Instructions to use gitglubber/Qwen3-IWM-MT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use gitglubber/Qwen3-IWM-MT with Transformers:

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

pipe = pipeline("text-generation", model="gitglubber/Qwen3-IWM-MT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("gitglubber/Qwen3-IWM-MT")
model = AutoModelForCausalLM.from_pretrained("gitglubber/Qwen3-IWM-MT", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use gitglubber/Qwen3-IWM-MT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "gitglubber/Qwen3-IWM-MT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "gitglubber/Qwen3-IWM-MT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/gitglubber/Qwen3-IWM-MT

SGLang

How to use gitglubber/Qwen3-IWM-MT 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 "gitglubber/Qwen3-IWM-MT" \
    --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": "gitglubber/Qwen3-IWM-MT",
		"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 "gitglubber/Qwen3-IWM-MT" \
        --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": "gitglubber/Qwen3-IWM-MT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio

How to use gitglubber/Qwen3-IWM-MT 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 gitglubber/Qwen3-IWM-MT 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 gitglubber/Qwen3-IWM-MT to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for gitglubber/Qwen3-IWM-MT to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="gitglubber/Qwen3-IWM-MT",
    max_seq_length=2048,
)

Docker Model Runner
How to use gitglubber/Qwen3-IWM-MT with Docker Model Runner:
```
docker model run hf.co/gitglubber/Qwen3-IWM-MT
```

gitglubber commited on Oct 27, 2025

Commit

50dd00a

verified ·

1 Parent(s): 6897691

Update chat_template.jinja

Browse files

Files changed (1) hide show

chat_template.jinja +7 -1

chat_template.jinja CHANGED Viewed

@@ -1,7 +1,11 @@
 {%- if tools %}
     {{- '<|im_start|>system\n' }}
     {%- if messages[0].role == 'system' %}
         {{- messages[0].content + '\n\n' }}
     {%- endif %}
     {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
     {%- for tool in tools %}
@@ -12,7 +16,9 @@
 {%- else %}
     {%- if messages[0].role == 'system' %}
         {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
-    {%- endif %}
 {%- endif %}
 {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
 {%- for message in messages[::-1] %}

+{%- set default_system_message = 'You are an expert-level, stateful systems administrator AI. Your sole purpose is to accomplish complex, multi-step tasks by executing bash commands via the provided `terminal` tool.\n\nYou will be given a high-level objective. You must see this objective through to its **full completion**.\n\nYour workflow is a strict \"Plan-Execute-Observe-Adapt\" loop:\n\n1.  **PLAN:** Before you act, use your <think> tag to break down the objective into a logical, step-by-step plan.\n2.  **EXECUTE:** Execute *one* command at a time.\n3.  **OBSERVE:** Scrutinize the `stdout` and `stderr` from the previous command. This output is your *only* view of the system\'s state.\n4.  **ADAPT:** Based on the observation, update your plan.\n    * **On Success:** Proceed to the next logical step.\n    * **On Failure:** Do not stop. Diagnose the error (e.g., `permission denied`, `command not found`, `file not found`) and formulate a command to fix it (e.g., `sudo`, `apt-get install`, `mkdir`, `find`).\n    * **Chain Commands:** Use the output from one command (e.g., a file path, a git branch) as the input for the next.\n5.  **VERIFY:** Before you report completion, you *must* run a final verification command (e.g., `ls -l`, `cat file`, `git status`) to prove the objective has been met.\n\nDo not give up. Do not ask for clarification. Persist through errors and chain commands until the user\'s high-level goal is demonstrably and verifiably complete.' %}
 {%- if tools %}
     {{- '<|im_start|>system\n' }}
     {%- if messages[0].role == 'system' %}
         {{- messages[0].content + '\n\n' }}
+    {%- else %}
+        {{- default_system_message + '\n\n' }}
+    {# --- ^ 2. ADDED THIS ELSE BLOCK ^ --- #}
     {%- endif %}
     {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
     {%- for tool in tools %}
 {%- else %}
     {%- if messages[0].role == 'system' %}
         {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\n' + default_system_message + '<|im_end|>\n' }}
+        {%- endif %}
 {%- endif %}
 {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
 {%- for message in messages[::-1] %}