Instructions to use minjaechoi/EXAONE-4.0-1.2B-W4A16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use minjaechoi/EXAONE-4.0-1.2B-W4A16 with Transformers:

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

pipe = pipeline("text-generation", model="minjaechoi/EXAONE-4.0-1.2B-W4A16")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("minjaechoi/EXAONE-4.0-1.2B-W4A16")
model = AutoModelForCausalLM.from_pretrained("minjaechoi/EXAONE-4.0-1.2B-W4A16", 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 minjaechoi/EXAONE-4.0-1.2B-W4A16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "minjaechoi/EXAONE-4.0-1.2B-W4A16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "minjaechoi/EXAONE-4.0-1.2B-W4A16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/minjaechoi/EXAONE-4.0-1.2B-W4A16

SGLang

How to use minjaechoi/EXAONE-4.0-1.2B-W4A16 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 "minjaechoi/EXAONE-4.0-1.2B-W4A16" \
    --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": "minjaechoi/EXAONE-4.0-1.2B-W4A16",
		"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 "minjaechoi/EXAONE-4.0-1.2B-W4A16" \
        --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": "minjaechoi/EXAONE-4.0-1.2B-W4A16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use minjaechoi/EXAONE-4.0-1.2B-W4A16 with Docker Model Runner:
```
docker model run hf.co/minjaechoi/EXAONE-4.0-1.2B-W4A16
```

EXAONE-4.0-1.2B — W4A16 (GPTQ, int4 weight-only)

4-bit weight-only (W4A16) quantization of LGAI-EXAONE/EXAONE-4.0-1.2B, produced with LLM Compressor's GPTQModifier. All Linear layers are quantized except lm_head, which is kept at full precision.

This is a research/educational-use-only derivative of an EXAONE model. The EXAONE license (see LICENSE in this repo) prohibits commercial use and external distribution beyond research/educational purposes without a separate written agreement with LG AI Research; the modified model name is prefixed with EXAONE per license section 2.1.c, and a copy of the license agreement is included in this repo per section 2.1.d. By using these weights you agree to the terms of that license, not just this card.

Quantization method


Method	GPTQ (`GPTQModifier`, LLM Compressor)
Scheme	W4A16 (4-bit weights, 16-bit activations)
Weight dtype	INT4, symmetric
Group size	128
Quantized targets	`Linear` (all layers except `lm_head`)
Activation ordering	static
Dampening frac	0.01
Output format	`compressed-tensors` (`pack-quantized`)

Recipe used (recipe.yaml, included in this repo):

default_stage:
  default_modifiers:
    GPTQModifier:
      targets: [Linear]
      ignore: [lm_head]
      scheme: W4A16
      block_size: 128
      dampening_frac: 0.01
      actorder: static
      requires_calibration_data: true

Calibration set

256 packed sequences × 2048 tokens = 524,288 calibration tokens, sampled (seed=42) from a mixed Korean/English instruction + function-calling corpus, targeting the following source composition:

Source	Target %	Actual % (this model)	Tokens (this model)
KRX-Data/Won-Instruct	35%	34.51%	180,954
heegyu/glaive-function-calling-v2-ko	30%	30.37%	159,242
NousResearch/hermes-function-calling-v1 (json-mode-agentic.json)	10%	9.90%	51,906
heegyu/open-korean-instructions	10%	9.88%	51,805
kuotient/gsm8k-ko	5%	5.11%	26,806
in-house synthetic data (SafeCommit project, programmatically generated)	10%	10.22%	53,575

Raw pool before filtering/dedup: 49,083 examples (49,017 after removing 65 duplicates)
Quota-sampled for calibration: 954 examples → 877 packed into the final 256×2048 blocks
Tool-calling trajectories in the pool: 8,897 (2,819 multi-tool)
Contamination check: gsm8k-ko: train split only used; no BFCL/AgentDojo/tau-bench/SafeCommit-eval sources included

safecommit_synth is unreleased in-house synthetic data from the SafeCommit project, not a public HF dataset.

How to run

vLLM (recommended — required for the `compressed-tensors` W4A16 kernels used here)

vllm serve minjaechoi/EXAONE-4.0-1.2B-W4A16 --served-model-name exaone4-1.2b

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "exaone4-1.2b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Transformers (requires the `compressed-tensors` package for int4 dequant kernels)

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "minjaechoi/EXAONE-4.0-1.2B-W4A16"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

msgs = [{"role": "user", "content": "Hello!"}]
inputs = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt").to(model.device)
out = model.generate(inputs, max_new_tokens=256)
print(tok.decode(out[0], skip_special_tokens=True))

Benchmarks

Evaluated on this W4A16 checkpoint (not compared against an FP16 baseline run in this project yet):

AgentDojo (banking suite, utility-only, 16 tasks): 5/16 tasks passed (31.2%) — utility-only scoring, no prompt-injection attacks included.
BFCL v4: attempted, but the evaluation harness used had a request-routing bug (--local-model-path was not substituted into the API request, so every call 404'd against the vLLM server) — the resulting scores are invalid and are intentionally not published here. Will be updated after a corrected re-run.

Files

model.safetensors — quantized weights (compressed-tensors pack-quantized format)
config.json — includes the quantization_config (compressed-tensors) needed by vLLM/transformers to load this checkpoint
recipe.yaml — the exact LLM Compressor recipe used to produce this checkpoint
tokenizer.json, tokenizer_config.json, chat_template.jinja — tokenizer/chat template, copied unmodified from the base model
LICENSE — base model license, included per its terms

License

This checkpoint is a derivative of LGAI-EXAONE/EXAONE-4.0-1.2B and is distributed under the same license (other, see LICENSE in this repo). No additional restrictions are added beyond the base model's license (see the notice above for EXAONE-specific terms).

Downloads last month: -

Safetensors

Model size

1B params

Tensor type

I32

BF16

Model tree for minjaechoi/EXAONE-4.0-1.2B-W4A16

Base model

LGAI-EXAONE/EXAONE-4.0-1.2B

Quantized

(35)

this model