{
  "artifacts": [
    {
      "file": "LFM2-700M_int4.litertlm",
      "sha256": "4ecfc71299a304e97e8f0ef5134085ecc4b5d414755b8c1709f6b552d20250b9",
      "size_mb": 458.521
    }
  ],
  "benchmarks": [],
  "conversion": {
    "command": "LiquidAI/LFM2-700M rev 86f49fc9: unquantized export (11 prefill signatures, 1024 down to 1, plus decode, over a 4096-position KV cache) -> post-hoc int4 quantization (no zero or subnormal block scale in its 92 blockwise tensors) -> metadata repack with LiteRT-LM's lfm2 configuration (card L110-L111)",
    "quantization": "int4 blockwise-32 + OCTAV on linears, int8 lm_head + embedding, convs float (card L28)",
    "tool": "litert-torch 0.9.4; repro = hf-to-litertlm lfm_work/convert_lfm2_700m.sh (card L110)",
    "tool_version": "litert-torch 0.9.4, litert-converter 0.4.0, ai-edge-quantizer 0.9.0, transformers 5.14.1, torch 2.13.0; packed by the litert-lm 0.18.0 CLI"
  },
  "cross_runtime": [],
  "delegation": null,
  "device": {
    "records": [
      {
        "device": "galaxy-s26",
        "run": {
          "accelerator": "cpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 57.12,
          "delegated_ops": 788,
          "env": {
            "device": "Galaxy S26 (SM-S942Q)",
            "machine_label": "galaxy-s26-aab1",
            "os_build": "Android 16",
            "runtime": "litert-lm",
            "runtime_version": "prebuilt-adac974b",
            "soc": "Qualcomm SM8850",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "binary: Google's prebuilt litert_lm_advanced_main + 6 .so (gs://litert/binaries/latest/android_arm64/litert_lm/, object dated 2026-09-18, 40,382,152 B, sha256 adac974b…)",
            "r1 hold 2026-10-12 20:06:49-20:22:44 JST: [20:14:09] bundle on device: 4ecfc71299a304e97e8f0ef5134085ecc4b5d414755b8c1709f6b552d20250b9 (480794160 B, push 12.1 s) = Mac",
            "bench args: --backend=cpu --model_path=/data/local/tmp/aa2_700m/LFM2-700M_int4.litertlm --benchmark=true --benchmark_prefill_tokens=1024 --benchmark_decode_tokens=256 --max_num_tokens=2048 --num_iterations=3 --report_peak_memory_footprint=true --metric_proto_file_path=/data/local/tmp/aa2_700m/metrics.pb",
            "VERBOSE: Replacing 487 out of 560 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 88 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 455 out of 519 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 92 partitions for subgraph 11 (decode).",
            "VERBOSE: Replacing 788 out of 818 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 14 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 737 out of 756 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 14 partitions for subgraph 11 (decode).",
            "[20:16:24] parsed: i4_cpu_timing cpu: gate PASS 1/1 | bench prefill 346.84000000000003 decode 57.120000000000005 init 1218.48 cold 1594.2 peak 1652.52 capped 10/11 | answers: 'The capital of France is Paris.'",
            "timing run prompt process: \"The capital of France is Paris.\""
          ],
          "failure_class": null,
          "full_delegation": false,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "boot_epoch_s": 1791590918.91,
            "capped_samples": 10.0,
            "cold_init_ms": 1594.2,
            "cpu_capped_samples": 10.0,
            "decode_tok_s_it1": 69.58,
            "decode_tok_s_it2": 59.24,
            "decode_tok_s_it3": 55.0,
            "decode_tokens": 256.0,
            "init_ms": 1218.48,
            "iterations": 3.0,
            "kgsl_ceiling_min_mhz": 726.0,
            "kgsl_temp_max_c": 71.7,
            "max_num_tokens": 2048.0,
            "memavail_at_bench_start_kib": 7574796.0,
            "memavail_min_kib": 6391812.0,
            "prefill_tok_s_it1": 435.73,
            "prefill_tok_s_it2": 406.11,
            "prefill_tok_s_it3": 287.57,
            "prefill_tokens": 1024.0,
            "sampler_samples": 11.0,
            "skin_c_at_bench_start": 45.0,
            "t_1024_256_s": 7.4342,
            "thermal_status_at_bench_start": 3.0,
            "ttft_s_it1": 2.36,
            "ttft_s_it2": 2.54,
            "ttft_s_it3": 3.58
          },
          "output_match": null,
          "peak_mem_mb": 1652.52,
          "prefill_tokens_per_s": 346.84,
          "provenance": "measured",
          "runs": true,
          "total_ops": 818,
          "ttft_ms": 3060.0
        },
        "source": "data/device_runs/prebuilt-adac974b/2026-10-12/lfm2-700m-int4__galaxy-s26.json"
      },
      {
        "device": "galaxy-s26",
        "run": {
          "accelerator": "gpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 73.25,
          "delegated_ops": 560,
          "env": {
            "device": "Galaxy S26 (SM-S942Q)",
            "machine_label": "galaxy-s26-aab1",
            "os_build": "Android 16",
            "runtime": "litert-lm",
            "runtime_version": "prebuilt-adac974b",
            "soc": "Qualcomm SM8850",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "binary: Google's prebuilt litert_lm_advanced_main + 6 .so (gs://litert/binaries/latest/android_arm64/litert_lm/, object dated 2026-09-18, 40,382,152 B, sha256 adac974b…)",
            "r3 hold 2026-10-12 21:40:50-21:47:53 JST: [21:42:59] bundle on device: 4ecfc71299a304e97e8f0ef5134085ecc4b5d414755b8c1709f6b552d20250b9 (480794160 B, push 12.4 s) = Mac",
            "bench args: --backend=gpu --model_path=/data/local/tmp/aa2_700m/LFM2-700M_int4.litertlm --benchmark=true --benchmark_prefill_tokens=1024 --benchmark_decode_tokens=256 --max_num_tokens=2048 --num_iterations=3 --report_peak_memory_footprint=true --metric_proto_file_path=/data/local/tmp/aa2_700m/metrics.pb",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 1 (prefill_512).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 2 (prefill_256).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 3 (prefill_128).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 4 (prefill_64).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 5 (prefill_32).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 6 (prefill_16).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 7 (prefill_8).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 8 (prefill_4).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 9 (prefill_2).",
            "VERBOSE: Replacing 501 out of 501 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 10 (prefill_1).",
            "VERBOSE: Replacing 519 out of 519 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 11 (decode).",
            "[21:44:38] parsed: i4_gpu_timing_rested gpu: gate PASS 1/1 | bench prefill 2427.335 decode 73.25 init 1821.29 cold 4966.79 peak 534.43 capped 2/7 | answers: 'The capital of France is Paris.'",
            "timing run prompt process: \"The capital of France is Paris.\""
          ],
          "failure_class": null,
          "full_delegation": true,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "boot_epoch_s": 1791590918.66,
            "capped_samples": 2.0,
            "cold_init_ms": 4966.79,
            "cpu_capped_samples": 5.0,
            "decode_tok_s_it1": 69.22,
            "decode_tok_s_it2": 73.52,
            "decode_tok_s_it3": 72.98,
            "decode_tokens": 256.0,
            "init_ms": 1821.29,
            "iterations": 3.0,
            "kgsl_ceiling_min_mhz": 1100.0,
            "kgsl_temp_max_c": 84.9,
            "max_num_tokens": 2048.0,
            "memavail_at_bench_start_kib": 7061784.0,
            "memavail_min_kib": 5444904.0,
            "prefill_tok_s_it1": 2494.64,
            "prefill_tok_s_it2": 2512.62,
            "prefill_tok_s_it3": 2342.05,
            "prefill_tokens": 1024.0,
            "sampler_samples": 7.0,
            "skin_c_at_bench_start": 37.9,
            "t_1024_256_s": 3.9167,
            "thermal_status_at_bench_start": 0.0,
            "ttft_s_it1": 0.42,
            "ttft_s_it2": 0.42,
            "ttft_s_it3": 0.45
          },
          "output_match": null,
          "peak_mem_mb": 534.43,
          "prefill_tokens_per_s": 2427.335,
          "provenance": "measured",
          "runs": true,
          "total_ops": 560,
          "ttft_ms": 435.0
        },
        "source": "data/device_runs/prebuilt-adac974b/2026-10-12/lfm2-700m-int4__galaxy-s26.json"
      },
      {
        "device": "mac-studio-m4-max",
        "run": {
          "accelerator": "cpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 105.965,
          "delegated_ops": null,
          "env": {
            "device": "Mac Studio (M4 Max)",
            "machine_label": "mac-studio-m4-max",
            "os_build": "macOS 27.0",
            "runtime": "litert-lm",
            "runtime_version": "0.18.0",
            "soc": "Apple M4 Max",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "Prefill 599.41 / decode 112.19 tok/s, TTFT 1.7175 s, init 2.3496 s",
            "Prefill 562.0 / decode 99.74 tok/s, TTFT 1.8477 s, init 2.3285 s",
            "2026-10-12 (cpu window) cpu int4 p1 exit 0 Prefill speed: 599.41 tokens/s Decode speed: 112.19 tokens/s",
            "2026-10-12 (cpu window) cpu int4 p2 exit 0 Prefill speed: 562.00 tokens/s Decode speed: 99.74 tokens/s"
          ],
          "failure_class": null,
          "full_delegation": null,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "decode_tokens": 256.0,
            "init_s": 2.3391,
            "iterations": 3.0,
            "max_num_tokens": 2048.0,
            "prefill_tokens": 1024.0,
            "processes": 2.0,
            "run1_decode_tok_s": 112.19,
            "run1_init_s": 2.3496,
            "run1_prefill_tok_s": 599.41,
            "run1_ttft_s": 1.7175,
            "run2_decode_tok_s": 99.74,
            "run2_init_s": 2.3285,
            "run2_prefill_tok_s": 562.0,
            "run2_ttft_s": 1.8477,
            "t_1024_256_s": 4.1793
          },
          "output_match": null,
          "peak_mem_mb": null,
          "prefill_tokens_per_s": 580.705,
          "provenance": "measured",
          "runs": true,
          "total_ops": null,
          "ttft_ms": 1782.6
        },
        "source": "data/device_runs/0.18.0/2026-10-12/lfm2-700m-int4__mac-studio-m4-max.json"
      },
      {
        "device": "mac-studio-m4-max",
        "run": {
          "accelerator": "gpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 348.26,
          "delegated_ops": null,
          "env": {
            "device": "Mac Studio (M4 Max)",
            "machine_label": "mac-studio-m4-max",
            "os_build": "macOS 27.0",
            "runtime": "litert-lm",
            "runtime_version": "0.18.0",
            "soc": "Apple M4 Max",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "Prefill 7839.15 / decode 348.26 tok/s, TTFT 0.1335 s, init 1.4619 s",
            "2026-10-12 19:54:06 JST gpu int4 verbose exit 0 validation_error=0 shape_mismatch=0 replacing=24",
            "2026-10-12 19:53:45 JST gpu int4 exit 0"
          ],
          "failure_class": null,
          "full_delegation": null,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "decode_tokens": 256.0,
            "init_s": 1.4619,
            "iterations": 3.0,
            "max_num_tokens": 2048.0,
            "prefill_tokens": 1024.0,
            "processes": 1.0,
            "t_1024_256_s": 0.8657
          },
          "output_match": null,
          "peak_mem_mb": null,
          "prefill_tokens_per_s": 7839.15,
          "provenance": "measured",
          "runs": true,
          "total_ops": null,
          "ttft_ms": 133.5
        },
        "source": "data/device_runs/0.18.0/2026-10-12/lfm2-700m-int4__mac-studio-m4-max.json"
      },
      {
        "device": "pixel-8a",
        "run": {
          "accelerator": "cpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 30.325,
          "delegated_ops": 788,
          "env": {
            "device": "Pixel 8a",
            "machine_label": "pixel-8a-aab1",
            "os_build": "Android 16",
            "runtime": "litert-lm",
            "runtime_version": "prebuilt-adac974b",
            "soc": "Tensor G3",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "binary: Google's prebuilt litert_lm_advanced_main + 6 .so (gs://litert/binaries/latest/android_arm64/litert_lm/, object dated 2026-09-18, 40,382,152 B, sha256 adac974b…)",
            "r1 hold 2026-10-12 20:40:55-21:06:15 JST: [20:56:16] bundle on device: 4ecfc71299a304e97e8f0ef5134085ecc4b5d414755b8c1709f6b552d20250b9 (480794160 B, push 10.2 s) = Mac",
            "bench args: --backend=cpu --model_path=/data/local/tmp/aa2_700m/LFM2-700M_int4.litertlm --benchmark=true --benchmark_prefill_tokens=1024 --benchmark_decode_tokens=256 --max_num_tokens=2048 --num_iterations=3 --report_peak_memory_footprint=true --metric_proto_file_path=/data/local/tmp/aa2_700m/metrics.pb",
            "VERBOSE: Replacing 487 out of 560 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 88 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 455 out of 519 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 92 partitions for subgraph 11 (decode).",
            "VERBOSE: Replacing 788 out of 818 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 14 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 737 out of 756 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 14 partitions for subgraph 11 (decode).",
            "[21:00:51] parsed: i4_cpu_timing cpu: gate PASS 1/1 | bench prefill 152.95499999999998 decode 30.325 init 2620.46 cold 3054.77 peak 1250.67 capped 13/17 | answers: 'The capital of France is Paris.'",
            "timing run prompt process: \"The capital of France is Paris.\""
          ],
          "failure_class": null,
          "full_delegation": false,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "boot_epoch_s": 1791442100.14,
            "capped_samples": 13.0,
            "cold_init_ms": 3054.77,
            "cpu_capped_samples": 13.0,
            "decode_tok_s_it1": 32.45,
            "decode_tok_s_it2": 31.99,
            "decode_tok_s_it3": 28.66,
            "decode_tokens": 256.0,
            "init_ms": 2620.46,
            "iterations": 3.0,
            "max_num_tokens": 2048.0,
            "memavail_at_bench_start_kib": 3978500.0,
            "memavail_min_kib": 3245492.0,
            "prefill_tok_s_it1": 171.77,
            "prefill_tok_s_it2": 166.68,
            "prefill_tok_s_it3": 139.23,
            "prefill_tokens": 1024.0,
            "sampler_samples": 17.0,
            "t_1024_256_s": 15.1367,
            "thermal_status_at_bench_start": 1.0,
            "ttft_s_it1": 5.99,
            "ttft_s_it2": 6.17,
            "ttft_s_it3": 7.39
          },
          "output_match": null,
          "peak_mem_mb": 1250.67,
          "prefill_tokens_per_s": 152.955,
          "provenance": "measured",
          "runs": true,
          "total_ops": 818,
          "ttft_ms": 6780.0
        },
        "source": "data/device_runs/prebuilt-adac974b/2026-10-12/lfm2-700m-int4__pixel-8a.json"
      },
      {
        "device": "pixel-8a",
        "run": {
          "accelerator": "gpu",
          "context_length": null,
          "date": "2026-10-12",
          "decode_tokens_per_s": 39.905,
          "delegated_ops": 560,
          "env": {
            "device": "Pixel 8a",
            "machine_label": "pixel-8a-aab1",
            "os_build": "Android 16",
            "runtime": "litert-lm",
            "runtime_version": "prebuilt-adac974b",
            "soc": "Tensor G3",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "binary: Google's prebuilt litert_lm_advanced_main + 6 .so (gs://litert/binaries/latest/android_arm64/litert_lm/, object dated 2026-09-18, 40,382,152 B, sha256 adac974b…)",
            "r2 hold 2026-10-12 21:25:52-21:36:25 JST: [21:32:54] bundle on device: 4ecfc71299a304e97e8f0ef5134085ecc4b5d414755b8c1709f6b552d20250b9 (480794160 B, push 10.1 s) = Mac",
            "bench args: --backend=gpu --model_path=/data/local/tmp/aa2_700m/LFM2-700M_int4.litertlm --benchmark=true --benchmark_prefill_tokens=1024 --benchmark_decode_tokens=256 --max_num_tokens=2048 --num_iterations=3 --report_peak_memory_footprint=true --metric_proto_file_path=/data/local/tmp/aa2_700m/metrics.pb",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 0 (prefill_1024).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 1 (prefill_512).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 2 (prefill_256).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 3 (prefill_128).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 4 (prefill_64).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 5 (prefill_32).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 6 (prefill_16).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 7 (prefill_8).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 8 (prefill_4).",
            "VERBOSE: Replacing 560 out of 560 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 9 (prefill_2).",
            "VERBOSE: Replacing 501 out of 501 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 10 (prefill_1).",
            "VERBOSE: Replacing 519 out of 519 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 11 (decode).",
            "[21:35:34] parsed: i4_gpu_timing gpu: gate PASS 1/1 | bench prefill 619.655 decode 39.905 init 3123.54 cold 8950.39 peak 1026.02 capped 0/11 | answers: 'The capital of France is Paris.'",
            "timing run prompt process: \"The capital of France is Paris.\""
          ],
          "failure_class": null,
          "full_delegation": true,
          "latency_p50_ms": null,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "boot_epoch_s": 1791442099.93,
            "capped_samples": 0.0,
            "cold_init_ms": 8950.39,
            "cpu_capped_samples": 5.0,
            "decode_tok_s_it1": 37.39,
            "decode_tok_s_it2": 39.94,
            "decode_tok_s_it3": 39.87,
            "decode_tokens": 256.0,
            "init_ms": 3123.54,
            "iterations": 3.0,
            "max_num_tokens": 2048.0,
            "memavail_at_bench_start_kib": 4194456.0,
            "memavail_min_kib": 2791452.0,
            "prefill_tok_s_it1": 606.52,
            "prefill_tok_s_it2": 620.83,
            "prefill_tok_s_it3": 618.48,
            "prefill_tokens": 1024.0,
            "sampler_samples": 11.0,
            "t_1024_256_s": 8.0678,
            "thermal_status_at_bench_start": 1.0,
            "ttft_s_it1": 1.72,
            "ttft_s_it2": 1.67,
            "ttft_s_it3": 1.68
          },
          "output_match": null,
          "peak_mem_mb": 1026.02,
          "prefill_tokens_per_s": 619.655,
          "provenance": "measured",
          "runs": true,
          "total_ops": 560,
          "ttft_ms": 1675.0
        },
        "source": "data/device_runs/prebuilt-adac974b/2026-10-12/lfm2-700m-int4__pixel-8a.json"
      }
    ]
  },
  "model": {
    "family": "lfm2",
    "id": "lfm2-700m-int4",
    "license": "other (lfm1.0)",
    "source_url": "https://huggingface.co/litert-community/LFM2-700M",
    "task": "text-generation"
  },
  "pitfalls": [
    "Runtime: tested on litert-lm 0.18.0 on a Mac; this file was not tried on litert-lm 0.17.1, where the int8 file of the same conversion answered as if no question had been asked (card L21).",
    "Template: Liquid AI's chat template fails at render on litert-lm 0.18.0 (message.get(\"content\")), so the bundle carries LiteRT-LM's lfm2 template (byte-identical to models/lfm2/chat_template.jinja at commit 86a7e6d0), which reads message content as a list of parts; stops <|im_end|> and <|endoftext|>, start token <|startoftext|>; marked non-thinking, with the configuration's thought channel kept and no reply in the checks containing its strings (card L71, L112).",
    "Tool calling is declared (model type lfm2): the file returned a parsed get_weather call (Python API, CPU, greedy) and used its result with automatic tool calling; with tools the template omits Liquid AI's <|tool_list_start|> / <|tool_list_end|> and <|tool_response_start|> / <|tool_response_end|> markers, wraps the tool result in [ … ] and quotes call arguments with double quotes (card L73).",
    "One Engine per conversation: the card advises a new Engine for each conversation (its probe on the int8 file, two different system prompts, found no change on a shared Engine; LiteRT-LM issue #3165 reports recurrent state reused across conversations that share prompt tokens) (card L75).",
    "Conversation preface with tools and no messages: LiteRT-LM renders the chat template with an empty message list, and the template reads messages[0] without checking that the list is non-empty, so the render fails with 'Failed to apply template: undefined value' when the preface is prefilled at creation (prefillPrefaceOnInit, default false; creating the conversation fails) and in the preface render (renderPrefaceIntoString and its C++ and C API forms); the same with extra context (such as enable_thinking) and no messages, and in the preface render with an empty preface (run in the runtime on one Qwen3 template; the other templates rendered with minijinja 2.14 and jinja2); read in the source of LiteRT-LM 0.17.0, 0.18.1 and main, measured with the 0.18.1 PyPI package through its C API on these files' template text; the default path, where the first message is rendered with the preface, is not affected (the 0.18.1 Python API and CLI expose neither the flag nor the preface render); a system message or any initial message in the conversation config avoids it (card L116).",
    "Sampler: the bundle's default sampler is temperature 0.1, top-k 50, top-p 1.0; Liquid AI recommends temperature 0.3, min_p 0.15 and repetition_penalty 1.05, which the bundle does not carry (no min_p or repetition penalty); --top-k 1 gives greedy output (card L69).",
    "Cache: without --cache, `run` writes a compiled-weights cache next to the model (465 MB for this file on CPU); without --backend, `run` uses the CPU (card L67).",
    "Speed: this file is faster than int8 on the Mac and Galaxy S26 GPUs (0.87 vs 0.97 s and 3.92 vs 4.64 s for a 1024-token prefill plus 256 decoded tokens) and equal on the Pixel 8a GPU (8.07 s each); on the CPUs it is faster on the Pixel 8a (15.13 vs 16.21 s) and slower on the Mac and the Galaxy S26; 40.5 % smaller than the int8 file; not measured on iOS, not run on the NPU (card L30-L39, L106).",
    "Galaxy S26 rows of this file: the GPU row ran with the GPU clock limit falling from 1300 to 1100 MHz in the last 2 of 7 samples (the unflagged first iteration read 2494.6 / 69.22 tok/s); the CPU row is from an earlier run that started on a hot phone (thermal status 3), the CPU clock lowered in 10 of 11 samples (card L93-L97).",
    "Pixel 8a: every run started at thermal status 1; the CPU clock was lowered in 13 of 17 samples of the CPU run and 5 of 11 of the GPU run (card L106).",
    "Accuracy (greedy, litert-lm on the Mac CPU, comparable only within the card's table): IFEval-lite 67 of 120 (bf16 77), GSM8K 52 and 57 of 100 with two wordings of the prompt (bf16 59 and 59); 8-question check 7/8 on CPU and GPU, the miss being the rhyme question as for bf16 (card L43, L49-L55).",
    "License: the LFM Open License v1.0's section 5 conditions commercial use on an annual revenue below 10 million US dollars (card L120)."
  ],
  "schema_version": "1.2"
}
