{"id":"run_Y576sLBflzGiCVyS","owner":"anonymous-EVjB2S","publishedAt":"2026-08-31T04:53:43.808Z","manifest":{"schema_version":"1.0","kind":"run","client_run_id":"rps-49be6801e9cc80accba2e79bf9959ebb1594311b","name":"meta-models/Muse-Glimmer-30B-GGUF | baseline | 256->1024 | C8","notes":"Aggregate-only synthetic online serving measurement. It reports serving performance and does not evaluate model quality.","visibility":"public","license":"CC-BY-4.0","outcome":"completed","subject":{"type":"model","source":"huggingface","adapters":[],"revision":"70bf1b61ac09f91b24d39038091b41c582bc5d7a","identifier":"meta-models/Muse-Glimmer-30B-GGUF","modalities":["text"],"architecture":"MuseGlimmerForConditionalGeneration with DFlash","quantization":{"format":"KQuant Dynamic Q4_K_XL target plus Q4_K_M DFlash"},"modifications":["Speculative dflash-draft checkpoint: meta-models/Muse-Glimmer-30B-GGUF@70bf1b61ac09f91b24d39038091b41c582bc5d7a."],"revision_kind":"commit","parameter_count_total":29776626688,"parameter_count_active":29776626688},"environment":{"os":"Linux-6.8.0-136-generic-x86_64-with-glibc2.39","cpu":{"model":"x86_64"},"kernel":"6.8.0-136-generic","topology":"One server engine on one GPU.","accelerators":[{"count":1,"model":"NVIDIA GeForce RTX 5090","driver":"580.173.02","vendor":"NVIDIA","architecture":"Blackwell SM120","interconnect":"single device","memory_bytes_per_device":34190917632}],"driver_versions":{"nvidia":"580.173.02"},"host_architecture":"x86_64","system_memory_bytes":269585743872},"runtime":{"engine":"llama.cpp","version":"b10689","extensions":{"runpile.server_variant":"baseline","runpile.auxiliary_models":[{"role":"dflash-draft","revision":"70bf1b61ac09f91b24d39038091b41c582bc5d7a","repository":"meta-models/Muse-Glimmer-30B-GGUF","quantization":"Q4_K_M","weight_bytes":1631208128,"artifact_sha256":"b2e808bf656086fe86bd0d0bd990f01d33e377537a07c02d45371517c8b264ef","artifact_filename":"dflash-Muse-Glimmer-30B-Q4_K_M.gguf"}]},"parameters":{"server":{"speculative":{"type":"draft-dflash","n_max":15},"ubatch_size":512,"context_size":147456,"max_num_seqs":16,"flash_attention":"on","weight_artifact":"Muse-Glimmer-30B-KQuant-Dynamic-Q4_K_XL.gguf","max_model_len_per_slot":9216,"max_num_batched_tokens":16384}},"parallelism":{"data":1,"expert":1,"tensor":1,"pipeline":1,"replicas":1},"launch_command":"CUDA_VISIBLE_DEVICES=0 <local-path> serve --model meta-models/Muse-Glimmer-30B-GGUF --alias meta-models/Muse-Glimmer-30B-GGUF --host 127.0.0.1 --port 9000 --ctx-size 147456 --parallel 16 --batch-size 16384 --ubatch-size 512 --n-gpu-layers all --flash-attn on --jinja --metrics --no-webui --no-cache-prompt --cache-ram 0 --no-cache-idle-slots --ctx-checkpoints 0 --model-draft meta-models/Muse-Glimmer-30B-GGUF --spec-type draft-dflash --spec-draft-n-max 15 --n-gpu-layers-draft all","build_identifier":"ghcr.io/ggml-org/llama.cpp@sha256:9fe590cd7c117a0ba171bc2979bb8a31d2bc71247c9ca6612f4634372a0feecb","safe_environment":{"HF_HUB_OFFLINE":"1","CUDA_VISIBLE_DEVICES":"0","TOKENIZERS_PARALLELISM":"false"}},"workload":{"mode":"online","task":"text_generation","client":{"name":"vllm bench serve","backend":"vllm","version":"0.28.0","placement":"same-host"},"dataset":{"sha256":"f60e32b508001f401d8ec671c7740634fd6b0748d6a35bae98aa87c266d47b86","generator":"runpile-sweeps-tokenizer-verified-jsonl","synthetic":true,"generator_revision":"a3d215d3624741ac0031b7b6ec24bc4600b7dadd2dcdf9ba58f18e56ea0e3b64","shared_prefix_tokens":0},"sampling":{"mode":"sampling","top_k":64,"top_p":0.95,"temperature":1},"stopping":{"ignore_eos":true,"max_output_tokens":1024},"streaming":true,"tokenizer":{"revision":"a4e59da52a7bc87ae7251dd5545c0dd437c44b68","identifier":"meta-models/Muse-Glimmer-30B"},"extensions":{"runpile.profile_id":"serve-fixed-256-1k-v1"},"concurrency":8,"random_seed":0,"input_tokens":{"kind":"fixed","value":256},"endpoint_type":"openai-compatible-completions","output_tokens":{"kind":"fixed","value":1024},"request_count":64,"arrival_process":"closed_loop","prefix_behavior":{"cache_state":"warm-host","cache_enabled":false,"shared_prefix_tokens":0},"warmup_request_count":4,"request_rate_per_second":null},"measurement":{"notes":"Aggregate-only output.","warmup":"4 unmeasured warmups per invocation after one readiness request.","harness":"vllm bench serve","version":"0.28.0","extensions":{"runpile.declared_warmup_count":4,"runpile.max_gpu_temperature_c":65,"runpile.preemption_log_mentions":0,"runpile.readiness_request_count":1,"runpile.controller_source_sha256":"a3d215d3624741ac0031b7b6ec24bc4600b7dadd2dcdf9ba58f18e56ea0e3b64","runpile.speculative_accept_length":15.98047323684815,"runpile.independent_repetition_index":2},"repetitions":1,"timing_scope":"client-observed streaming request latency","benchmark_command":"vllm bench serve --backend vllm --base-url http://127.0.0.1:9000 --endpoint /v1/completions --model meta-models/Muse-Glimmer-30B-GGUF --tokenizer meta-models/Muse-Glimmer-30B --dataset-name custom --dataset-path <EXACT_SYNTHETIC_DATASET> --custom-output-len 1024 --skip-chat-template --disable-shuffle --num-prompts 64 --max-concurrency 8 --request-rate inf --seed 0 --ignore-eos --temperature 1.0 --top-p 0.95 --top-k 64 --percentile-metrics ttft,tpot,itl,e2el --metric-percentiles 50,90,95,99 --disable-tqdm --save-result --result-dir <local-path> --result-filename glimmer30b-dynq4-dflash15-gpu0__baseline__serve-fixed-256-1k-v1__c8__r2__attempt1.json --num-warmups 4 --metadata campaign=runpile-rtx5090-glimmer-gguf-dflash15-20260831-v3 config_id=glimmer30b-dynq4-dflash15-gpu0 server_variant=baseline workload_key=serve-fixed-256-1k-v1 repetition=2 aggregate_only=true","retry_failure_treatment":"Failed requests counted; no client retries.","client_overhead_included":true},"results":{"latency_ms":{"end_to_end":{"p50":5623.486869037151,"p90":6338.239907007664,"p95":6434.030976914801,"p99":6577.432154146954,"mean":5805.846097973699,"stddev":438.97560486076634,"sample_count":64,"measurement_status":"measured"},"inter_token":{"p50":0.031147850677371025,"p90":0.07029715925455093,"p95":68.78436962142587,"p99":86.22890450060368,"mean":5.280496153204695,"stddev":22.30220682466054,"sample_count":64,"measurement_status":"measured"},"time_to_first_token":{"p50":372.1389069687575,"p90":658.5996897425503,"p95":659.5333974808455,"p99":681.4976440696046,"mean":398.618256986083,"stddev":142.50008091880122,"sample_count":64,"measurement_status":"measured"},"time_per_output_token":{"p50":5.194534632520257,"p90":5.84599502746085,"p95":5.902804844063194,"p99":6.004070588455827,"mean":5.285657713575382,"stddev":0.43928229852121853,"sample_count":64,"measurement_status":"measured"}},"throughput":{"requests_per_second":{"value":1.3710025431461976,"sample_count":64,"measurement_status":"measured"},"total_tokens_per_second":{"value":1754.8832552271328,"sample_count":64,"measurement_status":"measured"},"output_tokens_per_second":{"value":1403.9066041817064,"sample_count":64,"measurement_status":"measured"}},"failed_requests":0,"duration_seconds":46.68116796715185,"total_input_tokens":16384,"successful_requests":64,"total_output_tokens":65536},"artifacts":[{"kind":"vllm_bench_serve_json","sha256":"5f56931398319e5f7923072bc750390aebf2fce0bf7164c545c897ba51f544bd","license":"CC-BY-4.0","filename":"glimmer30b-dynq4-dflash15-gpu0__baseline__serve-fixed-256-1k-v1__c8__r2.json","media_type":"application/json","size_bytes":2971,"visibility":"private","client_artifact_id":"raw-vllm-serve","raw_requests_included":false}],"provenance":{"tool":"runpile HTTP importer","source":"import","tool_version":"1.0"},"warnings":["Review imported facts against the local model, hardware, launch command, and benchmark command before publication."]},"indexingState":"indexed","resultKey":"res_UTL6TNMDJj8oE8tzeWBCXxv0","eligibilityReasons":[],"artifactBacked":true,"downloads":[]}