Instructions to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw")
model = AutoModelForCausalLM.from_pretrained("brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Trellis

How to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with Trellis:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw

SGLang

How to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw with Docker Model Runner:
```
docker model run hf.co/brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw
```

brandonmusic commited on Jul 22

Commit

2f42e31

verified ·

1 Parent(s): 8d3b1c6

Make the packaged server local-only by default

Browse files

Bind the published API to loopback and clarify that localhost refers to each downloader own machine.

Files changed (3) hide show

README.md +7 -6
docker-compose.yml +1 -1
server.sh +2 -1

README.md CHANGED Viewed

@@ -82,9 +82,10 @@ chmod +x server.sh
 ./server.sh logs
 ```
-From the inference host, the OpenAI-compatible endpoint is available at
-`http://localhost:8000/v1`. Remote clients should use
-`http://<server-address>:8000/v1`.
 ```bash
 curl http://localhost:8000/v1/chat/completions \
@@ -126,6 +127,7 @@ The defaults can be overridden without editing the files:
 | `MODEL_DIR` | Directory containing `server.sh` | Model mount |
 | `CACHE_DIR` | `~/.cache/glm52-exl3-sparkinfer` | Persistent JIT cache |
 | `PORT` | `8000` | Host API port |
 | `CUDA_VISIBLE_DEVICES` | `3,1,2,0` | Physical GPU to TP-rank order |
 | `GPU_MEMORY_UTILIZATION` | `0.93` | vLLM memory reservation |
 | `MAX_MODEL_LEN` | `524288` | Per-request context cap |
@@ -134,9 +136,8 @@ The tested GPU order intentionally keeps physical GPU 3 away from TP rank 3.
 On another host, set `CUDA_VISIBLE_DEVICES=0,1,2,3` or use the order appropriate
 for that machine.
-The Compose file publishes port 8000 on all host interfaces. Restrict the port
-binding or place an authenticated reverse proxy and firewall in front of it
-before exposing the API to an untrusted network.
 ## Runtime validation

 ./server.sh logs
 ```
+The OpenAI-compatible endpoint is available locally at
+`http://localhost:8000/v1`. Here, `localhost` always means the machine on
+which the downloader starts this model; it does not refer to the model
+publisher's machine.
 ```bash
 curl http://localhost:8000/v1/chat/completions \
 | `MODEL_DIR` | Directory containing `server.sh` | Model mount |
 | `CACHE_DIR` | `~/.cache/glm52-exl3-sparkinfer` | Persistent JIT cache |
 | `PORT` | `8000` | Host API port |
+| `BIND_ADDRESS` | `127.0.0.1` | Local-only host binding |
 | `CUDA_VISIBLE_DEVICES` | `3,1,2,0` | Physical GPU to TP-rank order |
 | `GPU_MEMORY_UTILIZATION` | `0.93` | vLLM memory reservation |
 | `MAX_MODEL_LEN` | `524288` | Per-request context cap |
 On another host, set `CUDA_VISIBLE_DEVICES=0,1,2,3` or use the order appropriate
 for that machine.
+The supplied Compose file binds only to loopback by default, so it does not
+publish the API to the LAN or internet.
 ## Runtime validation

docker-compose.yml CHANGED Viewed

@@ -3,7 +3,7 @@ services:
     image: ${IMAGE:-verdictai/glm52-exl3-sparkinfer:v1-gg-60c82d972-spi1937274-cu132-sm120a}
     container_name: glm52-exl3-sparkinfer
     ports:
-      - "0.0.0.0:${PORT:-8000}:8000"
     gpus: all
     shm_size: "32g"
     ipc: host

     image: ${IMAGE:-verdictai/glm52-exl3-sparkinfer:v1-gg-60c82d972-spi1937274-cu132-sm120a}
     container_name: glm52-exl3-sparkinfer
     ports:
+      - "${BIND_ADDRESS:-127.0.0.1}:${PORT:-8000}:8000"
     gpus: all
     shm_size: "32g"
     ipc: host

server.sh CHANGED Viewed

@@ -7,6 +7,7 @@ export IMAGE="${IMAGE:-verdictai/glm52-exl3-sparkinfer:v1-gg-60c82d972-spi193727
 export MODEL_DIR="${MODEL_DIR:-$SCRIPT_DIR}"
 export CACHE_DIR="${CACHE_DIR:-$HOME/.cache/glm52-exl3-sparkinfer}"
 export PORT="${PORT:-8000}"
 export CUDA_VISIBLE_DEVICES="${CUDA_VISIBLE_DEVICES:-3,1,2,0}"
 export GPU_MEMORY_UTILIZATION="${GPU_MEMORY_UTILIZATION:-0.93}"
 export MAX_MODEL_LEN="${MAX_MODEL_LEN:-524288}"
@@ -20,7 +21,7 @@ usage() {
 Usage: ./server.sh [start|stop|restart|logs|status|pull]
 Environment overrides:
-  IMAGE, MODEL_DIR, CACHE_DIR, PORT, CUDA_VISIBLE_DEVICES,
   GPU_MEMORY_UTILIZATION, MAX_MODEL_LEN, COMPOSE_PROJECT_NAME, COMPOSE_FILE
 EOF
 }

 export MODEL_DIR="${MODEL_DIR:-$SCRIPT_DIR}"
 export CACHE_DIR="${CACHE_DIR:-$HOME/.cache/glm52-exl3-sparkinfer}"
 export PORT="${PORT:-8000}"
+export BIND_ADDRESS="${BIND_ADDRESS:-127.0.0.1}"
 export CUDA_VISIBLE_DEVICES="${CUDA_VISIBLE_DEVICES:-3,1,2,0}"
 export GPU_MEMORY_UTILIZATION="${GPU_MEMORY_UTILIZATION:-0.93}"
 export MAX_MODEL_LEN="${MAX_MODEL_LEN:-524288}"
 Usage: ./server.sh [start|stop|restart|logs|status|pull]
 Environment overrides:
+  IMAGE, MODEL_DIR, CACHE_DIR, PORT, BIND_ADDRESS, CUDA_VISIBLE_DEVICES,
   GPU_MEMORY_UTILIZATION, MAX_MODEL_LEN, COMPOSE_PROJECT_NAME, COMPOSE_FILE
 EOF
 }