.5518 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,448B, BPFP=3.6284 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,604B, BPFP=10.4014 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,628B, BPFP=3.1511 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,364B, BPFP=9.1221 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,032B, BPFP=2.3505 ⌛️ [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, 182, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 182, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004107 layer.1.conv_state 0.00049898 0.22640064 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00015473 0.04360270 layer.3.ssm_state 0.00000001 0.00000266 layer.3.conv_state 0.00007806 0.03806149 layer.4.ssm_state 0.00000005 0.00000517 layer.4.conv_state 0.00037297 0.09335004 layer.4.output 0.00000094 0.00016460 ------------------------------------------------------------------------------------- TOTAL 0.00002666 0.01015419 (elements=1,564,672) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1564672 Total Bytes 725556 BPFP 3.7097 bits/point EBPFP 6.2995 equivalent bits/point MSE 0.010154 ---------------------- -------------------------------------------------------- Time: 4.910s Load: 0.009s, Pack+Encode: 2.581s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample50-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample50-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample51-layer4-item1.zst (65/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample51-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 195, 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, 195, 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, 195, 4096]) -> torch.Size([1, 1, 195, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,116B, BPFP=4.6458 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,152B, BPFP=2.5117 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,488B, BPFP=8.9082 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,704B, BPFP=4.9868 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,284B, BPFP=10.5674 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,432B, BPFP=3.6274 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,588B, BPFP=10.3975 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,140B, BPFP=3.1824 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,232B, BPFP=9.0898 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 258,072B, BPFP=2.5849 ⌛️ [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, 195, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.376s [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, 195, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000025 0.00004103 layer.1.conv_state 0.00051868 0.22863117 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00013950 0.04406607 layer.3.ssm_state 0.00000001 0.00000278 layer.3.conv_state 0.00011491 0.03775459 layer.4.ssm_state 0.00000004 0.00000473 layer.4.conv_state 0.00031385 0.08666397 layer.4.output 0.00000197 0.00015037 ------------------------------------------------------------------------------------- TOTAL 0.00002597 0.00973130 (elements=1,617,920) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1617920 Total Bytes 765352 BPFP 3.7844 bits/point EBPFP 6.2927 equivalent bits/point MSE 0.009731 ---------------------- -------------------------------------------------------- Time: 5.016s Load: 0.012s, Pack+Encode: 2.628s, Decode+Unpack: 2.376s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 195, 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 0.0097 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample51-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample51-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample52-layer4-item1.zst (66/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample52-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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, 179, 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, 179, 4096]) -> torch.Size([1, 1, 179, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,308B, BPFP=4.6575 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,292B, BPFP=2.5203 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,576B, BPFP=8.9297 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,460B, BPFP=4.8499 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,224B, BPFP=10.5527 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,324B, BPFP=3.6208 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,524B, BPFP=10.3818 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,904B, BPFP=3.1680 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,176B, BPFP=9.0762 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,500B, BPFP=2.4169 ⌛️ [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, 179, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.318s [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, 179, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000011 0.00004021 layer.1.conv_state 0.00049627 0.22472034 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00016598 0.04391602 layer.3.ssm_state 0.00000001 0.00000260 layer.3.conv_state 0.00011926 0.03770894 layer.4.ssm_state 0.00000004 0.00000473 layer.4.conv_state 0.00026642 0.08675953 layer.4.output 0.00000098 0.00016653 ------------------------------------------------------------------------------------- TOTAL 0.00002568 0.01005865 (elements=1,552,384) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1552384 Total Bytes 726432 BPFP 3.7436 bits/point EBPFP 6.3457 equivalent bits/point MSE 0.010059 ---------------------- -------------------------------------------------------- Time: 4.913s Load: 0.009s, Pack+Encode: 2.586s, Decode+Unpack: 2.318s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample52-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample52-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample53-layer4-item1.zst (67/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample53-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 190, 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, 190, 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, 190, 4096]) -> torch.Size([1, 1, 190, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,640B, BPFP=4.6777 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,220B, BPFP=2.5159 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,960B, BPFP=4.9414 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,356B, BPFP=10.5850 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,408B, BPFP=3.6260 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,552B, BPFP=10.3887 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,856B, BPFP=3.1650 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,344B, BPFP=9.1172 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 226,036B, BPFP=2.3236 ⌛️ [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, 190, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 190, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004224 layer.1.conv_state 0.00050992 0.22743210 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00012018 0.04350451 layer.3.ssm_state 0.00000001 0.00000278 layer.3.conv_state 0.00007056 0.03712477 layer.4.ssm_state 0.00000002 0.00000469 layer.4.conv_state 0.00041272 0.08896521 layer.4.output 0.00000212 0.00021081 ------------------------------------------------------------------------------------- TOTAL 0.00002688 0.00988185 (elements=1,597,440) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1597440 Total Bytes 732996 BPFP 3.6709 bits/point EBPFP 6.2097 equivalent bits/point MSE 0.009882 ---------------------- -------------------------------------------------------- Time: 4.910s Load: 0.010s, Pack+Encode: 2.576s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 190, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample53-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample53-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample54-layer4-item1.zst (68/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample54-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 185, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 185, 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, 185, 4096]) -> torch.Size([1, 1, 185, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,336B, BPFP=4.5981 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,176B, BPFP=2.5132 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,540B, BPFP=8.9209 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,352B, BPFP=4.9043 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,204B, BPFP=10.5479 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,432B, BPFP=3.6274 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,608B, BPFP=10.4023 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,780B, BPFP=3.2214 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,248B, BPFP=9.0938 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 223,116B, BPFP=2.3555 ⌛️ [2/4] FRONTEND: Frontend time: 2.590s (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, 185, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 185, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000138 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000024 0.00003976 layer.1.conv_state 0.00050545 0.22910380 layer.2.ssm_state 0.00000001 0.00000219 layer.2.conv_state 0.00017560 0.04371614 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00011401 0.03798592 layer.4.ssm_state 0.00000001 0.00000428 layer.4.conv_state 0.00021709 0.08445189 layer.4.output 0.00000095 0.00016760 ------------------------------------------------------------------------------------- TOTAL 0.00002455 0.00994968 (elements=1,576,960) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1576960 Total Bytes 728936 BPFP 3.6979 bits/point EBPFP 6.2640 equivalent bits/point MSE 0.009950 ---------------------- -------------------------------------------------------- Time: 4.922s Load: 0.011s, Pack+Encode: 2.590s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 185, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample54-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample54-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample55-layer4-item1.zst (69/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample55-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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: 0.017s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 176, 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, 176, 4096]) -> torch.Size([1, 1, 176, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,232B, BPFP=4.6528 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,288B, BPFP=2.5200 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,516B, BPFP=8.9150 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,888B, BPFP=4.8149 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,284B, BPFP=10.5674 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,508B, BPFP=3.5710 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,532B, BPFP=10.3838 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,180B, BPFP=3.1848 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,380B, BPFP=9.1260 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,448B, BPFP=2.4242 ⌛️ [2/4] FRONTEND: Frontend time: 2.572s (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, 176, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.311s [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, 176, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000025 0.00004032 layer.1.conv_state 0.00049999 0.22842470 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00014724 0.04401245 layer.3.ssm_state 0.00000001 0.00000263 layer.3.conv_state 0.00007475 0.03757970 layer.4.ssm_state 0.00000004 0.00000444 layer.4.conv_state 0.00021006 0.08747436 layer.4.output 0.00000096 0.00018346 ------------------------------------------------------------------------------------- TOTAL 0.00002342 0.01023882 (elements=1,540,096) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1540096 Total Bytes 722400 BPFP 3.7525 bits/point EBPFP 6.3703 equivalent bits/point MSE 0.010239 ---------------------- -------------------------------------------------------- Time: 4.900s Load: 0.017s, Pack+Encode: 2.572s, Decode+Unpack: 2.311s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample55-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample55-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample58-layer4-item1.zst (70/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample58-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 192, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 192, 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, 192, 4096]) -> torch.Size([1, 1, 192, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,192B, BPFP=4.6504 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,328B, BPFP=2.5225 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,544B, BPFP=8.9219 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,836B, BPFP=4.9338 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,280B, BPFP=10.5664 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,944B, BPFP=3.6587 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,596B, BPFP=10.3994 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,584B, BPFP=3.1484 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,248B, BPFP=9.0938 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 217,748B, BPFP=2.2150 ⌛️ [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, 192, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.321s [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, 192, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000137 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000025 0.00004079 layer.1.conv_state 0.00050455 0.22996123 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00014463 0.04411694 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00011943 0.03775381 layer.4.ssm_state 0.00000003 0.00000464 layer.4.conv_state 0.00029170 0.08835885 layer.4.output 0.00000214 0.00016932 ------------------------------------------------------------------------------------- TOTAL 0.00002569 0.00987662 (elements=1,605,632) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1605632 Total Bytes 724444 BPFP 3.6095 bits/point EBPFP 6.1341 equivalent bits/point MSE 0.009877 ---------------------- -------------------------------------------------------- Time: 4.917s Load: 0.013s, Pack+Encode: 2.583s, Decode+Unpack: 2.321s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 192, 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 0.0099 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample58-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample58-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample59-layer4-item1.zst (71/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample59-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 194, 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, 194, 4096]) -> torch.Size([1, 1, 194, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,972B, BPFP=4.6370 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,144B, BPFP=2.5112 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,480B, BPFP=8.9062 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,548B, BPFP=4.8552 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,280B, BPFP=10.5664 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,780B, BPFP=3.5876 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,544B, BPFP=10.3867 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,512B, BPFP=3.2051 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,432B, BPFP=9.1387 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 252,744B, BPFP=2.5445 ⌛️ [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, 194, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.388s [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, 194, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00004164 layer.1.conv_state 0.00050596 0.22822411 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00015065 0.04472810 layer.3.ssm_state 0.00000001 0.00000265 layer.3.conv_state 0.00011979 0.03803526 layer.4.ssm_state 0.00000001 0.00000446 layer.4.conv_state 0.00026273 0.08747636 layer.4.output 0.00000092 0.00014700 ------------------------------------------------------------------------------------- TOTAL 0.00002453 0.00978135 (elements=1,613,824) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1613824 Total Bytes 757580 BPFP 3.7555 bits/point EBPFP 6.2580 equivalent bits/point MSE 0.009781 ---------------------- -------------------------------------------------------- Time: 5.027s Load: 0.011s, Pack+Encode: 2.628s, Decode+Unpack: 2.388s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 194, 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 0.0098 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample59-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample59-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample61-layer4-item1.zst (72/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample61-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 163, 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, 163, 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, 163, 4096]) -> torch.Size([1, 1, 163, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,480B, BPFP=4.6069 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,148B, BPFP=2.5115 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,464B, BPFP=8.9023 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,520B, BPFP=4.9146 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,144B, BPFP=10.5332 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,900B, BPFP=3.5950 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,448B, BPFP=10.3633 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,292B, BPFP=3.1917 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,184B, BPFP=9.0781 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 214,936B, BPFP=2.5754 ⌛️ [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, 163, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.317s [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, 163, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000024 0.00004090 layer.1.conv_state 0.00050555 0.22523049 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00012632 0.04391606 layer.3.ssm_state 0.00000001 0.00000263 layer.3.conv_state 0.00011799 0.03730524 layer.4.ssm_state 0.00000001 0.00000438 layer.4.conv_state 0.00022430 0.08913077 layer.4.output 0.00000101 0.00017540 ------------------------------------------------------------------------------------- TOTAL 0.00002517 0.01055330 (elements=1,486,848) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1486848 Total Bytes 719660 BPFP 3.8721 bits/point EBPFP 6.5878 equivalent bits/point MSE 0.010553 ---------------------- -------------------------------------------------------- Time: 4.907s Load: 0.009s, Pack+Encode: 2.581s, Decode+Unpack: 2.317s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 163, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample61-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample61-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample63-layer4-item1.zst (73/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample63-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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, 175, 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, 175, 4096]) -> torch.Size([1, 1, 175, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,568B, BPFP=4.6733 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,236B, BPFP=2.5168 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,580B, BPFP=4.9182 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,200B, BPFP=10.5469 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,896B, BPFP=3.6558 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,536B, BPFP=10.3848 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,544B, BPFP=3.1460 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,188B, BPFP=9.0791 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 224,396B, BPFP=2.5044 ⌛️ [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, 175, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.313s [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, 175, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00003901 layer.1.conv_state 0.00049528 0.22867334 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00016355 0.04355894 layer.3.ssm_state 0.00000001 0.00000263 layer.3.conv_state 0.00007492 0.03753166 layer.4.ssm_state 0.00000004 0.00000453 layer.4.conv_state 0.00026845 0.08884916 layer.4.output 0.00000230 0.00017876 ------------------------------------------------------------------------------------- TOTAL 0.00002558 0.01028727 (elements=1,536,000) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1536000 Total Bytes 730824 BPFP 3.8064 bits/point EBPFP 6.4440 equivalent bits/point MSE 0.010287 ---------------------- -------------------------------------------------------- Time: 4.896s Load: 0.009s, Pack+Encode: 2.573s, Decode+Unpack: 2.313s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample63-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample63-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample64-layer4-item1.zst (74/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample64-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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, 182, 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, 182, 4096]) -> torch.Size([1, 1, 182, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,752B, BPFP=4.6846 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,264B, BPFP=2.5186 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,472B, BPFP=8.9043 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,152B, BPFP=4.8311 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,224B, BPFP=10.5527 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,268B, BPFP=3.6174 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,516B, BPFP=10.3799 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,720B, BPFP=3.2178 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,368B, BPFP=9.1230 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,840B, BPFP=2.3270 ⌛️ [2/4] FRONTEND: Frontend time: 2.584s (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, 182, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.315s [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, 182, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000031 0.00003964 layer.1.conv_state 0.00048971 0.22687350 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00016503 0.04427696 layer.3.ssm_state 0.00000001 0.00000288 layer.3.conv_state 0.00011771 0.03806885 layer.4.ssm_state 0.00000003 0.00000478 layer.4.conv_state 0.00025892 0.09038130 layer.4.output 0.00000093 0.00016037 ------------------------------------------------------------------------------------- TOTAL 0.00002514 0.01011405 (elements=1,564,672) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1564672 Total Bytes 722720 BPFP 3.6952 bits/point EBPFP 6.2817 equivalent bits/point MSE 0.010114 ---------------------- -------------------------------------------------------- Time: 4.910s Load: 0.012s, Pack+Encode: 2.584s, Decode+Unpack: 2.315s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample64-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample64-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample65-layer4-item1.zst (75/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample65-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 179, 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, 179, 4096]) -> torch.Size([1, 1, 179, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,232B, BPFP=4.6528 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,208B, BPFP=2.5151 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,360B, BPFP=4.7827 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,244B, BPFP=10.5576 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,848B, BPFP=3.5918 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,500B, BPFP=10.3760 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,028B, BPFP=3.1755 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,204B, BPFP=9.0830 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,956B, BPFP=2.3891 ⌛️ [2/4] FRONTEND: Frontend time: 2.580s (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, 179, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.316s [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, 179, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000009 0.00004104 layer.1.conv_state 0.00050147 0.22587754 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00016404 0.04381604 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00011869 0.03715871 layer.4.ssm_state 0.00000001 0.00000447 layer.4.conv_state 0.00020939 0.08356669 layer.4.output 0.00000216 0.00016758 ------------------------------------------------------------------------------------- TOTAL 0.00002509 0.01000251 (elements=1,552,384) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1552384 Total Bytes 722248 BPFP 3.7220 bits/point EBPFP 6.3157 equivalent bits/point MSE 0.010003 ---------------------- -------------------------------------------------------- Time: 4.908s Load: 0.011s, Pack+Encode: 2.580s, Decode+Unpack: 2.316s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample65-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample65-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample66-layer4-item1.zst (76/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample66-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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, 175, 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, 175, 4096]) -> torch.Size([1, 1, 175, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,716B, BPFP=4.6213 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,192B, BPFP=2.5142 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,504B, BPFP=8.9121 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,640B, BPFP=4.9219 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,216B, BPFP=10.5508 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,696B, BPFP=3.6436 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,504B, BPFP=10.3770 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,508B, BPFP=3.1438 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,304B, BPFP=9.1074 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,392B, BPFP=2.4821 ⌛️ [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, 175, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 175, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000146 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004074 layer.1.conv_state 0.00051073 0.22885391 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00016593 0.04413203 layer.3.ssm_state 0.00000001 0.00000273 layer.3.conv_state 0.00007184 0.03751760 layer.4.ssm_state 0.00000002 0.00000478 layer.4.conv_state 0.00035513 0.08994661 layer.4.output 0.00000218 0.00017831 ------------------------------------------------------------------------------------- TOTAL 0.00002769 0.01032643 (elements=1,536,000) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1536000 Total Bytes 727816 BPFP 3.7907 bits/point EBPFP 6.4231 equivalent bits/point MSE 0.010326 ---------------------- -------------------------------------------------------- Time: 4.924s Load: 0.012s, Pack+Encode: 2.592s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample66-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample66-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample68-layer4-item1.zst (77/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample68-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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, 176, 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, 176, 4096]) -> torch.Size([1, 1, 176, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,828B, BPFP=4.6282 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,216B, BPFP=2.5156 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,464B, BPFP=8.9023 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,712B, BPFP=4.9263 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,192B, BPFP=10.5449 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,552B, BPFP=3.6348 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,540B, BPFP=10.3857 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,484B, BPFP=3.1423 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,196B, BPFP=9.0811 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,520B, BPFP=2.4694 ⌛️ [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, 176, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.326s [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, 176, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000022 0.00003971 layer.1.conv_state 0.00050862 0.22764464 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00017603 0.04371645 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00006868 0.03759557 layer.4.ssm_state 0.00000007 0.00000495 layer.4.conv_state 0.00030923 0.09317963 layer.4.output 0.00000105 0.00018756 ------------------------------------------------------------------------------------- TOTAL 0.00002624 0.01033957 (elements=1,540,096) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1540096 Total Bytes 727848 BPFP 3.7808 bits/point EBPFP 6.4057 equivalent bits/point MSE 0.010340 ---------------------- -------------------------------------------------------- Time: 4.914s Load: 0.012s, Pack+Encode: 2.575s, Decode+Unpack: 2.326s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 176, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample68-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample68-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample71-layer4-item1.zst (78/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample71-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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, 178, 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, 178, 4096]) -> torch.Size([1, 1, 178, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,680B, BPFP=4.6802 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,236B, BPFP=2.5168 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,500B, BPFP=8.9111 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,792B, BPFP=4.8701 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,220B, BPFP=10.5518 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,192B, BPFP=3.6128 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,516B, BPFP=10.3799 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,348B, BPFP=3.1951 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,272B, BPFP=9.0996 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,580B, BPFP=2.4423 ⌛️ [2/4] FRONTEND: Frontend time: 2.580s (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, 178, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.322s [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, 178, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004011 layer.1.conv_state 0.00051443 0.22798711 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00013474 0.04382544 layer.3.ssm_state 0.00000001 0.00000265 layer.3.conv_state 0.00007456 0.03768951 layer.4.ssm_state 0.00000004 0.00000468 layer.4.conv_state 0.00023149 0.09029844 layer.4.output 0.00000221 0.00017151 ------------------------------------------------------------------------------------- TOTAL 0.00002436 0.01022887 (elements=1,548,288) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1548288 Total Bytes 728480 BPFP 3.7641 bits/point EBPFP 6.3780 equivalent bits/point MSE 0.010229 ---------------------- -------------------------------------------------------- Time: 4.912s Load: 0.010s, Pack+Encode: 2.580s, Decode+Unpack: 2.322s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample71-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample71-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample72-layer4-item1.zst (79/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample72-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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: 0.013s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 184, 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, 184, 4096]) -> torch.Size([1, 1, 184, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,152B, BPFP=4.5869 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,176B, BPFP=2.5132 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,516B, BPFP=8.9150 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,864B, BPFP=4.9355 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,312B, BPFP=10.5742 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,092B, BPFP=3.6067 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,528B, BPFP=10.3828 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,100B, BPFP=3.1799 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,224B, BPFP=9.0879 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,564B, BPFP=2.3625 ⌛️ [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, 184, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.318s [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, 184, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00004043 layer.1.conv_state 0.00050142 0.22849867 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012834 0.04423264 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00006642 0.03788278 layer.4.ssm_state 0.00000001 0.00000455 layer.4.conv_state 0.00021693 0.08438088 layer.4.output 0.00000094 0.00016648 ------------------------------------------------------------------------------------- TOTAL 0.00002253 0.00996922 (elements=1,572,864) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1572864 Total Bytes 727672 BPFP 3.7011 bits/point EBPFP 6.2702 equivalent bits/point MSE 0.009969 ---------------------- -------------------------------------------------------- Time: 4.925s Load: 0.013s, Pack+Encode: 2.594s, Decode+Unpack: 2.318s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 184, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample72-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample72-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample73-layer4-item1.zst (80/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample73-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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, 172, 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, 172, 4096]) -> torch.Size([1, 1, 172, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,308B, BPFP=4.6575 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,292B, BPFP=2.5203 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,972B, BPFP=4.8811 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,360B, BPFP=10.5859 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,636B, BPFP=3.6399 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,620B, BPFP=10.4053 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,436B, BPFP=3.2004 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,288B, BPFP=9.1035 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,228B, BPFP=2.4894 ⌛️ [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, 172, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 172, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000149 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000018 0.00004374 layer.1.conv_state 0.00051489 0.22626618 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00015859 0.04441116 layer.3.ssm_state 0.00000001 0.00000284 layer.3.conv_state 0.00007387 0.03733850 layer.4.ssm_state 0.00000004 0.00000470 layer.4.conv_state 0.00025418 0.09124385 layer.4.output 0.00000227 0.00018741 ------------------------------------------------------------------------------------- TOTAL 0.00002575 0.01038714 (elements=1,523,712) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1523712 Total Bytes 725768 BPFP 3.8105 bits/point EBPFP 6.4700 equivalent bits/point MSE 0.010387 ---------------------- -------------------------------------------------------- Time: 4.897s Load: 0.009s, Pack+Encode: 2.574s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample73-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample73-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample76-layer4-item1.zst (81/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample76-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 172, 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, 172, 4096]) -> torch.Size([1, 1, 172, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,768B, BPFP=4.6245 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,220B, BPFP=2.5159 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,540B, BPFP=8.9209 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,308B, BPFP=4.8406 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,320B, BPFP=10.5762 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,864B, BPFP=3.5928 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,644B, BPFP=10.4111 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,176B, BPFP=3.1846 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,432B, BPFP=9.1387 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,556B, BPFP=2.5159 ⌛️ [2/4] FRONTEND: Frontend time: 2.595s (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, 172, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 172, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000024 0.00004142 layer.1.conv_state 0.00051342 0.22522280 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00016370 0.04381084 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00007350 0.03819366 layer.4.ssm_state 0.00000001 0.00000477 layer.4.conv_state 0.00023274 0.09143198 layer.4.output 0.00000099 0.00019000 ------------------------------------------------------------------------------------- TOTAL 0.00002477 0.01037521 (elements=1,523,712) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1523712 Total Bytes 725972 BPFP 3.8116 bits/point EBPFP 6.4600 equivalent bits/point MSE 0.010375 ---------------------- -------------------------------------------------------- Time: 4.920s Load: 0.011s, Pack+Encode: 2.595s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 172, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample76-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample76-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample77-layer4-item1.zst (82/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample77-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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, 179, 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, 179, 4096]) -> torch.Size([1, 1, 179, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,592B, BPFP=4.6138 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,304B, BPFP=2.5210 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,516B, BPFP=8.9150 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,900B, BPFP=4.8767 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,220B, BPFP=10.5518 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,672B, BPFP=3.6421 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,624B, BPFP=10.4062 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,928B, BPFP=3.1694 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,252B, BPFP=9.0947 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 225,624B, BPFP=2.4619 ⌛️ [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, 179, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.313s [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, 179, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00003950 layer.1.conv_state 0.00050384 0.22742008 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00014578 0.04352612 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00006954 0.03787462 layer.4.ssm_state 0.00000001 0.00000466 layer.4.conv_state 0.00022371 0.08503648 layer.4.output 0.00000104 0.00017380 ------------------------------------------------------------------------------------- TOTAL 0.00002349 0.01007791 (elements=1,552,384) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1552384 Total Bytes 730776 BPFP 3.7660 bits/point EBPFP 6.3692 equivalent bits/point MSE 0.010078 ---------------------- -------------------------------------------------------- Time: 4.901s Load: 0.012s, Pack+Encode: 2.577s, Decode+Unpack: 2.313s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample77-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample77-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample78-layer4-item1.zst (83/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample78-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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, 173, 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, 173, 4096]) -> torch.Size([1, 1, 173, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,888B, BPFP=4.6318 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,124B, BPFP=2.5100 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,508B, BPFP=8.9131 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,204B, BPFP=4.8953 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,216B, BPFP=10.5508 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,132B, BPFP=3.6091 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,540B, BPFP=10.3857 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 53,004B, BPFP=3.2351 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,160B, BPFP=9.0723 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,572B, BPFP=2.4789 ⌛️ [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, 173, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.329s [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, 173, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00004290 layer.1.conv_state 0.00050351 0.22704899 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012771 0.04406469 layer.3.ssm_state 0.00000001 0.00000264 layer.3.conv_state 0.00011854 0.03722550 layer.4.ssm_state 0.00000006 0.00000426 layer.4.conv_state 0.00024363 0.08310118 layer.4.output 0.00000222 0.00018662 ------------------------------------------------------------------------------------- TOTAL 0.00002550 0.01019159 (elements=1,527,808) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1527808 Total Bytes 725492 BPFP 3.7989 bits/point EBPFP 6.4480 equivalent bits/point MSE 0.010192 ---------------------- -------------------------------------------------------- Time: 4.926s Load: 0.012s, Pack+Encode: 2.586s, Decode+Unpack: 2.329s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample78-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample78-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample79-layer4-item1.zst (84/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample79-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 164, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 164, 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, 164, 4096]) -> torch.Size([1, 1, 164, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 77,856B, BPFP=4.7520 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,204B, BPFP=2.5149 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,552B, BPFP=8.9238 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 81,832B, BPFP=4.9946 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,264B, BPFP=10.5625 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 60,156B, BPFP=3.6716 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,484B, BPFP=10.3721 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,984B, BPFP=3.1729 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,184B, BPFP=9.0781 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 212,192B, BPFP=2.5271 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 164, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.320s [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, 164, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000013 0.00003958 layer.1.conv_state 0.00048863 0.22759467 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00016665 0.04358893 layer.3.ssm_state 0.00000001 0.00000276 layer.3.conv_state 0.00011922 0.03788121 layer.4.ssm_state 0.00000005 0.00000507 layer.4.conv_state 0.00036108 0.09333809 layer.4.output 0.00000224 0.00018634 ------------------------------------------------------------------------------------- TOTAL 0.00002920 0.01067957 (elements=1,490,944) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1490944 Total Bytes 721852 BPFP 3.8733 bits/point EBPFP 6.6080 equivalent bits/point MSE 0.010680 ---------------------- -------------------------------------------------------- Time: 4.922s Load: 0.011s, Pack+Encode: 2.591s, Decode+Unpack: 2.320s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 164, 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 0.0107 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample79-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample79-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample8-layer4-item1.zst (85/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample8-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 231, 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, 231, 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, 231, 4096]) -> torch.Size([1, 1, 231, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,472B, BPFP=4.6064 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,148B, BPFP=2.5115 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,832B, BPFP=4.8726 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,188B, BPFP=10.5439 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,212B, BPFP=3.6140 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,536B, BPFP=10.3848 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,396B, BPFP=3.1980 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,400B, BPFP=9.1309 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 287,032B, BPFP=2.4269 ⌛️ [2/4] FRONTEND: Frontend time: 2.630s (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, 231, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.380s [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, 231, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000031 0.00003998 layer.1.conv_state 0.00050949 0.22686824 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00013166 0.04336898 layer.3.ssm_state 0.00000001 0.00000264 layer.3.conv_state 0.00011605 0.03761037 layer.4.ssm_state 0.00000004 0.00000458 layer.4.conv_state 0.00023796 0.08644477 layer.4.output 0.00000160 0.00013625 ------------------------------------------------------------------------------------- TOTAL 0.00002207 0.00887096 (elements=1,765,376) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1765376 Total Bytes 791896 BPFP 3.5886 bits/point EBPFP 5.8764 equivalent bits/point MSE 0.008871 ---------------------- -------------------------------------------------------- Time: 5.022s Load: 0.012s, Pack+Encode: 2.630s, Decode+Unpack: 2.380s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 231, 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 0.0089 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample8-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample8-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample81-layer4-item1.zst (86/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample81-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 171, 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, 171, 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, 171, 4096]) -> torch.Size([1, 1, 171, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,284B, BPFP=4.6560 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 40,996B, BPFP=2.5022 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,568B, BPFP=8.9277 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,772B, BPFP=4.8079 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,352B, BPFP=10.5840 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,920B, BPFP=3.5962 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,788B, BPFP=10.4463 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,644B, BPFP=3.2131 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,480B, BPFP=9.1504 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,308B, BPFP=2.5392 ⌛️ [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, 171, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 171, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000025 0.00004000 layer.1.conv_state 0.00050732 0.22817056 layer.2.ssm_state 0.00000001 0.00000213 layer.2.conv_state 0.00011782 0.04478267 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00007792 0.03827268 layer.4.ssm_state 0.00000003 0.00000436 layer.4.conv_state 0.00023472 0.08746523 layer.4.output 0.00000098 0.00017446 ------------------------------------------------------------------------------------- TOTAL 0.00002385 0.01039604 (elements=1,519,616) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1519616 Total Bytes 727256 BPFP 3.8286 bits/point EBPFP 6.4869 equivalent bits/point MSE 0.010396 ---------------------- -------------------------------------------------------- Time: 4.916s Load: 0.009s, Pack+Encode: 2.588s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 171, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample81-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample81-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample82-layer4-item1.zst (87/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample82-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 170, 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, 170, 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, 170, 4096]) -> torch.Size([1, 1, 170, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,748B, BPFP=4.6233 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,172B, BPFP=2.5129 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,496B, BPFP=8.9102 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,636B, BPFP=4.8606 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,292B, BPFP=10.5693 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 60,004B, BPFP=3.6624 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,640B, BPFP=10.4102 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,432B, BPFP=3.1392 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,172B, BPFP=9.0752 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 222,168B, BPFP=2.5525 ⌛️ [2/4] FRONTEND: Frontend time: 2.590s (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, 170, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.317s [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, 170, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004143 layer.1.conv_state 0.00051054 0.22659793 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00015881 0.04427860 layer.3.ssm_state 0.00000001 0.00000278 layer.3.conv_state 0.00007716 0.03803890 layer.4.ssm_state 0.00000001 0.00000472 layer.4.conv_state 0.00033254 0.08641540 layer.4.output 0.00000101 0.00017350 ------------------------------------------------------------------------------------- TOTAL 0.00002696 0.01035073 (elements=1,515,520) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1515520 Total Bytes 726904 BPFP 3.8371 bits/point EBPFP 6.5015 equivalent bits/point MSE 0.010351 ---------------------- -------------------------------------------------------- Time: 4.918s Load: 0.010s, Pack+Encode: 2.590s, Decode+Unpack: 2.317s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 170, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample82-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample82-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample84-layer4-item1.zst (88/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample84-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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, 180, 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, 180, 4096]) -> torch.Size([1, 1, 180, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,368B, BPFP=4.6611 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,236B, BPFP=2.5168 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,468B, BPFP=8.9033 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,020B, BPFP=4.8840 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,248B, BPFP=10.5586 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,036B, BPFP=3.6033 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,628B, BPFP=10.4072 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,420B, BPFP=3.1995 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,428B, BPFP=9.1377 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 216,656B, BPFP=2.3509 ⌛️ [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, 180, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 180, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000014 0.00004167 layer.1.conv_state 0.00051556 0.22921161 layer.2.ssm_state 0.00000001 0.00000215 layer.2.conv_state 0.00013957 0.04462051 layer.3.ssm_state 0.00000001 0.00000278 layer.3.conv_state 0.00012063 0.03826775 layer.4.ssm_state 0.00000004 0.00000434 layer.4.conv_state 0.00023889 0.08576980 layer.4.output 0.00000092 0.00017316 ------------------------------------------------------------------------------------- TOTAL 0.00002489 0.01013618 (elements=1,556,480) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1556480 Total Bytes 722652 BPFP 3.7143 bits/point EBPFP 6.3150 equivalent bits/point MSE 0.010136 ---------------------- -------------------------------------------------------- Time: 4.898s Load: 0.010s, Pack+Encode: 2.570s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample84-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample84-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample85-layer4-item1.zst (89/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample85-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 183, 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, 183, 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, 183, 4096]) -> torch.Size([1, 1, 183, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,992B, BPFP=4.6382 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,144B, BPFP=2.5112 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,524B, BPFP=8.9170 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,280B, BPFP=4.7778 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,316B, BPFP=10.5752 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,564B, BPFP=3.6355 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,640B, BPFP=10.4102 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,928B, BPFP=3.1694 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,292B, BPFP=9.1045 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 223,376B, BPFP=2.3841 ⌛️ [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, 183, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.323s [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, 183, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000145 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004187 layer.1.conv_state 0.00049509 0.22663319 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00014259 0.04457862 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00007565 0.03797072 layer.4.ssm_state 0.00000001 0.00000475 layer.4.conv_state 0.00024574 0.08737294 layer.4.output 0.00000097 0.00016679 ------------------------------------------------------------------------------------- TOTAL 0.00002356 0.01002769 (elements=1,568,768) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1568768 Total Bytes 727200 BPFP 3.7084 bits/point EBPFP 6.2777 equivalent bits/point MSE 0.010028 ---------------------- -------------------------------------------------------- Time: 4.912s Load: 0.010s, Pack+Encode: 2.579s, Decode+Unpack: 2.323s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 183, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample85-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample85-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample86-layer4-item1.zst (90/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample86-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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, 175, 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, 175, 4096]) -> torch.Size([1, 1, 175, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,376B, BPFP=4.6006 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,196B, BPFP=2.5144 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,544B, BPFP=8.9219 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,436B, BPFP=4.9094 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,256B, BPFP=10.5605 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,056B, BPFP=3.6045 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,540B, BPFP=10.3857 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,768B, BPFP=3.1597 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,268B, BPFP=9.0986 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 221,580B, BPFP=2.4730 ⌛️ [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, 175, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.321s [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, 175, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004132 layer.1.conv_state 0.00050931 0.22678110 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00014221 0.04385502 layer.3.ssm_state 0.00000001 0.00000277 layer.3.conv_state 0.00006953 0.03766498 layer.4.ssm_state 0.00000004 0.00000504 layer.4.conv_state 0.00030746 0.09128862 layer.4.output 0.00000095 0.00017398 ------------------------------------------------------------------------------------- TOTAL 0.00002552 0.01030613 (elements=1,536,000) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1536000 Total Bytes 726164 BPFP 3.7821 bits/point EBPFP 6.4101 equivalent bits/point MSE 0.010306 ---------------------- -------------------------------------------------------- Time: 4.910s Load: 0.010s, Pack+Encode: 2.579s, Decode+Unpack: 2.321s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 175, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample86-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample86-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample89-layer4-item1.zst (91/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample89-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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, 179, 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, 179, 4096]) -> torch.Size([1, 1, 179, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,892B, BPFP=4.6321 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,164B, BPFP=2.5125 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,472B, BPFP=8.9043 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,156B, BPFP=4.8923 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,192B, BPFP=10.5449 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,100B, BPFP=3.6072 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,576B, BPFP=10.3945 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,704B, BPFP=3.2168 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,408B, BPFP=9.1328 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,732B, BPFP=2.3867 ⌛️ [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, 179, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.321s [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, 179, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00004135 layer.1.conv_state 0.00050671 0.22572681 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00014788 0.04436813 layer.3.ssm_state 0.00000001 0.00000275 layer.3.conv_state 0.00011867 0.03756142 layer.4.ssm_state 0.00000004 0.00000440 layer.4.conv_state 0.00026927 0.08590980 layer.4.output 0.00000218 0.00016079 ------------------------------------------------------------------------------------- TOTAL 0.00002615 0.01006577 (elements=1,552,384) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1552384 Total Bytes 724540 BPFP 3.7338 bits/point EBPFP 6.3404 equivalent bits/point MSE 0.010066 ---------------------- -------------------------------------------------------- Time: 4.919s Load: 0.012s, Pack+Encode: 2.586s, Decode+Unpack: 2.321s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample89-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample89-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample9-layer4-item1.zst (92/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample9-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 222, 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: 0.011s ------------------------------------------------------------ FalconMamba Features Summary ------------------------------------------------------------ Number of layers: 5 SSM state shape: (1, 8192, 16) Conv state shape: (1, 8192, 4) Output shape: (1, 222, 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, 222, 4096]) -> torch.Size([1, 1, 222, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 75,364B, BPFP=4.5999 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,220B, BPFP=2.5159 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,488B, BPFP=8.9082 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,036B, BPFP=4.8850 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,232B, BPFP=10.5547 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,676B, BPFP=3.6423 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,584B, BPFP=10.3965 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,288B, BPFP=3.1914 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,228B, BPFP=9.0889 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 282,180B, BPFP=2.4826 ⌛️ [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, 222, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.377s [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, 222, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004008 layer.1.conv_state 0.00049966 0.22756499 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00014998 0.04391404 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00007558 0.03737194 layer.4.ssm_state 0.00000001 0.00000473 layer.4.conv_state 0.00024683 0.08724256 layer.4.output 0.00000168 0.00014232 ------------------------------------------------------------------------------------- TOTAL 0.00002209 0.00909461 (elements=1,728,512) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1728512 Total Bytes 787440 BPFP 3.6445 bits/point EBPFP 5.9829 equivalent bits/point MSE 0.009095 ---------------------- -------------------------------------------------------- Time: 5.028s Load: 0.011s, Pack+Encode: 2.640s, Decode+Unpack: 2.377s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 222, 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 0.0091 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample9-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample9-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample90-layer4-item1.zst (93/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample90-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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, 179, 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, 179, 4096]) -> torch.Size([1, 1, 179, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,276B, BPFP=4.6555 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,304B, BPFP=2.5210 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,528B, BPFP=8.9180 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,328B, BPFP=4.8418 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,184B, BPFP=10.5430 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,472B, BPFP=3.6299 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,644B, BPFP=10.4111 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,836B, BPFP=3.2249 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,424B, BPFP=9.1367 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,816B, BPFP=2.3985 ⌛️ [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, 179, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 179, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000138 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000029 0.00004248 layer.1.conv_state 0.00050231 0.22559394 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00012015 0.04465353 layer.3.ssm_state 0.00000001 0.00000281 layer.3.conv_state 0.00006912 0.03804892 layer.4.ssm_state 0.00000006 0.00000454 layer.4.conv_state 0.00021996 0.08941206 layer.4.output 0.00000096 0.00016535 ------------------------------------------------------------------------------------- TOTAL 0.00002281 0.01015547 (elements=1,552,384) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1552384 Total Bytes 725956 BPFP 3.7411 bits/point EBPFP 6.3494 equivalent bits/point MSE 0.010155 ---------------------- -------------------------------------------------------- Time: 4.907s Load: 0.010s, Pack+Encode: 2.583s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 179, 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 0.0102 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample90-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample90-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample91-layer4-item1.zst (94/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample91-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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, 173, 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, 173, 4096]) -> torch.Size([1, 1, 173, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,340B, BPFP=4.6594 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,248B, BPFP=2.5176 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,520B, BPFP=8.9160 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,480B, BPFP=4.9121 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,212B, BPFP=10.5498 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,548B, BPFP=3.5735 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,540B, BPFP=10.3857 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,472B, BPFP=3.2026 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,404B, BPFP=9.1318 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 220,780B, BPFP=2.4925 ⌛️ [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, 173, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.318s [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, 173, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000142 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000027 0.00004066 layer.1.conv_state 0.00050455 0.22571018 layer.2.ssm_state 0.00000001 0.00000214 layer.2.conv_state 0.00016590 0.04348157 layer.3.ssm_state 0.00000001 0.00000261 layer.3.conv_state 0.00012109 0.03798436 layer.4.ssm_state 0.00000001 0.00000471 layer.4.conv_state 0.00024452 0.08860859 layer.4.output 0.00000102 0.00018826 ------------------------------------------------------------------------------------- TOTAL 0.00002586 0.01028537 (elements=1,527,808) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1527808 Total Bytes 726688 BPFP 3.8051 bits/point EBPFP 6.4542 equivalent bits/point MSE 0.010285 ---------------------- -------------------------------------------------------- Time: 4.909s Load: 0.012s, Pack+Encode: 2.579s, Decode+Unpack: 2.318s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 173, 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 0.0103 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample91-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample91-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample92-layer4-item1.zst (95/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample92-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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, 182, 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, 182, 4096]) -> torch.Size([1, 1, 182, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,152B, BPFP=4.6479 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,148B, BPFP=2.5115 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,568B, BPFP=8.9277 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,996B, BPFP=4.8826 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,272B, BPFP=10.5645 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,276B, BPFP=3.6179 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,608B, BPFP=10.4023 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,740B, BPFP=3.1580 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,320B, BPFP=9.1113 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,084B, BPFP=2.3511 ⌛️ [2/4] FRONTEND: Frontend time: 2.568s (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, 182, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.319s [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, 182, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000143 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004115 layer.1.conv_state 0.00048835 0.22440459 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00017267 0.04362957 layer.3.ssm_state 0.00000001 0.00000279 layer.3.conv_state 0.00007600 0.03737459 layer.4.ssm_state 0.00000004 0.00000492 layer.4.conv_state 0.00032314 0.08887091 layer.4.output 0.00000092 0.00015890 ------------------------------------------------------------------------------------- TOTAL 0.00002572 0.01000205 (elements=1,564,672) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1564672 Total Bytes 724308 BPFP 3.7033 bits/point EBPFP 6.2865 equivalent bits/point MSE 0.010002 ---------------------- -------------------------------------------------------- Time: 4.896s Load: 0.009s, Pack+Encode: 2.568s, Decode+Unpack: 2.319s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 182, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample92-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample92-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample94-layer4-item1.zst (96/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample94-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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, 178, 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, 178, 4096]) -> torch.Size([1, 1, 178, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,228B, BPFP=4.6526 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,340B, BPFP=2.5232 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,484B, BPFP=8.9072 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,992B, BPFP=4.8823 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,188B, BPFP=10.5439 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,340B, BPFP=3.6218 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,472B, BPFP=10.3691 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,252B, BPFP=3.1892 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,312B, BPFP=9.1094 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 218,008B, BPFP=2.3921 ⌛️ [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, 178, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.324s [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, 178, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000012 0.00004104 layer.1.conv_state 0.00049680 0.22755887 layer.2.ssm_state 0.00000001 0.00000218 layer.2.conv_state 0.00014561 0.04458842 layer.3.ssm_state 0.00000001 0.00000272 layer.3.conv_state 0.00011773 0.03761918 layer.4.ssm_state 0.00000001 0.00000447 layer.4.conv_state 0.00023886 0.08382133 layer.4.output 0.00000218 0.00016386 ------------------------------------------------------------------------------------- TOTAL 0.00002527 0.01009385 (elements=1,548,288) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1548288 Total Bytes 723760 BPFP 3.7397 bits/point EBPFP 6.3529 equivalent bits/point MSE 0.010094 ---------------------- -------------------------------------------------------- Time: 4.906s Load: 0.010s, Pack+Encode: 2.573s, Decode+Unpack: 2.324s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 178, 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 0.0101 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample94-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample94-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample96-layer4-item1.zst (97/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample96-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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, 165, 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, 165, 4096]) -> torch.Size([1, 1, 165, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,092B, BPFP=4.6443 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,140B, BPFP=2.5110 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,520B, BPFP=8.9160 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 78,896B, BPFP=4.8154 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,252B, BPFP=10.5596 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,968B, BPFP=3.5991 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,652B, BPFP=10.4131 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,292B, BPFP=3.1917 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,472B, BPFP=9.1484 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 220,016B, BPFP=2.6044 ⌛️ [2/4] FRONTEND: Frontend time: 2.582s (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, 165, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.321s [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, 165, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004081 layer.1.conv_state 0.00050456 0.22666262 layer.2.ssm_state 0.00000001 0.00000211 layer.2.conv_state 0.00011481 0.04480083 layer.3.ssm_state 0.00000001 0.00000276 layer.3.conv_state 0.00007970 0.03856218 layer.4.ssm_state 0.00000004 0.00000462 layer.4.conv_state 0.00026671 0.09001518 layer.4.output 0.00000102 0.00019333 ------------------------------------------------------------------------------------- TOTAL 0.00002485 0.01060227 (elements=1,495,040) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1495040 Total Bytes 724444 BPFP 3.8765 bits/point EBPFP 6.5757 equivalent bits/point MSE 0.010602 ---------------------- -------------------------------------------------------- Time: 4.914s Load: 0.012s, Pack+Encode: 2.582s, Decode+Unpack: 2.321s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 165, 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 0.0106 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample96-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample96-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample97-layer4-item1.zst (98/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample97-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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, 169, 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, 169, 4096]) -> torch.Size([1, 1, 169, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,580B, BPFP=4.6741 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,188B, BPFP=2.5139 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,536B, BPFP=8.9199 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 80,380B, BPFP=4.9060 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,320B, BPFP=10.5762 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 58,964B, BPFP=3.5989 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,604B, BPFP=10.4014 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,384B, BPFP=3.1973 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,296B, BPFP=9.1055 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 220,448B, BPFP=2.5477 ⌛️ [2/4] FRONTEND: Frontend time: 2.595s (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, 169, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.314s [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, 169, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000141 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000010 0.00004112 layer.1.conv_state 0.00050728 0.22917387 layer.2.ssm_state 0.00000001 0.00000213 layer.2.conv_state 0.00017692 0.04380749 layer.3.ssm_state 0.00000001 0.00000268 layer.3.conv_state 0.00007651 0.03753615 layer.4.ssm_state 0.00000001 0.00000448 layer.4.conv_state 0.00023506 0.08665317 layer.4.output 0.00000099 0.00017301 ------------------------------------------------------------------------------------- TOTAL 0.00002522 0.01041791 (elements=1,511,424) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1511424 Total Bytes 726844 BPFP 3.8472 bits/point EBPFP 6.5276 equivalent bits/point MSE 0.010418 ---------------------- -------------------------------------------------------- Time: 4.921s Load: 0.012s, Pack+Encode: 2.595s, Decode+Unpack: 2.314s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 169, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample97-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample97-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample98-layer4-item1.zst (99/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample98-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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, 167, 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, 167, 4096]) -> torch.Size([1, 1, 167, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,308B, BPFP=4.6575 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,068B, BPFP=2.5066 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,500B, BPFP=8.9111 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,652B, BPFP=4.8616 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,288B, BPFP=10.5684 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,164B, BPFP=3.6111 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,476B, BPFP=10.3701 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 51,272B, BPFP=3.1294 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,272B, BPFP=9.0996 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,152B, BPFP=2.5631 ⌛️ [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, 167, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.316s [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, 167, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000000 0.00000140 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00003981 layer.1.conv_state 0.00050724 0.22432269 layer.2.ssm_state 0.00000001 0.00000217 layer.2.conv_state 0.00015963 0.04524369 layer.3.ssm_state 0.00000001 0.00000271 layer.3.conv_state 0.00007484 0.03727098 layer.4.ssm_state 0.00000002 0.00000508 layer.4.conv_state 0.00026827 0.08739710 layer.4.output 0.00000233 0.00016852 ------------------------------------------------------------------------------------- TOTAL 0.00002627 0.01040764 (elements=1,503,232) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1503232 Total Bytes 723296 BPFP 3.8493 bits/point EBPFP 6.5323 equivalent bits/point MSE 0.010408 ---------------------- -------------------------------------------------------- Time: 4.921s Load: 0.009s, Pack+Encode: 2.596s, Decode+Unpack: 2.316s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 167, 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 0.0104 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample98-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample98-layer4-item1.zst 💪 Processing: ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample99-layer4-item1.zst (100/100) [1/4] FRONTEND: Loading features from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample99-layer4-item1.zst... Original data structure: root: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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, 180, 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, 180, 4096]) -> torch.Size([1, 1, 180, 4096]) Using per-key quantization points (layer.0.ssm_state: torch.Size([256])) for layer.0.ssm_state layer.0.ssm_state: 76,104B, BPFP=4.6450 Using per-key quantization points (layer.0.conv_state: torch.Size([256])) for layer.0.conv_state layer.0.conv_state: 37,144B, BPFP=9.0684 Using per-key quantization points (layer.1.ssm_state: torch.Size([256])) for layer.1.ssm_state layer.1.ssm_state: 41,128B, BPFP=2.5103 Using per-key quantization points (layer.1.conv_state: torch.Size([256])) for layer.1.conv_state layer.1.conv_state: 36,496B, BPFP=8.9102 Using per-key quantization points (layer.2.ssm_state: torch.Size([256])) for layer.2.ssm_state layer.2.ssm_state: 79,892B, BPFP=4.8762 Using per-key quantization points (layer.2.conv_state: torch.Size([256])) for layer.2.conv_state layer.2.conv_state: 43,284B, BPFP=10.5674 Using per-key quantization points (layer.3.ssm_state: torch.Size([256])) for layer.3.ssm_state layer.3.ssm_state: 59,292B, BPFP=3.6189 Using per-key quantization points (layer.3.conv_state: torch.Size([256])) for layer.3.conv_state layer.3.conv_state: 42,524B, BPFP=10.3818 Using per-key quantization points (layer.4.ssm_state: torch.Size([256])) for layer.4.ssm_state layer.4.ssm_state: 52,728B, BPFP=3.2183 Using per-key quantization points (layer.4.conv_state: torch.Size([256])) for layer.4.conv_state layer.4.conv_state: 37,356B, BPFP=9.1201 Using per-key quantization points (layer.4.output: torch.Size([256])) for layer.4.output layer.4.output: 219,732B, BPFP=2.3842 ⌛️ [2/4] FRONTEND: Frontend time: 2.591s (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, 180, 4096]) ⌛️ [3/4] BACKEND: Backend time: 2.309s [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, 180, 4096]) Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): Key Quant-MSE Total-MSE ------------------------------------------------------------------------------------- layer.0.ssm_state 0.00000001 0.00000144 layer.0.conv_state 0.00014612 0.07949343 layer.1.ssm_state 0.00000008 0.00004114 layer.1.conv_state 0.00048137 0.22597519 layer.2.ssm_state 0.00000001 0.00000216 layer.2.conv_state 0.00017169 0.04441203 layer.3.ssm_state 0.00000001 0.00000276 layer.3.conv_state 0.00007691 0.03765545 layer.4.ssm_state 0.00000001 0.00000448 layer.4.conv_state 0.00020990 0.08348551 layer.4.output 0.00000214 0.00017511 ------------------------------------------------------------------------------------- TOTAL 0.00002388 0.01000357 (elements=1,556,480) ---------------------- -------------------------------------------------------- SAMPLE-WISE STATISTICS ---------------------- -------------------------------------------------------- Handler falconmamba Strategy individual Architecture elic-featurecoding ---------------------- -------------------------------------------------------- Total Elements 1556480 Total Bytes 725680 BPFP 3.7299 bits/point EBPFP 6.3303 equivalent bits/point MSE 0.010004 ---------------------- -------------------------------------------------------- Time: 4.912s Load: 0.012s, Pack+Encode: 2.591s, Decode+Unpack: 2.309s ---------------------- -------------------------------------------------------- Restored Feature Format: [dict] with 3 keys key['output']: [Tensor] shape=torch.Size([1, 180, 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 0.0100 MSE: from ../datasets/FalconMamba-500features-L5wCache/tiiuae/falcon-mamba-7b-instruct/fc_arc_challenge/sample99-layer4-item1.zst to output-fixed/falconmamba/lambda0.02/elic-featurecoding-8bit-individual/fc_arc_challenge/sample99-layer4-item1.zst ------------------------ ---------------------------- TOTAL PROCESSING SUMMARY ------------------------ ---------------------------- Total files 100 Avg BPFP 3.7454 bits/point Avg EBPFP 6.3183 equivalent bits/point Avg MSE 0.010005 Avg Time 4.950s ------------------------ ----------------------------