{ "artifacts": [ { "file": "Mordant-3B-Think_int8.litertlm", "sha256": "85205709015e412915ca86ca1e790177f204d438fbee288317b2c7cc2deb484e", "size_mb": 3587.91 } ], "benchmarks": [], "conversion": { "command": "python scripts/convert.py Kezmark/Mordant-3B-Think (hf-to-litertlm, 2026-08-25)", "quantization": "dynamic int8 on linears + embedding (README file table)", "tool": "hf-to-litertlm, one command (python scripts/convert.py Kezmark/Mordant-3B-Think, 2026-08-25) (README Conversion & verification)", "tool_version": "TODO (the README names the converter but not its version)" }, "cross_runtime": [], "delegation": null, "device": { "records": [ { "device": "galaxy-s26", "run": { "accelerator": "cpu", "context_length": null, "date": "2026-09-05", "decode_tokens_per_s": 7.7, "delegated_ops": 2172, "env": { "device": "Galaxy S26 (SM-S942Q)", "machine_label": "galaxy-s26-cold-cache-cooled", "os_build": "Android 16", "runtime": "litert-lm", "runtime_version": "0.16.0", "soc": "Qualcomm SM8850", "vendor_sdk": null }, "error": null, "evidence": [ "VERBOSE: Replacing 1619 out of 1784 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 238 partitions for subgraph 0 (prefill_1024).", "VERBOSE: Replacing 1619 out of 1784 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 238 partitions for subgraph 1 (prefill_256).", "VERBOSE: Replacing 1619 out of 1784 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 238 partitions for subgraph 2 (prefill_64).", "VERBOSE: Replacing 1619 out of 1784 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 238 partitions for subgraph 3 (prefill_16).", "results block: prefill=62.91 decode=7.7 tokens/s" ], "failure_class": null, "full_delegation": false, "latency_p50_ms": null, "loads": true, "max_abs_diff": null, "max_rel_diff": null, "metrics": { "decode_tokens": 503.0, "init_s": 5.80061, "prefill_tokens": 206.0 }, "output_match": null, "peak_mem_mb": null, "prefill_tokens_per_s": 62.91, "provenance": "measured", "runs": true, "total_ops": 2258, "ttft_ms": 3400.0 }, "source": "data/device_runs/0.16.0/2026-09-05/mordant-3b-think-int8__galaxy-s26.json" }, { "device": "galaxy-s26", "run": { "accelerator": "gpu", "context_length": null, "date": "2026-09-05", "decode_tokens_per_s": 9.59, "delegated_ops": 1784, "env": { "device": "Galaxy S26 (SM-S942Q)", "machine_label": "galaxy-s26-cold-cache-cooled", "os_build": "Android 16", "runtime": "litert-lm", "runtime_version": "0.16.0", "soc": "Qualcomm SM8850", "vendor_sdk": null }, "error": null, "evidence": [ "VERBOSE: Replacing 1784 out of 1784 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 0 (prefill_1024).", "VERBOSE: Replacing 1784 out of 1784 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 1 (prefill_256).", "VERBOSE: Replacing 1784 out of 1784 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 2 (prefill_64).", "VERBOSE: Replacing 1784 out of 1784 node(s) with delegate (LITERT_CL) node, yielding 1 partitions for subgraph 3 (prefill_16).", "results block: prefill=232.68 decode=9.59 tokens/s" ], "failure_class": null, "full_delegation": true, "latency_p50_ms": null, "loads": true, "max_abs_diff": null, "max_rel_diff": null, "metrics": { "decode_tokens": 487.0, "init_s": 14.14932, "prefill_tokens": 206.0 }, "output_match": null, "peak_mem_mb": null, "prefill_tokens_per_s": 232.68, "provenance": "measured", "runs": true, "total_ops": 1784, "ttft_ms": 990.0 }, "source": "data/device_runs/0.16.0/2026-09-05/mordant-3b-think-int8__galaxy-s26.json" }, { "device": "mac-studio-m4-max", "run": { "accelerator": "cpu", "context_length": null, "date": "2026-08-25", "decode_tokens_per_s": 20.18, "delegated_ops": null, "env": { "device": "mac-studio-m4-max", "machine_label": "mac-studio-m4-max", "os_build": null, "runtime": "litert-lm", "runtime_version": "0.16.0", "soc": "Apple M4 Max", "vendor_sdk": null }, "error": null, "evidence": [ "results block: prefill=97.25 decode=20.18 tokens/s", "token counts supplied by the caller from recorded evidence (this log states none): prefill=256 decode=256" ], "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, "prefill_tokens": 256.0 }, "output_match": null, "peak_mem_mb": null, "prefill_tokens_per_s": 97.25, "provenance": "measured", "runs": true, "total_ops": null, "ttft_ms": 2682.0 }, "source": "data/device_runs/0.16.0/2026-08-25/mordant-3b-think-int8__mac-studio-m4-max.json" }, { "device": "mac-studio-m4-max", "run": { "accelerator": "gpu", "context_length": null, "date": "2026-08-25", "decode_tokens_per_s": 71.85, "delegated_ops": null, "env": { "device": "mac-studio-m4-max", "machine_label": "mac-studio-m4-max", "os_build": null, "runtime": "litert-lm", "runtime_version": "0.16.0", "soc": "Apple M4 Max", "vendor_sdk": null }, "error": null, "evidence": [ "results block: prefill=1128.98 decode=71.85 tokens/s", "token counts supplied by the caller from recorded evidence (this log states none): prefill=256 decode=256" ], "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, "prefill_tokens": 256.0 }, "output_match": null, "peak_mem_mb": null, "prefill_tokens_per_s": 1128.98, "provenance": "measured", "runs": true, "total_ops": null, "ttft_ms": 240.7 }, "source": "data/device_runs/0.16.0/2026-08-25/mordant-3b-think-int8__mac-studio-m4-max.json" } ] }, "model": { "family": "granite", "id": "mordant-3b-think-int8", "license": "apache-2.0", "source_url": "https://huggingface.co/mlboydaisuke/Mordant-3B-Think-LiteRT", "task": "text-generation" }, "pitfalls": [ "A full fine-tune of ibm-granite/granite-4.1-3b for AI image-generation prompt composition with chain-of-thought reasoning; the finetune's thinking-form chat template (opens the assistant turn with ) is embedded verbatim (byte-equal 1474/1474) (README intro + Conversion).", "The spurious metadata start token is dropped: this family declares bos == eos == <|end_of_text|> and its template never renders a leading BOS, so an engine-prepended start token reads as 'this document already ended' — measured on this checkpoint it flips HF bf16 greedy output into a code-fence loop (README Conversion).", "Reduced 7-signature prefill ladder + externalized embedder (the >=3B ship shape; the full 11-signature ladder is killed by iOS at Metal init on this family); quality gate 8/8 (think-aware budget), non-degenerate (README Conversion).", "License Apache-2.0, inherited from the source model and its granite-4.1 base (README License).", "LiteRT-LM .litertlm bundle: LiteRT.js cannot run it, so delegation stays null and there is no browser block; device rows come from data/device_runs/." ], "schema_version": "1.2" }