Instructions to use jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32 with Transformers:

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

pipe = pipeline("image-text-to-text", model="jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForCausalLM

processor = AutoProcessor.from_pretrained("jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32", trust_remote_code=True, device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32

SGLang

How to use jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32 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 "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32" \
    --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": "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32" \
        --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": "jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32 with Docker Model Runner:
```
docker model run hf.co/jkim96/EXAONE-4.5-33B-DASHQ-INT4-g32
```

EXAONE-4.5-33B-DASHQ-INT4-g32 / config.json

jkim96

Add DASH-Q remote-code inference (Triton decode kernel)

5d3bf36 verified about 1 month ago

Raw

History Blame Contribute Delete

4.33 kB

	{
	"architectures": [
	"DashQExaone4_5_ForConditionalGeneration"
	],
	"auto_map": {
	"AutoModelForCausalLM": "modeling_dashq.DashQExaone4_5_ForConditionalGeneration",
	"AutoModelForImageTextToText": "modeling_dashq.DashQExaone4_5_ForConditionalGeneration"
	},
	"dashq": {
	"format": "dashq-packed-linear",
	"format_version": 1,
	"layer_metadata": "dashq_config.json",
	"method": "dashq",
	"n_quantized_modules": 448,
	"params": {
	"bits": 4,
	"group_size": 32,
	"low_memory_optimization": false,
	"moe_hessian_scope": "shared",
	"n_samples": 128,
	"scale_zero_dtype": "float16",
	"symmetric": false,
	"use_error_compensation": true,
	"use_optimal_shrinkage": true,
	"use_weighted_quantization": true
	}
	},
	"dtype": "bfloat16",
	"image_token_id": 67,
	"model_type": "exaone4_5",
	"text_config": {
	"_num_mtp_layers": 1,
	"architectures": [
	"Exaone4ForCausalLM"
	],
	"attention_dropout": 0.0,
	"bos_token_id": 1,
	"dtype": "bfloat16",
	"eos_token_id": 53,
	"hidden_act": "silu",
	"hidden_size": 5120,
	"initializer_range": 0.02,
	"intermediate_size": 27392,
	"layer_types": [
	"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"
	],
	"max_position_embeddings": 262144,
	"model_type": "exaone4",
	"mtp_loss_scaling_factor": 0.05,
	"mtp_share_layers": true,
	"num_attention_heads": 40,
	"num_hidden_layers": 64,
	"num_key_value_heads": 8,
	"num_nextn_predict_layers": 1,
	"pad_token_id": 0,
	"rms_norm_eps": 1e-05,
	"rope_parameters": {
	"factor": 16.0,
	"high_freq_factor": 4.0,
	"low_freq_factor": 1.0,
	"original_max_position_embeddings": 8192,
	"rope_theta": 1000000.0,
	"rope_type": "llama3"
	},
	"sliding_window": 4096,
	"sliding_window_pattern": "LLLG",
	"tie_word_embeddings": false,
	"use_cache": true,
	"vocab_size": 153600
	},
	"tie_word_embeddings": false,
	"transformers_version": "5.8.0",
	"use_cache": false,
	"video_token_id": 68,
	"vision_config": {
	"depth": 28,
	"dtype": "bfloat16",
	"fullatt_block_indexes": [
	6,
	13,
	20,
	27
	],
	"hidden_act": "silu",
	"hidden_size": 2048,
	"in_channels": 3,
	"initializer_range": 0.02,
	"intermediate_size": 5120,
	"model_type": "exaone4_5_vision",
	"num_heads": 32,
	"num_key_value_heads": 8,
	"out_hidden_size": 5120,
	"patch_size": 14,
	"spatial_merge_size": 2,
	"temporal_patch_size": 2,
	"tokens_per_second": 2,
	"window_size": 112
	},
	"vision_end_token_id": 74,
	"vision_start_token_id": 73,
	"vision_token_id": 67,
	"vocab_size": 153600
	}