Field	Value
Source checkpoint	`ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth`
Base model	`unsloth/qwen3-4b-unsloth-bnb-4bit`
Dataset	`Salesforce/xlam-function-calling-60k`
Training	1 full epoch (effective batch=8)
Final training loss	0.2309 (job 36885894, runtime 2h 10m, peak VRAM 15.21 GB)
Conversion	Unsloth `save_pretrained` → llama.cpp `convert_hf_to_gguf.py` → `llama-quantize`

Available quantizations

File	Bits	Size	Notes
`qwen3-4b-function-calling-xlam-unsloth.q2_k.gguf`	2-bit	1.67 GB	Smallest; aggressive quality loss
`qwen3-4b-function-calling-xlam-unsloth.q3_k_m.gguf`	3-bit	2.08 GB	Small; noticeable quality loss
`qwen3-4b-function-calling-xlam-unsloth.q4_k_m.gguf`	4-bit	2.50 GB	Recommended — best size/quality balance
`qwen3-4b-function-calling-xlam-unsloth.q5_k_m.gguf`	5-bit	2.89 GB	Near-full quality
`qwen3-4b-function-calling-xlam-unsloth.q6_k.gguf`	6-bit	3.31 GB	Very close to Q8_0 at smaller size
`qwen3-4b-function-calling-xlam-unsloth.q8_0.gguf`	8-bit	4.28 GB	Largest; closest to bf16 source

Recommended default: Q4_K_M. For maximum fidelity, use Q8_0.

Usage

llama.cpp

llama-cli -hf ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth-GGUF --jinja -p "Find flights from SFO to NYC on December 25th" -n 256

Ollama

ollama run hf.co/ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth-GGUF:Q4_K_M

llama-cpp-python

python

from llama_cpp import Llama
llm = Llama.from_pretrained(
    repo_id="ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth-GGUF",
    filename="*q4_k_m.gguf",
    n_ctx=2048,
)
out = llm.create_chat_completion(
    messages=[{"role": "user", "content": "Find flights from SFO to NYC on December 25th"}],
    max_tokens=256,
)
print(out["choices"][0]["message"]["content"])

Intended use

For research and non-commercial experimentation. Outputs should be independently verified before any downstream use.

Limitations

GGUF quantizations have unavoidable quality loss vs the bf16 source. Use Q5_K_M+ for best fidelity.
Inherits all limitations of the source merged checkpoint (ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth).

Citation

bibtex

@misc{ qwen3_4b_xlam_unsloth_2026_gguf ,
  author = {Ermia Azarkhalili},
  title = { Qwen3-4B-Function-Calling-xLAM-Unsloth — GGUF quantized },
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/ermiaazarkhalili/Qwen3-4B-Function-Calling-xLAM-Unsloth-GGUF}}
}

This qwen3 model was trained 2× faster with Unsloth and Hugging Face's TRL library.