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
```

LG-AI-EXAONE commited on about 21 hours ago

Commit

9d153d5

1 Parent(s): 54b1c9e

Update config.json

Browse files

Files changed (1) hide show

config.json +16 -14

config.json CHANGED Viewed

@@ -3,11 +3,23 @@
     "ExaoneMoeForCausalLM"
   ],
   "attention_dropout": 0.0,
-  "block_size": 7,
   "bos_token_id": 1,
-  "confidence_head_with_markov": true,
   "dtype": "bfloat16",
-  "enable_confidence_head": true,
   "eos_token_id": 53,
   "head_dim": 128,
   "hidden_act": "silu",
@@ -94,8 +106,7 @@
     "sliding_attention",
     "full_attention"
   ],
-  "markov_head_type": "vanilla",
-  "markov_rank": 256,
   "max_position_embeddings": 262144,
   "mlp_layer_types": [
     "dense",
@@ -196,7 +207,6 @@
     0
   ],
   "n_group": 1,
-  "noise_token_id": 84,
   "norm_topk_prob": true,
   "num_attention_heads": 64,
   "num_experts": 256,
@@ -205,7 +215,6 @@
   "num_key_value_heads": 8,
   "num_nextn_predict_layers": 5,
   "num_shared_experts": 1,
-  "num_target_layers": 78,
   "pad_token_id": 0,
   "rms_norm_eps": 1e-05,
   "rope_parameters": {
@@ -374,13 +383,6 @@
     7.0,
     7.0
   ],
-  "target_layer_ids": [
-    2,
-    20,
-    38,
-    56,
-    74
-  ],
   "tie_word_embeddings": false,
   "tokenizer_class": "GPT2Tokenizer",
   "topk_group": 1,

     "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",
     "sliding_attention",
     "full_attention"
   ],
+  "mask_token_id": 84,
   "max_position_embeddings": 262144,
   "mlp_layer_types": [
     "dense",
     0
   ],
   "n_group": 1,
   "norm_topk_prob": true,
   "num_attention_heads": 64,
   "num_experts": 256,
   "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": {
     7.0,
     7.0
   ],
   "tie_word_embeddings": false,
   "tokenizer_class": "GPT2Tokenizer",
   "topk_group": 1,