{ "case": "base-q2_k", "machine": "macos-15-arm64", "run": "ci-1", "args": [ "--model", "models/qwen2.5-0.5b-instruct-q2_k.gguf", "--contextsize", "4096" ], "started": "2026-11-13T12:40:12+0000", "start": { "ready": true, "seconds": 29.64, "model": { "result": "koboldcpp/qwen2.5-0.5b-instruct-q2_k" } }, "gen": [ { "status": 200, "wall_s": 19.566, "text_head": " looked at the lighthouse. The lighthouse was still lit, but the light was dim. She had been there for a long time, and it was time to check the machine. She wa", "prompt_tokens": 38, "gen_tokens": 128, "pp_tps": 94.76, "tg_tps": 6.69 }, { "status": 200, "wall_s": 21.666, "text_head": " looked at the lighthouse. She had never seen anything like it before. The lighthouse was a beacon of hope, a beacon of light that guided ships to safety. She h", "prompt_tokens": 38, "gen_tokens": 128, "pp_tps": 7.14, "tg_tps": 5.98 }, { "status": 200, "wall_s": 23.01, "text_head": " looked at the lighthouse. She had never seen anything like it before. The lighthouse was a beacon of hope, a beacon of light that guided ships to safety. She h", "prompt_tokens": 38, "gen_tokens": 128, "pp_tps": 9.8, "tg_tps": 5.61 } ], "long": [ { "status": 200, "wall_s": 21.886, "text_head": " The quick brown fox jumps over the lazy dog near the old stone bridge. The quick brown fox jumps over the lazy dog near the old stone bridge. The quick", "prompt_tokens": 1650, "gen_tokens": 32, "pp_tps": 108.44, "tg_tps": 4.91 }, { "status": 200, "wall_s": 4.86, "text_head": " The quick brown fox jumps over the lazy dog near the old stone bridge. The quick brown fox jumps over the lazy dog near the old stone bridge. The quick", "prompt_tokens": 1650, "gen_tokens": 32, "pp_tps": 4.74, "tg_tps": 7.09 } ], "version": { "result": "KoboldCpp", "version": "1.124.2", "protected": false, "llm": true, "txt2img": false, "vision": false, "audio": false, "transcribe": false, "multiplayer": false, "websearch": false, "tts": false, "embeddings": false, "music": false, "musicllm": false, "musicfamily": "none", "savedata": false, "admin": 0, "router": false, "guidance": false, "jinja": false, "mcp": false, "parallelrequests": 1, "defaults_modified": false }, "peak_rss_mb": 671.4, "log_lines": [ "Unable to determine GPU Memory", "Unable to determine available RAM", "Initializing dynamic library: koboldcpp_default.so", "Namespace(usecuda=None, usevulkan=None, usecpu=False, model=['models/qwen2.5-0.5b-instruct-q2_k.gguf'], model_param='models/qwen2.5-0.5b-instruct-q2_k.gguf', config=None, contextsize=4096, gpulayers=200, host='', launch=False, tensor_split=None, threads=1, port=49178, port_param=49178, version=False, agent=False, agentload='', analyze='', autofit=False, autofitpadding=1024, batchsize=512, ubatchsize=-1, benchmark=None, blasthreads=0, chatcompletionsadapter='AutoGuess', cli=False, debugmode=0, defaultgenamt=2048, device='', downloaddir='', draftamount=4, draftgpulayers=999, draftgpusplit=None, draftmodel='', enableguidance=False, exportconfig='', exporttemplate='', failsafe=False, foreground=False, gendefaults='', gendefaultsoverwrite=False, genlimit=0, highpriority=False, ignoremissing=False, jinja=False, jinja_kwargs='', jinjatemplate='', jinjathink='default', lora=None, loramult=1.0, lowvram=False, maingpu=-1, maxrequestsize=32, mcpfile='', mmproj='', mmprojcpu=False, lazy_mode='auto', moe_cache_mib=0, moecpu=0, ffncpu=0, moeexperts=-1, multiplayer=False, nocertify=False, nofastforward=False, noflashattention=False, noavx2=False, nobostoken= nommq=False, nomodel=False, noshift=False, noswa=False, onready='', allow_config_onready=False, overridekv='', overridenativecontext=0, overridetensors='', parallelrequests=1, paralleltotalcontext=0, password= preloadstory='', prompt='', quantkv='f16', queuelimit=10, quiet=False, ratelimit=0, reasoningeffort='default', remotetunnel=False, ropeconfig=[0.0, 10000.0], savedatafile='', singleinstance=False, smartcache=0, smartcontext=False, splitmode='layer', ssl=None, swapadding=0, unpack='', usemtp=False, usemlock=False, usemmap=False, usedirectio=False, visionmaxres=1024, visionmaxtokens=-1, visionmintokens=-1, websearch=False, showgui=False, skiplauncher=True, hordemodelname='', hordeworkername='', hordekey='', hordemaxctx=0, hordegenlen=0, sdaudiovae='', sdclamped=0, sdclampedsoft=0, sdclip1='', sdclip2='', sdclipdevice=-2, sdconvdirect='off', sdflashattention=False, sdlora=[], sdloramult=[1.0], sdmaingpu=None, sdmodel='', sdoffloadcpu=False, sdphotomaker='', sdquant=0, sdllm='', sdthreads=0, sdtiledvae=512, sdupscaler='', sdvae='', sdvaeauto=False, sdvaedevice=-1, sdvramlimit=0, whispermodel='', ttsmodel='', ttswavtokenizer='', ttsgpu=False, ttsmaxlen=4096, ttsthreads=0, ttsdir='', musicllm='', musicembeddings='', musicdiffusion='', musicvae='', musiclowvram=False, embeddingsmodel='', embeddingsmaxctx=4096, embeddingsgpu=False, rpcmode='disabled', rpcport=5551, rpchost='0.0.0.0', rpcdevice='', rpccache='', rpctargets='', admin=False, adminpassword= admindir='', adminunloadtimeout=0, routermode=False, reqtimeout=600, autoswapmode=False, autoswapthreshold=256, baseconfig='', hordeconfig=None, sdconfig=None, noblas=False, nommap=False, nopipelineparallel=False, pipelineparallel=False, sdnotile=False, sdt5xxl=None, sdvaecpu=False, sdclipgpu=False, forceversion=False, sdgendefaults=False, flashattention=False, useswa=False, multiuser=None, testmemory=False, run_bundled_agent=False, agent_base_url=None, agent_api_key= proxy_port=None)", "print_info: file size = 390.28 MiB (5.20 BPW)", "print_info: n_ctx_train = 32768", "print_info: n_ctx_orig_yarn = 32768", "done_getting_tensors: tensor 'token_embd.weight' (iq4_nl) (and 169 others) moved from CPU_REPACK, using CPU instead", "load_tensors: CPU model buffer size = 117.69 MiB", "load_tensors: CPU_REPACK model buffer size = 272.31 MiB", "llama_context: n_ctx = 4352", "llama_context: n_ctx_seq = 4352", "llama_context: flash_attn = enabled", "llama_context: n_ctx_seq (4352) < n_ctx_train (32768) -- the full capacity of the model will not be utilized", "llama_kv_cache: MTL0 KV buffer size = 51.00 MiB", "sched_reserve: MTL0 compute buffer size = 10.50 MiB", "sched_reserve: CPU compute buffer size = 300.25 MiB", "Load Text Model OK: True", "Starting Kobold API on port 49178 at http://localhost:49178/api/", "Please connect to custom endpoint at http://localhost:49178", "Threadpool set to 1 threads and 1 blasthreads..." ], "finished": "2026-11-13T12:42:13+0000" }