Instructions to use MrObiKenobi/mermaid-gemma-3-270m-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MrObiKenobi/mermaid-gemma-3-270m-it with Transformers:

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

pipe = pipeline("text-generation", model="MrObiKenobi/mermaid-gemma-3-270m-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MrObiKenobi/mermaid-gemma-3-270m-it")
model = AutoModelForCausalLM.from_pretrained("MrObiKenobi/mermaid-gemma-3-270m-it", 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 MrObiKenobi/mermaid-gemma-3-270m-it with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MrObiKenobi/mermaid-gemma-3-270m-it"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MrObiKenobi/mermaid-gemma-3-270m-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MrObiKenobi/mermaid-gemma-3-270m-it

SGLang

How to use MrObiKenobi/mermaid-gemma-3-270m-it 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 "MrObiKenobi/mermaid-gemma-3-270m-it" \
    --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": "MrObiKenobi/mermaid-gemma-3-270m-it",
		"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 "MrObiKenobi/mermaid-gemma-3-270m-it" \
        --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": "MrObiKenobi/mermaid-gemma-3-270m-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MrObiKenobi/mermaid-gemma-3-270m-it with Docker Model Runner:
```
docker model run hf.co/MrObiKenobi/mermaid-gemma-3-270m-it
```

MrObiKenobi commited on Feb 6

Commit

07fa1ff

verified ·

1 Parent(s): 1b3531b

Update README.md

Browse files

Files changed (1) hide show

README.md +53 -12

README.md CHANGED Viewed

@@ -1,36 +1,75 @@
 ---
 base_model: google/gemma-3-270m-it
 library_name: transformers
-model_name: checkpoint_models
 tags:
 - generated_from_trainer
 - sft
 - trl
 licence: license
 ---
-# Model Card for checkpoint_models
-This model is a fine-tuned version of [google/gemma-3-270m-it](https://huggingface.co/google/gemma-3-270m-it).
-It has been trained using [TRL](https://github.com/huggingface/trl).
 ## Quick start
 ```python
-from transformers import pipeline
-question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="None", device="cuda")
-output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
-print(output["generated_text"])
 ```
 ## Training procedure
-This model was trained with SFT.
 ### Framework versions
@@ -40,9 +79,11 @@ This model was trained with SFT.
 - Datasets: 4.0.0
 - Tokenizers: 0.22.2
-## Citations
 Cite TRL as:

 ---
 base_model: google/gemma-3-270m-it
 library_name: transformers
+model_name: mermaid-gemma-3-270m-it
 tags:
 - generated_from_trainer
 - sft
 - trl
 licence: license
+license: mit
+datasets:
+- Celiadraw/text-to-mermaid
+- ibm-research/MermaidSeqBench
+language:
+- en
 ---
+# Model Card for `mermaid-gemma-3-270m-it`
+This model is a fine-tuned variant of [google/gemma-3-270m-it](https://huggingface.co/google/gemma-3-270m-it), specifically trained to transform natural-language descriptions into structured Mermaid diagram code.
+## Example Input/Output
+**Input**:
+```
+Design a sequence diagram for a video conferencing application, illustrating
+interactions between users, scheduling system, video call establishment,
+audio transmission, and chat messaging.
+```
+**Output**:
+```mermaid
+sequenceDiagram
+  participant User1
+  participant User2
+  participant SchedulingSystem as Scheduling System
+  participant VideoCall as Video Call
+  User1 ->> SchedulingSystem: Schedule Meeting
+  SchedulingSystem ->> User2: Meeting Invitation
+  User1 ->> VideoCall: Start Call
+  VideoCall ->> User2: Receive Call
+```
 ## Quick start
 ```python
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
+model = AutoModelForCausalLM.from_pretrained(
+    "MrObiKenobi/mermaid-gemma-3-270m-it",
+    device_map="auto",
+    dtype="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained("MrObiKenobi/mermaid-gemma-3-270m-it")
+pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
+prompt = "Design a sequence diagram for a login system with user, frontend, and backend."
+messages = [{"role": "user", "content": prompt}]
+formatted = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+output = pipe(formatted, max_new_tokens=256)
+print(output[0]["generated_text"][len(formatted):])
 ```
 ## Training procedure
+This model was trained using supervised fine-tuning (SFT) on the dataset [Celiadraw/text-to-mermaid](https://huggingface.co/datasets/Celiadraw/text-to-mermaid), using only 1,000 samples.
+This work is intended as an academic exercise; however, we are confident that training on the full dataset would lead to significantly improved performance.
+Please refer to the accompanying notebook for detailed fine-tuning procedures and configuration.
 ### Framework versions
 - Datasets: 4.0.0
 - Tokenizers: 0.22.2
+## Credits
+Based on tutorial by Daniel Bourke: [Small LLM Fine-tuning Tutorial](https://www.youtube.com/watch?v=2hoNAr-id-E)
+## Citations
 Cite TRL as: