Gemma4-E4B-Function-Calling-xLAM-Unsloth

This model is a fine-tuned version of Gemma4-E4B-it optimized for function calling using Unsloth for 2x faster training and 60% less VRAM.

Trained on the Salesforce/xlam-function-calling-60k dataset, which contains 60,000 function calling examples with queries, tool definitions, and structured answers.

Overview

Property	Value
Developed by	ermiaazarkhalili
License	GEMMA
Language	English
Base Model	Gemma4-E4B-it
Model Size	4B parameters
Training Framework	Unsloth + TRL
Training Method	SFT with QLoRA (4-bit)
Context Length	2,048 tokens
GGUF Available	Gemma4-E4B-Function-Calling-xLAM-Unsloth-GGUF

Training Configuration

SFT + LoRA Settings

Parameter	Value
Unsloth Class	FastModel
Chat Template	gemma-4
Learning Rate	2e-4
Batch Size	2 per device
Gradient Accumulation	4 steps
Effective Batch Size	8
Max Steps	1,000
Optimizer	AdamW 8-bit
LR Scheduler	Linear
Warmup Steps	5
Precision	Auto (BF16/FP16)
Gradient Checkpointing	Enabled (Unsloth optimized)
Seed	3407

LoRA Configuration

Parameter	Value
LoRA Rank (r)	16
LoRA Alpha	16
LoRA Dropout	0
Quantization	4-bit QLoRA
Target Modules	attention + MLP (via FastModel)

Dataset

Property	Value
Dataset	xLAM Function Calling 60K
Training Samples	60,000
Format	XML-tagged: <query>, <tools>, <answers>

Hardware

Property	Value
GPU	NVIDIA H100 80GB HBM3 (MIG 3g.40gb slice)
Cluster	DRAC Fir (Compute Canada)
Execution	Papermill on SLURM

Usage

Quick Start (Transformers)

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {"role": "user", "content": "Check if the numbers 8 and 1233 are powers of two."}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7, do_sample=True)
response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
print(response)

Using with Unsloth (Fastest)

from unsloth import FastModel

model, tokenizer = FastModel.from_pretrained(
    "ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth",
    max_seq_length=2048,
    load_in_4bit=True,
)

4-bit Quantized Inference

from transformers import AutoModelForCausalLM, BitsAndBytesConfig
import torch

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
)

model = AutoModelForCausalLM.from_pretrained(
    "ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth",
    quantization_config=quantization_config,
    device_map="auto",
)

GGUF Versions

Quantized GGUF versions for CPU and edge inference are available at: Gemma4-E4B-Function-Calling-xLAM-Unsloth-GGUF

Format	Description
Q4_K_M	Recommended — good balance of quality and size
Q5_K_M	Higher quality, slightly larger
Q8_0	Near-lossless, largest GGUF size

Using with Ollama

ollama pull hf.co/ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth-GGUF:Q4_K_M
ollama run hf.co/ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth-GGUF:Q4_K_M "Check if the numbers 8 and 1233 are powers of two."

Using with llama.cpp

./llama-cli -m Gemma4-E4B-Function-Calling-xLAM-Unsloth-Q4_K_M.gguf -p "Check if the numbers 8 and 1233 are powers of two." -n 512

Limitations

• Language: Primarily trained on English data
• Knowledge Cutoff: Limited to base model's training data cutoff
• Hallucinations: May generate plausible-sounding but incorrect information
• Context Length: Fine-tuned with 2,048 token context window
• Safety: Not extensively safety-tuned; use with appropriate guardrails

Training Framework Versions

Package	Version
Unsloth	2026.4.4
TRL	0.24.0
Transformers	5.5.0
PyTorch	2.9.0
Datasets	4.3.0
PEFT	0.18.1
BitsAndBytes	0.49.2

Citation

@misc{ermiaazarkhalili_gemma4_e4b_function_calling_xlam_unsloth,
    author = {ermiaazarkhalili},
    title = {Gemma4-E4B-Function-Calling-xLAM-Unsloth: Fine-tuned Gemma4-E4B-it with Unsloth},
    year = {2026},
    publisher = {Hugging Face},
    howpublished = {\url{https://huggingface.co/ermiaazarkhalili/Gemma4-E4B-Function-Calling-xLAM-Unsloth}}
}

Acknowledgments

• Unsloth for 2x faster fine-tuning
• Base model developers (google)
• Hugging Face TRL Team for the training library
• Salesforce xLAM for the function calling dataset
• Compute Canada / DRAC for HPC resources