Instructions to use LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark with Transformers:

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

pipe = pipeline("text-generation", model="LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark")
model = AutoModelForCausalLM.from_pretrained("LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark", 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 LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark

SGLang

How to use LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark 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 "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark" \
    --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": "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark",
		"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 "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark" \
        --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": "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark with Docker Model Runner:
```
docker model run hf.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B-DSpark
```

K-EXAONE-2.0-750B-A37B-DSpark / config.json

LG-AI-EXAONE

Update config.json

9d153d5 about 21 hours ago

Raw

History Blame Contribute Delete

6.09 kB

	{
	"architectures": [
	"ExaoneMoeForCausalLM"
	],
	"attention_dropout": 0.0,
	"bos_token_id": 1,
	"dspark_block_size": 7,
	"dspark_confidence_head_with_markov": true,
	"dspark_enable_confidence_head": true,
	"dspark_markov_head_type": "vanilla",
	"dspark_markov_rank": 256,
	"dspark_noise_token_id": 84,
	"dspark_num_hidden_layers": 5,
	"dspark_num_target_layers": 78,
	"dspark_target_layer_ids": [
	2,
	20,
	38,
	56,
	74
	],
	"dtype": "bfloat16",
	"eos_token_id": 53,
	"head_dim": 128,
	"hidden_act": "silu",
	"hidden_size": 6144,
	"initializer_range": 0.02,
	"intermediate_size": 18432,
	"layer_types": [
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention",
	"sliding_attention",
	"sliding_attention",
	"sliding_attention",
	"full_attention"
	],
	"mask_token_id": 84,
	"max_position_embeddings": 262144,
	"mlp_layer_types": [
	"dense",
	"dense",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse"
	],
	"model_type": "exaone_moe",
	"moe_intermediate_size": 2048,
	"mtp_layer_types": [
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention"
	],
	"mtp_loss_scaling_factor": 0.05,
	"mtp_num_speculative_steps": 4,
	"mtp_sliding_windows": [
	0,
	0,
	0,
	0,
	0
	],
	"n_group": 1,
	"norm_topk_prob": true,
	"num_attention_heads": 64,
	"num_experts": 256,
	"num_experts_per_tok": 8,
	"num_hidden_layers": 78,
	"num_key_value_heads": 8,
	"num_nextn_predict_layers": 5,
	"num_shared_experts": 1,
	"pad_token_id": 0,
	"rms_norm_eps": 1e-05,
	"rope_parameters": {
	"rope_theta": 1000000,
	"rope_type": "default"
	},
	"routed_scaling_factor": 2.5,
	"scoring_func": "sigmoid",
	"sliding_windows": [
	0,
	4096,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0,
	128,
	128,
	128,
	0
	],
	"swiglu_limits": [
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	0.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0,
	7.0
	],
	"tie_word_embeddings": false,
	"tokenizer_class": "GPT2Tokenizer",
	"topk_group": 1,
	"topk_method": "noaux_tc",
	"transformers_version": "5.9.0",
	"use_cache": true,
	"vocab_size": 153600
	}