Instructions to use kth8/gemma-3-1b-it-IFeval with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kth8/gemma-3-1b-it-IFeval with Transformers:

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

pipe = pipeline("text-generation", model="kth8/gemma-3-1b-it-IFeval")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kth8/gemma-3-1b-it-IFeval")
model = AutoModelForCausalLM.from_pretrained("kth8/gemma-3-1b-it-IFeval")
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]:]))

Inference
Local Apps Settings

vLLM

How to use kth8/gemma-3-1b-it-IFeval with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kth8/gemma-3-1b-it-IFeval"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kth8/gemma-3-1b-it-IFeval",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/kth8/gemma-3-1b-it-IFeval

SGLang

How to use kth8/gemma-3-1b-it-IFeval 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 "kth8/gemma-3-1b-it-IFeval" \
    --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": "kth8/gemma-3-1b-it-IFeval",
		"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 "kth8/gemma-3-1b-it-IFeval" \
        --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": "kth8/gemma-3-1b-it-IFeval",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio

How to use kth8/gemma-3-1b-it-IFeval 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 kth8/gemma-3-1b-it-IFeval 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 kth8/gemma-3-1b-it-IFeval to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for kth8/gemma-3-1b-it-IFeval to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="kth8/gemma-3-1b-it-IFeval",
    max_seq_length=2048,
)

Docker Model Runner
How to use kth8/gemma-3-1b-it-IFeval with Docker Model Runner:
```
docker model run hf.co/kth8/gemma-3-1b-it-IFeval
```

A fine-tune of unsloth/gemma-3-1b-it on the HuggingFaceH4/ifeval-like-data dataset.

Usage example

System prompt

You are a helpful assistant.

User prompt

We want to teach our daughter the value of a clean home. Can you suggest chores that are appropriate for a child who is who 5 years old?

Assistant response

No Problem! Here are some great chores for a 5-year-old to do. 

     • Vacuuming: Most kids love the vacuum so this is a great starter chore for a five-year-old. Make sure to go through the do’s and don’t of vacuuming and remind your child not everything gets picked up by a 
       vacuum. 
     • Bedroom Cleanup: Teach your child to take care of their bedrooms. Start with picking up toys, this can be made into a game by color or items. Another good way of teaching children to fine-tune their bedrooms is 
       lining up their shoes, making their beds, and ensuring nothing is out of order. 
     • Separate Laundry: At 5 a children can easily distinguish between colors, whites, and towels. Have your children separate the laundry to make the task of washing clothes easier and more efficient. 
     • Folding Laundry: This is a great time to teach children how to fold laundry. At 5 they won’t be perfect but little hands can handle washcloths, hand towels, and mating socks. 
     • Putting Away Groceries: This is a good way for children to learn everything has a place to go. With your guidance, a small task like putting away groceries will teach them later on how to put away and keep things 
       neat. 
     • Dusting: Grab a duster and teach your child how to carefully remove things to dust tables, handrails, or other kid-friendly dustable areas. 
     • Set and Clean The Table: Most households with small kids have kid-friendly items for the table. Allow your child to set the table before dinner.  After dinner teach your child to remove their items and take them to 
        the kitchen sink to be rinsed off. This chore will become a healthy habit. 
     • Make The Bed: At five this won’t be perfect, but a child will be able to learn to set the pillow properly and adjust their bedding so it's more tidy. If they use extra blankets at night teach them to put them away or 
       fold them at the bottom of the bed. Don’t forget to adjust the stuffed animals!

Model Details

Base Model: unsloth/gemma-3-1b-it
Parameter Count: 999,885,952
Precision: torch.bfloat16

Hardware

GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition
Announced: Mar 17th, 2025
Release Date: Mar 18th, 2025
Memory Type: GDDR7
Bandwidth: 1.79 TB/s
Memory Size: 96 GB
Memory Bus: 512 bit
Shading Units: 24064
TDP: 600W

Training Settings

PEFT

Rank: 32
LoRA alpha: 64
Modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Gradient checkpointing: unsloth

SFT

Epoch: 2
Batch size: 16
Gradient Accumulation steps: 1
Warmup ratio: 0.05
Learning rate: 0.0004
Optimizer: adamw_torch_fused
Learning rate scheduler: cosine

Training stats

Date: 2026-03-24T11:16:12.098497
Peak VRAM usage: 34.73 GB
Global step: 1850
Training runtime (seconds): 958.4314
Average training loss: 1.6259772120295344
Final validation loss: 1.7383391857147217

Framework versions

Unsloth: 2026.3.10
TRL: 0.22.2
Transformers: 4.56.2
Pytorch: 2.10.0+cu128
Datasets: 4.8.4
Tokenizers: 0.22.2

License

This model is released under the Gemma license. See the Gemma Terms of Use and Prohibited Use Policy regarding the use of Gemma-generated content.

Downloads last month: 19

Safetensors

Model size

1.0B params

Tensor type

BF16

Model tree for kth8/gemma-3-1b-it-IFeval

Base model

google/gemma-3-1b-pt

Finetuned

google/gemma-3-1b-it

Finetuned

unsloth/gemma-3-1b-it

Finetuned

(556)

this model

kth8
/

gemma-3-1b-it-IFeval

Usage example

Model Details

Hardware

Training Settings

PEFT

SFT

Training stats

Framework versions

License

Model tree for kth8/gemma-3-1b-it-IFeval

Dataset used to train kth8/gemma-3-1b-it-IFeval