{ "case": "feat-multimodal", "machine": "macos-15-arm64-usecpu", "run": "ci-2", "args": [ "--model", "models/SmolVLM-256M-Instruct-Q8_0.gguf", "--mmproj", "models/mmproj-SmolVLM-256M-Instruct-Q8_0.gguf", "--whispermodel", "models/ggml-tiny.en.bin", "--embeddingsmodel", "models/all-MiniLM-L6-v2-Q8_0.gguf", "--contextsize", "4096", "--usecpu" ], "started": "2026-11-13T13:29:57+0000", "start": { "ready": true, "seconds": 4.06, "model": { "result": "koboldcpp/SmolVLM-256M-Instruct-Q8_0" } }, "mm": { "version": { "result": "KoboldCpp", "version": "1.124.2", "protected": false, "llm": true, "txt2img": false, "vision": true, "audio": false, "transcribe": true, "multiplayer": false, "websearch": false, "tts": false, "embeddings": true, "music": false, "musicllm": false, "musicfamily": "none", "savedata": false, "admin": 0, "router": false, "guidance": false, "jinja": true, "mcp": false, "parallelrequests": 1, "defaults_modified": false }, "vision": { "status": 200, "s": 6.15, "text": " \\red\\quad\\blue\\quad", "prompt_tokens": 159 }, "whisper": { "status": 200, "s": 14.61, "text": " And so my fellow Americans ask not what your country can do for you, ask what you can do for your country." }, "embeddings": { "status": 200, "s": 0.133, "dim": 384, "cos_cat_kitten": 0.657, "cos_cat_stock": 0.078 } }, "peak_rss_mb": 921.9, "log_lines": [ "Unable to determine GPU Memory", "Unable to determine available RAM", "Initializing dynamic library: koboldcpp_default.so", "Namespace(usecuda=None, usevulkan=None, usecpu=True, model=['models/SmolVLM-256M-Instruct-Q8_0.gguf'], model_param='models/SmolVLM-256M-Instruct-Q8_0.gguf', config=None, contextsize=4096, gpulayers=-1, host='', launch=False, tensor_split=None, threads=1, port=49821, port_param=49821, 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='models/mmproj-SmolVLM-256M-Instruct-Q8_0.gguf', 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='models/ggml-tiny.en.bin', ttsmodel='', ttswavtokenizer='', ttsgpu=False, ttsmaxlen=4096, ttsthreads=0, ttsdir='', musicllm='', musicembeddings='', musicdiffusion='', musicvae='', musiclowvram=False, embeddingsmodel='models/all-MiniLM-L6-v2-Q8_0.gguf', 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 = 165.24 MiB (8.51 BPW)", "print_info: n_ctx_train = 8192", "print_info: n_ctx_orig_yarn = 8192", "done_getting_tensors: tensor 'token_embd.weight' (q8_0) (and 211 others) moved from CPU_REPACK, using CPU instead", "load_tensors: CPU model buffer size = 28.76 MiB", "load_tensors: CPU_REPACK model buffer size = 136.34 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 (8192) -- the full capacity of the model will not be utilized", "llama_kv_cache: MTL0 KV buffer size = 95.62 MiB", "sched_reserve: MTL0 compute buffer size = 4.50 MiB", "sched_reserve: CPU compute buffer size = 98.50 MiB", "print_info: file size = 23.13 MiB (8.60 BPW)", "print_info: n_ctx_train = 512", "print_info: n_ctx_orig_yarn = 512", "done_getting_tensors: tensor 'token_embd.weight' (q8_0) (and 64 others) moved from CPU_REPACK, using CPU instead", "load_tensors: CPU model buffer size = 12.37 MiB", "load_tensors: CPU_REPACK model buffer size = 10.76 MiB", "llama_context: n_ctx = 512", "llama_context: n_ctx_seq = 512", "llama_context: flash_attn = enabled", "Load Text Model OK: True", "Chat template heuristics failed to identify chat completions format. Jinja will be used.", "Starting Kobold API on port 49821 at http://localhost:49821/api/", "Please connect to custom endpoint at http://localhost:49821", "Threadpool set to 1 threads and 1 blasthreads...", "reserve_compute_meta: MTL0 compute buffer size = 21.00 MiB", "reserve_compute_meta: CPU compute buffer size = 17.39 MiB", "warmup: WARNING: the CLIP graph uses unsupported operators by the backend", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]", "warmup: FLASH_ATTN_EXT: type = f32, ne = [64 12 1024 1]" ], "finished": "2026-11-13T13:30:23+0000" }