=== Quantization Report === Model: Qwen/Qwen3.6-35B-A3B Date: 2026-06-21 00:28:05 Version: 0.14.2 CLI Arguments: --batch_size 8 --iters 1000 --nsamples 512 --seqlen 2048 --dataset opencode-instruct --output_dir ./models/Qwen3.6-35B-A3B-int4-AutoRound Memory Summary: Peak RAM: 30.21 GB Peak VRAM: 28.16 GB Sensitivity Analysis: ───────────────────────────────────────────────────────────────────────────────────────────────────── Layer Cosine Sim PSNR (dB) Iters uLoss Status ───────────────────────────────────────────────────────────────────────────────────────────────────── 🟢 model.language_model.layers.0 0.9995 63.3 976 11 → 0 PASS 🟢 model.language_model.layers.1 0.9994 63.3 947 3 → 1 PASS 🟢 model.language_model.layers.2 0.9995 67.1 993 7 → 1 PASS 🟢 model.language_model.layers.3 0.9991 62.3 879 9 → 2 PASS 🟢 model.language_model.layers.4 0.9989 61.7 906 6 → 3 PASS 🟢 model.language_model.layers.5 0.9988 61.0 950 8 → 4 PASS 🟢 model.language_model.layers.6 0.9988 72.1 874 9 → 4 PASS 🟢 model.language_model.layers.7 0.9980 68.0 944 15 → 6 PASS 🟢 model.language_model.layers.8 0.9978 66.7 990 13 → 7 PASS 🟢 model.language_model.layers.9 0.9977 65.7 682 15 → 9 PASS 🟢 model.language_model.layers.10 0.9982 80.4 898 23 → 11 PASS 🟢 model.language_model.layers.11 0.9976 79.3 985 29 → 14 PASS 🟢 model.language_model.layers.12 0.9973 79.2 899 24 → 14 PASS 🟢 model.language_model.layers.13 0.9970 78.2 978 26 → 18 PASS 🟢 model.language_model.layers.14 0.9973 77.6 798 33 → 20 PASS 🟢 model.language_model.layers.15 0.9956 76.0 674 49 → 28 PASS 🟢 model.language_model.layers.16 0.9954 75.3 967 53 → 33 PASS 🟢 model.language_model.layers.17 0.9948 74.8 467 57 → 35 PASS 🟢 model.language_model.layers.18 0.9946 73.4 839 75 → 49 PASS 🟢 model.language_model.layers.19 0.9926 71.8 362 100 → 70 PASS 🟢 model.language_model.layers.20 0.9928 71.8 555 95 → 73 PASS 🟢 model.language_model.layers.21 0.9925 71.5 824 112 → 72 PASS 🟢 model.language_model.layers.22 0.9917 69.8 907 178 → 108 PASS 🟠 model.language_model.layers.23 0.9896 69.2 667 192 → 128 WARN 🟠 model.language_model.layers.24 0.9894 69.2 885 169 → 124 WARN 🟠 model.language_model.layers.25 0.9888 68.6 415 179 → 139 WARN 🟢 model.language_model.layers.26 0.9902 68.4 541 223 → 154 PASS 🟠 model.language_model.layers.27 0.9882 67.5 382 250 → 188 WARN 🟠 model.language_model.layers.28 0.9895 67.4 772 276 → 201 WARN 🟠 model.language_model.layers.29 0.9895 67.3 828 283 → 195 WARN 🟢 model.language_model.layers.30 0.9902 66.5 739 323 → 233 PASS 🟠 model.language_model.layers.31 0.9883 49.0 438 423 → 302 WARN 🟠 model.language_model.layers.32 0.9881 47.8 783 457 → 323 WARN 🟠 model.language_model.layers.33 0.9898 48.9 814 619 → 420 WARN 🟠 model.language_model.layers.34 0.9897 58.4 988 825 → 511 WARN 🟠 model.language_model.layers.35 0.9894 49.3 982 1073 → 685 WARN 🟠 model.language_model.layers.36 0.9888 48.2 913 1142 → 830 WARN 🟢 model.language_model.layers.37 0.9901 48.5 482 1543 → 1013 PASS 🟢 model.language_model.layers.38 0.9908 55.8 503 2211 → 1449 PASS 🟢 model.language_model.layers.39 0.9909 54.7 582 3271 → 1949 PASS Summary: Total blocks: 40 Passed (🟢): 28 Warning (🟠): 12 Thresholds: Cosine Similarity < 0.99, PSNR < 45.0 dB