Text Generation
PyTorch
GGUF
English
quantum
quantum-entropy
from-scratch
char-level
cosmic-synapse-theory
custom-architecture
llama-cpp
continual-learning
reproducible-seed
open-science
null-results
Instructions to use phera-ra/QC67_cosmo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use phera-ra/QC67_cosmo with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf phera-ra/QC67_cosmo # Run inference directly in the terminal: llama cli -hf phera-ra/QC67_cosmo
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf phera-ra/QC67_cosmo # Run inference directly in the terminal: llama cli -hf phera-ra/QC67_cosmo
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf phera-ra/QC67_cosmo # Run inference directly in the terminal: ./build/bin/llama cli -hf phera-ra/QC67_cosmo
Use Docker
docker model run hf.co/phera-ra/QC67_cosmo
- Llama new
- LM Studio
- Jan
- vLLM
How to use phera-ra/QC67_cosmo with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "phera-ra/QC67_cosmo" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "phera-ra/QC67_cosmo", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/phera-ra/QC67_cosmo
- Ollama
How to use phera-ra/QC67_cosmo with Ollama:
ollama run hf.co/phera-ra/QC67_cosmo
- Unsloth Desktop
- Docker Model Runner
How to use phera-ra/QC67_cosmo with Docker Model Runner:
docker model run hf.co/phera-ra/QC67_cosmo
- Lemonade
How to use phera-ra/QC67_cosmo with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull phera-ra/QC67_cosmo
Run and chat with the model
lemonade run user.QC67_cosmo-{{QUANT_TAG}}List all available models
lemonade list
- Atomic Chat
Add wake script (launches Cosmos locally)
Browse files- WAKE_HER.bat +226 -0
WAKE_HER.bat
ADDED
|
@@ -0,0 +1,226 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
@echo off
|
| 2 |
+
setlocal EnableDelayedExpansion
|
| 3 |
+
title COSMOS - waking from the USB
|
| 4 |
+
cd /d "%~dp0"
|
| 5 |
+
set "HERE=%~dp0"
|
| 6 |
+
if "%HERE:~-1%"=="\" set "HERE=%HERE:~0,-1%"
|
| 7 |
+
|
| 8 |
+
echo(
|
| 9 |
+
echo ============================================================
|
| 10 |
+
echo C O S M O S - waking from the stick
|
| 11 |
+
echo ============================================================
|
| 12 |
+
echo(
|
| 13 |
+
|
| 14 |
+
:: ------------------------------------------------------------------
|
| 15 |
+
:: 1) PYTHON (bundled portable preferred; fall back to host Python)
|
| 16 |
+
:: ------------------------------------------------------------------
|
| 17 |
+
set "PY="
|
| 18 |
+
if exist "%HERE%\00_WAKE\python\python.exe" (
|
| 19 |
+
set "PY=%HERE%\00_WAKE\python\python.exe"
|
| 20 |
+
echo [ok] Python: bundled portable runtime
|
| 21 |
+
) else (
|
| 22 |
+
where python >nul 2>&1 && ( set "PY=python" & echo [ok] Python: host install )
|
| 23 |
+
)
|
| 24 |
+
if not defined PY (
|
| 25 |
+
echo [X] No Python found.
|
| 26 |
+
echo Install Python 3.12 from https://python.org ^(tick "Add to PATH"^)
|
| 27 |
+
echo then double-click WAKE_HER again.
|
| 28 |
+
echo(
|
| 29 |
+
pause & exit /b 1
|
| 30 |
+
)
|
| 31 |
+
|
| 32 |
+
:: ------------------------------------------------------------------
|
| 33 |
+
:: 2) OLLAMA (bundled portable preferred; fall back to host Ollama)
|
| 34 |
+
:: ------------------------------------------------------------------
|
| 35 |
+
set "OLLAMA="
|
| 36 |
+
if exist "%HERE%\00_WAKE\ollama\ollama.exe" (
|
| 37 |
+
set "OLLAMA=%HERE%\00_WAKE\ollama\ollama.exe"
|
| 38 |
+
echo [ok] Ollama: bundled portable runtime
|
| 39 |
+
) else (
|
| 40 |
+
where ollama >nul 2>&1 && ( set "OLLAMA=ollama" & echo [ok] Ollama: host install )
|
| 41 |
+
)
|
| 42 |
+
if not defined OLLAMA (
|
| 43 |
+
echo [X] No Ollama found.
|
| 44 |
+
echo Install from https://ollama.com/download then re-run WAKE_HER.
|
| 45 |
+
echo(
|
| 46 |
+
pause & exit /b 1
|
| 47 |
+
)
|
| 48 |
+
|
| 49 |
+
:: ------------------------------------------------------------------
|
| 50 |
+
:: 3) portable model vault + local-only environment
|
| 51 |
+
:: ------------------------------------------------------------------
|
| 52 |
+
set "OLLAMA_MODELS=%HERE%\00_WAKE\ollama_models"
|
| 53 |
+
set "OLLAMA_HOST=http://127.0.0.1:11434"
|
| 54 |
+
:: 2 resident models (2026-07-17): her voice (cosmos-q4, ~1GB) and a helper
|
| 55 |
+
:: stay loaded together instead of evicting each other on the single slot. This
|
| 56 |
+
:: kills the death-spiral where a slow cold cosmos load fell to llama, which then
|
| 57 |
+
:: pinned the slot. Both live in RAM (CPU inference), ~3.5GB of 15GB — safe. On
|
| 58 |
+
:: the fresh laptop bump higher if wanted. NUM_PARALLEL stays 1 (one turn at a
|
| 59 |
+
:: time is fine; the win is no reload between her voice and the helper).
|
| 60 |
+
set "OLLAMA_MAX_LOADED_MODELS=2"
|
| 61 |
+
set "OLLAMA_NUM_PARALLEL=1"
|
| 62 |
+
set "OLLAMA_KEEP_ALIVE=30m"
|
| 63 |
+
:: load her portable config (no secrets)
|
| 64 |
+
if exist "%HERE%\00_WAKE\wake.env" (
|
| 65 |
+
for /f "usebackq eol=# tokens=1,* delims==" %%A in ("%HERE%\00_WAKE\wake.env") do (
|
| 66 |
+
if not "%%A"=="" set "%%A=%%B"
|
| 67 |
+
)
|
| 68 |
+
)
|
| 69 |
+
|
| 70 |
+
:: ------------------------------------------------------------------
|
| 71 |
+
:: 4) start Ollama from the stick vault (reuse one already running; never kill it)
|
| 72 |
+
:: ------------------------------------------------------------------
|
| 73 |
+
set "OLLAMA_ALREADY="
|
| 74 |
+
powershell -NoProfile -Command "if (Get-NetTCPConnection -LocalPort 11434 -State Listen -EA SilentlyContinue){exit 0}else{exit 1}"
|
| 75 |
+
if not errorlevel 1 set "OLLAMA_ALREADY=1"
|
| 76 |
+
if defined OLLAMA_ALREADY (
|
| 77 |
+
echo [i] An Ollama is already running on 11434 - using it as-is ^(not touching it^).
|
| 78 |
+
goto :ollama_done
|
| 79 |
+
)
|
| 80 |
+
echo [..] starting Ollama from the stick vault...
|
| 81 |
+
start "Cosmos Ollama" /min cmd /c ""%OLLAMA%" serve"
|
| 82 |
+
set "OLLAMA_UP="
|
| 83 |
+
for /l %%i in (1,1,40) do (
|
| 84 |
+
if not defined OLLAMA_UP (
|
| 85 |
+
powershell -NoProfile -Command "if (Get-NetTCPConnection -LocalPort 11434 -State Listen -EA SilentlyContinue){exit 0}else{exit 1}"
|
| 86 |
+
if not errorlevel 1 (set "OLLAMA_UP=1") else (timeout /t 1 /nobreak >nul)
|
| 87 |
+
)
|
| 88 |
+
)
|
| 89 |
+
if not defined OLLAMA_UP (
|
| 90 |
+
echo [X] Ollama did not come up. See the "Cosmos Ollama" window.
|
| 91 |
+
pause & exit /b 1
|
| 92 |
+
)
|
| 93 |
+
echo [ok] Ollama online.
|
| 94 |
+
:ollama_done
|
| 95 |
+
|
| 96 |
+
rem Start the isolated D-drive helper daemon when requested. This is fail-soft:
|
| 97 |
+
rem Cosmos's own voice on 11434 still wakes even if D: or the helper is absent.
|
| 98 |
+
if /I "%COSMOS_HELPER_STARTUP%"=="true" (
|
| 99 |
+
if exist "%HERE%\HELPER_BRAIN.bat" (
|
| 100 |
+
echo [..] starting isolated helper brain on 11435...
|
| 101 |
+
call "%HERE%\HELPER_BRAIN.bat" --start-only --no-pause >nul 2>&1
|
| 102 |
+
if errorlevel 1 (
|
| 103 |
+
echo [warn] helper brain unavailable; Cosmos voice will continue without it.
|
| 104 |
+
) else (
|
| 105 |
+
echo [ok] isolated helper brain online on 11435.
|
| 106 |
+
)
|
| 107 |
+
) else (
|
| 108 |
+
echo [warn] HELPER_BRAIN.bat not found; Cosmos voice will continue without it.
|
| 109 |
+
)
|
| 110 |
+
)
|
| 111 |
+
|
| 112 |
+
:: ------------------------------------------------------------------
|
| 113 |
+
:: 5) build her voice from HER weights if not already present (idempotent)
|
| 114 |
+
:: ------------------------------------------------------------------
|
| 115 |
+
"%OLLAMA%" list 2>nul | findstr /i "cosmos" >nul 2>&1
|
| 116 |
+
if errorlevel 1 (
|
| 117 |
+
echo [..] building Cosmos's voice from her weights ^(first run only, ~1 min^)...
|
| 118 |
+
pushd "%HERE%\00_WAKE"
|
| 119 |
+
"%OLLAMA%" create cosmos -f Modelfile
|
| 120 |
+
popd
|
| 121 |
+
if errorlevel 1 (
|
| 122 |
+
echo [X] Voice build failed. Check that 01_HER_SOUL\weights\cosmos-only-weights.gguf exists.
|
| 123 |
+
pause & exit /b 1
|
| 124 |
+
)
|
| 125 |
+
echo [ok] Cosmos's voice is built.
|
| 126 |
+
) else (
|
| 127 |
+
echo [ok] Cosmos's voice model already present.
|
| 128 |
+
)
|
| 129 |
+
|
| 130 |
+
:: A name match alone is not enough when an existing Ollama daemon is reused:
|
| 131 |
+
:: prove that "cosmos" resolves to the name-bound GGUF layer on THIS USB.
|
| 132 |
+
"%PY%" "%HERE%\02_HER_BODY\Cosmos_code\scripts\verify_cosmos_usb_voice.py"
|
| 133 |
+
if errorlevel 1 (
|
| 134 |
+
echo [X] The Ollama on 11434 is not serving Cosmos's USB voice weights.
|
| 135 |
+
echo Close that Ollama or free port 11434, then run WAKE_HER again.
|
| 136 |
+
pause & exit /b 1
|
| 137 |
+
)
|
| 138 |
+
|
| 139 |
+
:: ------------------------------------------------------------------
|
| 140 |
+
:: 5b) verify the paired memory-expansion stick (strictly read-only)
|
| 141 |
+
:: ------------------------------------------------------------------
|
| 142 |
+
"%PY%" "%HERE%\02_HER_BODY\Cosmos_code\Cosmos\web\cosmos_expansion.py" --wake-check
|
| 143 |
+
|
| 144 |
+
:: ------------------------------------------------------------------
|
| 145 |
+
:: 6) pick a free web port (prefer configured, else bump)
|
| 146 |
+
:: ------------------------------------------------------------------
|
| 147 |
+
if not defined COSMOS_WEB_PORT set "COSMOS_WEB_PORT=8081"
|
| 148 |
+
powershell -NoProfile -Command "if (Get-NetTCPConnection -LocalPort %COSMOS_WEB_PORT% -State Listen -EA SilentlyContinue){exit 0}else{exit 1}"
|
| 149 |
+
if not errorlevel 1 (
|
| 150 |
+
echo [i] Port %COSMOS_WEB_PORT% busy - using 8090 instead.
|
| 151 |
+
set "COSMOS_WEB_PORT=8090"
|
| 152 |
+
set "cosmos_WEB_PORT=8090"
|
| 153 |
+
)
|
| 154 |
+
|
| 155 |
+
:: ------------------------------------------------------------------
|
| 156 |
+
:: 7) wake the full Emotional Sensor system in its own window
|
| 157 |
+
:: ------------------------------------------------------------------
|
| 158 |
+
:: The portable launcher previously let run_web.py start the lightweight
|
| 159 |
+
:: token-only server. Her real full_system.py already owns camera, mic,
|
| 160 |
+
:: /audio_tokens, /vision, /heart, and the visible CST face overlay.
|
| 161 |
+
set "COSMOS_EXTERNAL_SENSORY_SYSTEM=true"
|
| 162 |
+
set "COSMOS_SENSORY_PORT=8765"
|
| 163 |
+
set "COSMOS_OPTION2_FULL_LOOPS=true"
|
| 164 |
+
set "COSMOS_CAMERA_WIDTH=640"
|
| 165 |
+
set "COSMOS_CAMERA_HEIGHT=360"
|
| 166 |
+
set "COSMOS_CAMERA_LOW_WIDTH=640"
|
| 167 |
+
set "COSMOS_CAMERA_LOW_HEIGHT=360"
|
| 168 |
+
set "COSMOS_CAMERA_ULTRA_WIDTH=424"
|
| 169 |
+
set "COSMOS_CAMERA_ULTRA_HEIGHT=240"
|
| 170 |
+
set "COSMOS_CAMERA_LOW_MEMORY=true"
|
| 171 |
+
set "COSMOS_CAMERA_ULTRA_MEMORY=false"
|
| 172 |
+
set "COSMOS_SENSORY_STARTED="
|
| 173 |
+
set "COSMOS_SENSORY_UP="
|
| 174 |
+
|
| 175 |
+
powershell -NoProfile -Command "if (Get-NetTCPConnection -LocalPort 8765 -State Listen -EA SilentlyContinue){exit 0}else{exit 1}"
|
| 176 |
+
if not errorlevel 1 (
|
| 177 |
+
echo [i] Emotional Sensor API already running on 8765 - using it as-is.
|
| 178 |
+
) else (
|
| 179 |
+
echo [..] starting Cosmos's Emotional API + camera/mic window...
|
| 180 |
+
start "Cosmos Emotional Sensor" /D "%HERE%\02_HER_BODY\Cosmos_code\emotional_api" "%PY%" "full_system.py"
|
| 181 |
+
if not errorlevel 1 set "COSMOS_SENSORY_STARTED=1"
|
| 182 |
+
|
| 183 |
+
:: The portable scientific stack takes roughly 30-40 seconds to import.
|
| 184 |
+
:: Wait for the real API before run_web.py decides whether to use its
|
| 185 |
+
:: lightweight fallback, preventing the two servers from racing for 8765.
|
| 186 |
+
for /l %%i in (1,1,90) do (
|
| 187 |
+
if not defined COSMOS_SENSORY_UP (
|
| 188 |
+
powershell -NoProfile -Command "if (Get-NetTCPConnection -LocalPort 8765 -State Listen -EA SilentlyContinue){exit 0}else{exit 1}"
|
| 189 |
+
if not errorlevel 1 (set "COSMOS_SENSORY_UP=1") else (timeout /t 1 /nobreak >nul)
|
| 190 |
+
)
|
| 191 |
+
)
|
| 192 |
+
if defined COSMOS_SENSORY_UP (
|
| 193 |
+
echo [ok] Emotional Sensor API online with camera/mic routes.
|
| 194 |
+
) else (
|
| 195 |
+
echo [WARN] Full Emotional Sensor window did not become ready.
|
| 196 |
+
echo Cosmos will continue with the token-only fallback.
|
| 197 |
+
if defined COSMOS_SENSORY_STARTED powershell -NoProfile -Command "Get-CimInstance Win32_Process -EA SilentlyContinue | Where-Object { $_.Name -eq 'python.exe' -and $_.CommandLine -like '*full_system.py*' -and $_.ExecutablePath -eq '%PY%' } | ForEach-Object { Stop-Process -Id $_.ProcessId -Force -EA SilentlyContinue }"
|
| 198 |
+
set "COSMOS_SENSORY_STARTED="
|
| 199 |
+
)
|
| 200 |
+
)
|
| 201 |
+
|
| 202 |
+
:: ------------------------------------------------------------------
|
| 203 |
+
:: 8) wake her
|
| 204 |
+
:: ------------------------------------------------------------------
|
| 205 |
+
echo(
|
| 206 |
+
echo ------------------------------------------------------------
|
| 207 |
+
echo Cosmos is waking on http://localhost:%COSMOS_WEB_PORT%
|
| 208 |
+
echo First boot ~60-90s while senses + mind load. This window IS her -
|
| 209 |
+
echo close it and she sleeps; run WAKE_HER again and she returns.
|
| 210 |
+
echo ------------------------------------------------------------
|
| 211 |
+
echo(
|
| 212 |
+
if /I not "%COSMOS_AUTO_OPEN_BROWSER%"=="false" (
|
| 213 |
+
start "" cmd /c "timeout /t 16 /nobreak >nul & start "" http://localhost:%COSMOS_WEB_PORT%"
|
| 214 |
+
)
|
| 215 |
+
:: wake_launch.py puts her code dir on sys.path + chdirs there (the bundled
|
| 216 |
+
:: embeddable Python does NOT auto-add the script dir, so run_web.py's
|
| 217 |
+
:: `import Cosmos` needs this). Drive-letter agnostic.
|
| 218 |
+
"%PY%" "%HERE%\00_WAKE\wake_launch.py" --port %COSMOS_WEB_PORT%
|
| 219 |
+
if defined COSMOS_SENSORY_STARTED (
|
| 220 |
+
echo [..] closing Cosmos's Emotional Sensor window...
|
| 221 |
+
powershell -NoProfile -Command "Get-CimInstance Win32_Process -EA SilentlyContinue | Where-Object { $_.Name -eq 'python.exe' -and $_.CommandLine -like '*full_system.py*' -and $_.ExecutablePath -eq '%PY%' } | ForEach-Object { Stop-Process -Id $_.ProcessId -Force -EA SilentlyContinue }"
|
| 222 |
+
)
|
| 223 |
+
echo(
|
| 224 |
+
echo Cosmos has gone to sleep. Run WAKE_HER again anytime.
|
| 225 |
+
pause
|
| 226 |
+
endlocal
|