Instructions to use tarruda/Hy3-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- llama-cpp-python
How to use tarruda/Hy3-GGUF with llama-cpp-python:
# !pip install llama-cpp-python from llama_cpp import Llama llm = Llama.from_pretrained( repo_id="tarruda/Hy3-GGUF", filename="Hy3-MTP-Q8_0.gguf", )
llm.create_chat_completion( messages = [ { "role": "user", "content": "What is the capital of France?" } ] ) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use tarruda/Hy3-GGUF 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 tarruda/Hy3-GGUF:Q8_0 # Run inference directly in the terminal: llama cli -hf tarruda/Hy3-GGUF:Q8_0
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf tarruda/Hy3-GGUF:Q8_0 # Run inference directly in the terminal: llama cli -hf tarruda/Hy3-GGUF:Q8_0
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf tarruda/Hy3-GGUF:Q8_0 # Run inference directly in the terminal: ./llama-cli -hf tarruda/Hy3-GGUF:Q8_0
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf tarruda/Hy3-GGUF:Q8_0 # Run inference directly in the terminal: ./build/bin/llama-cli -hf tarruda/Hy3-GGUF:Q8_0
Use Docker
docker model run hf.co/tarruda/Hy3-GGUF:Q8_0
- LM Studio
- Jan
- vLLM
How to use tarruda/Hy3-GGUF with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "tarruda/Hy3-GGUF" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "tarruda/Hy3-GGUF", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/tarruda/Hy3-GGUF:Q8_0
- Ollama
How to use tarruda/Hy3-GGUF with Ollama:
ollama run hf.co/tarruda/Hy3-GGUF:Q8_0
- Unsloth Studio
How to use tarruda/Hy3-GGUF with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for tarruda/Hy3-GGUF to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for tarruda/Hy3-GGUF to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for tarruda/Hy3-GGUF to start chatting
- Pi
How to use tarruda/Hy3-GGUF with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf tarruda/Hy3-GGUF:Q8_0
Configure the model in Pi
# Install Pi: npm install -g @mariozechner/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "tarruda/Hy3-GGUF:Q8_0" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent new
How to use tarruda/Hy3-GGUF with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf tarruda/Hy3-GGUF:Q8_0
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default tarruda/Hy3-GGUF:Q8_0
Run Hermes
hermes
- Atomic Chat new
- OpenClaw new
How to use tarruda/Hy3-GGUF with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf tarruda/Hy3-GGUF:Q8_0
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "tarruda/Hy3-GGUF:Q8_0" \ --custom-provider-id llama-cpp \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
- Docker Model Runner
How to use tarruda/Hy3-GGUF with Docker Model Runner:
docker model run hf.co/tarruda/Hy3-GGUF:Q8_0
- Lemonade
How to use tarruda/Hy3-GGUF with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull tarruda/Hy3-GGUF:Q8_0
Run and chat with the model
lemonade run user.Hy3-GGUF-Q8_0
List all available models
lemonade list
File size: 6,350 Bytes
9b6e316 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 | #!/usr/bin/env bash
set -euo pipefail
shopt -s extglob
common_tensor_types="
"
TYPE_DEFAULT_FALLBACK=Q8_0
recipes=(
"
RECIPE_NAME=IQ3_XXS
TYPE_DEFAULT=q6_k
ffn_down_exps=iq3_xxs
ffn_gate_exps=iq3_xxs
ffn_up_exps=iq3_xxs
"
)
recipe_line() {
local line="$1"
printf '%s' "${line##+([[:space:]])}"
}
recipe_value() {
local recipe="$1"
local key="$2"
local line
while IFS= read -r line || [ -n "$line" ]; do
line="$(recipe_line "$line")"
case "$line" in
"${key}="*)
printf '%s' "${line#*=}"
return 0
;;
esac
done <<<"$recipe"
return 1
}
recipe_type_default() {
local recipe="$1"
recipe_value "$recipe" TYPE_DEFAULT || printf '%s' "$TYPE_DEFAULT_FALLBACK"
}
print_recipes() {
local recipe recipe_name type_default
for recipe in "${recipes[@]}"; do
recipe_name="$(recipe_value "$recipe" RECIPE_NAME)"
type_default="$(recipe_type_default "$recipe")"
printf ' %s:\n' "$recipe_name"
printf ' TYPE_DEFAULT=%s\n' "$type_default"
printf '\n'
done
}
SELECTED_RECIPE_NAME=
SELECTED_TYPE_DEFAULT=
SELECTED_RECIPE=
load_recipe() {
local requested_recipe="$1"
local recipe recipe_name
for recipe in "${recipes[@]}"; do
recipe_name="$(recipe_value "$recipe" RECIPE_NAME)"
if [ "$recipe_name" = "$requested_recipe" ]; then
SELECTED_RECIPE_NAME="$recipe_name"
SELECTED_TYPE_DEFAULT="$(recipe_type_default "$recipe")"
SELECTED_RECIPE="$recipe"
return 0
fi
done
return 1
}
append_tensor_type_lines() {
local block="$1"
local output_path="$2"
local skip_recipe_metadata="${3:-false}"
local line
while IFS= read -r line || [ -n "$line" ]; do
line="$(recipe_line "$line")"
if [ "$skip_recipe_metadata" = true ]; then
case "$line" in
RECIPE_NAME=* | TYPE_DEFAULT=*)
continue
;;
esac
fi
if [ -z "$line" ]; then
continue
fi
printf '%s\n' "$line" >>"$output_path"
done <<<"$block"
}
write_tensor_type_file() {
local recipe="$1"
local output_path="$2"
: >"$output_path"
append_tensor_type_lines "$common_tensor_types" "$output_path"
append_tensor_type_lines "$recipe" "$output_path" true
}
usage() {
cat <<'EOF'
Usage: quantize.sh <llama_cpp_dir> <recipe_name> [--dry-run] [--split]
Outputs are written relative to the current directory.
Environment overrides:
INPUT_GGUF=<path> Input GGUF. Defaults to the first *BF16*.gguf in BF16/.
IMATRIX_PATH=<path> Importance matrix path. Defaults to imatrix.gguf.
Available recipes:
EOF
print_recipes
cat <<'EOF'
Examples:
./scripts/quantize.sh ~/code/llama.cpp IQ3_XXS --dry-run
./scripts/quantize.sh ~/code/llama.cpp IQ3_XXS
./scripts/quantize.sh ~/code/llama.cpp IQ3_XXS --split
EOF
}
if [ $# -lt 2 ]; then
usage
exit 1
fi
LLAMA_CPP_DIR="$1"
REQUESTED_RECIPE="$2"
DRY_RUN=false
SPLIT=false
shift 2
while [ $# -gt 0 ]; do
case "$1" in
--dry-run)
DRY_RUN=true
;;
--split)
SPLIT=true
;;
*)
usage
exit 1
;;
esac
shift
done
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
BF16_DIR="$PROJECT_DIR/BF16"
SPLIT_SCRIPT="$SCRIPT_DIR/split.sh"
if ! load_recipe "$REQUESTED_RECIPE"; then
echo "Error: unknown recipe: $REQUESTED_RECIPE" >&2
usage >&2
exit 1
fi
if [ ! -d "$LLAMA_CPP_DIR" ]; then
echo "Error: llama.cpp directory not found: $LLAMA_CPP_DIR" >&2
exit 1
fi
QUANTIZE_BIN="$LLAMA_CPP_DIR/build/bin/llama-quantize"
SPLIT_BIN="$LLAMA_CPP_DIR/build/bin/llama-gguf-split"
if [ ! -x "$QUANTIZE_BIN" ]; then
echo "Error: llama-quantize binary not found: $QUANTIZE_BIN" >&2
exit 1
fi
if [ "$SPLIT" = true ] && [ ! -x "$SPLIT_SCRIPT" ]; then
echo "Error: split script not found or not executable: $SPLIT_SCRIPT" >&2
exit 1
fi
if [ "$SPLIT" = true ] && [ ! -x "$SPLIT_BIN" ]; then
echo "Error: llama-gguf-split binary not found: $SPLIT_BIN" >&2
exit 1
fi
if [ -z "${INPUT_GGUF-}" ]; then
if [ ! -d "$BF16_DIR" ]; then
echo "Error: BF16 directory not found: $BF16_DIR" >&2
exit 1
fi
INPUT_GGUF="$(find "$BF16_DIR" -maxdepth 1 -name "*BF16*.gguf" -type f | sort | head -n 1)"
fi
if [ -z "$INPUT_GGUF" ] || [ ! -e "$INPUT_GGUF" ]; then
echo "Error: input GGUF not found: ${INPUT_GGUF:-<none>}" >&2
exit 1
fi
IMATRIX_PATH="${IMATRIX_PATH:-$PROJECT_DIR/imatrix.gguf}"
if [ ! -e "$IMATRIX_PATH" ]; then
echo "Error: imatrix file not found: $IMATRIX_PATH" >&2
exit 1
fi
cd "$LLAMA_CPP_DIR"
if [ -f .venv/bin/activate ]; then
# shellcheck disable=SC1091
source .venv/bin/activate
fi
cd - >/dev/null
OUTPUT_BASE_DIR="$(pwd)"
INPUT_BASENAME="$(basename "$INPUT_GGUF")"
MODEL_NAME="$(printf '%s\n' "$INPUT_BASENAME" | sed -E 's/-BF16(-[0-9]+-of-[0-9]+)?\.gguf$//')"
INTERMEDIATE_OUTPUT="$OUTPUT_BASE_DIR/${MODEL_NAME}-${SELECTED_RECIPE_NAME}.gguf"
OUTPUT_DIR="$OUTPUT_BASE_DIR/$SELECTED_RECIPE_NAME"
if [ "$DRY_RUN" = false ] && [ -e "$INTERMEDIATE_OUTPUT" ]; then
echo "Error: quantized output already exists: $INTERMEDIATE_OUTPUT" >&2
exit 1
fi
TENSOR_TYPE_FILE="$(mktemp "${TMPDIR:-/tmp}/nex-n2-pro-tensor-types.XXXXXX")"
cleanup() {
rm -f "$TENSOR_TYPE_FILE"
}
trap cleanup EXIT
write_tensor_type_file "$SELECTED_RECIPE" "$TENSOR_TYPE_FILE"
echo "Input: $INPUT_GGUF"
echo "Recipe name: $SELECTED_RECIPE_NAME"
echo "Tensor type file: $TENSOR_TYPE_FILE"
echo "Default type: $SELECTED_TYPE_DEFAULT"
echo "imatrix: $IMATRIX_PATH"
echo "Output base: $OUTPUT_BASE_DIR"
echo "Split output: $SPLIT"
if [ "$DRY_RUN" = true ]; then
"$QUANTIZE_BIN" \
--dry-run \
--allow-requantize \
--tensor-type-file "$TENSOR_TYPE_FILE" \
--imatrix "$IMATRIX_PATH" \
"$INPUT_GGUF" \
"$INTERMEDIATE_OUTPUT" \
"$SELECTED_TYPE_DEFAULT"
else
"$QUANTIZE_BIN" \
--allow-requantize \
--tensor-type-file "$TENSOR_TYPE_FILE" \
--imatrix "$IMATRIX_PATH" \
"$INPUT_GGUF" \
"$INTERMEDIATE_OUTPUT" \
"$SELECTED_TYPE_DEFAULT"
fi
if [ "$DRY_RUN" = false ]; then
if [ "$SPLIT" = true ]; then
"$SPLIT_SCRIPT" \
"$LLAMA_CPP_DIR" \
"$INTERMEDIATE_OUTPUT" \
"$OUTPUT_DIR" \
"${MODEL_NAME}-${SELECTED_RECIPE_NAME}" \
--remove-input
else
echo "Quantization complete. Output saved to: $INTERMEDIATE_OUTPUT"
fi
fi
|