Instructions to use nvidia/Eagle2-9B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/Eagle2-9B with Transformers:

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

pipe = pipeline("image-text-to-text", model="nvidia/Eagle2-9B", 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 AutoModel
model = AutoModel.from_pretrained("nvidia/Eagle2-9B", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nvidia/Eagle2-9B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/Eagle2-9B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/Eagle2-9B",
		"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/nvidia/Eagle2-9B

SGLang

How to use nvidia/Eagle2-9B 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 "nvidia/Eagle2-9B" \
    --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": "nvidia/Eagle2-9B",
		"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 "nvidia/Eagle2-9B" \
        --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": "nvidia/Eagle2-9B",
		"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 nvidia/Eagle2-9B with Docker Model Runner:
```
docker model run hf.co/nvidia/Eagle2-9B
```

Deepspeed ZeRO3 Compatible Issue

by BK-Lee - opened Jan 28, 2025

Discussion

BK-Lee

Jan 28, 2025

•

edited Jan 28, 2025

I've faced the issue of compatiblity with DeepSpeed ZeRO3
Could you suggest a solution for it?

class SiglipMultiheadAttentionPoolingHead(nn.Module):
    """Multihead Attention Pooling."""

    def __init__(self, config: SiglipVisionConfig):
        super().__init__()

        self.probe = nn.Parameter(torch.randn(1, 1, config.hidden_size))
        self.attention = torch.nn.MultiheadAttention(config.hidden_size, config.num_attention_heads, batch_first=True) # this is the problem
        self.layernorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
        self.mlp = SiglipMLP(config)

    def forward(self, hidden_state):
        batch_size = hidden_state.shape[0]
        probe = self.probe.repeat(batch_size, 1, 1)

        hidden_state = self.attention(probe, hidden_state, hidden_state)[0] # this is the problem [The point Error Occrured!]

        residual = hidden_state
        hidden_state = self.layernorm(hidden_state)
        hidden_state = residual + self.mlp(hidden_state)

        return hidden_state[:, 0]

[rank3]:   File "lib/python3.11/site-packages/torch/nn/modules/activation.py", line 1275, in forward
[rank3]:     attn_output, attn_output_weights = F.multi_head_attention_forward(
[rank3]:                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank3]:   File "lib/python3.11/site-packages/torch/nn/functional.py", line 5533, in multi_head_attention_forward
[rank3]:     attn_output = linear(attn_output, out_proj_weight, out_proj_bias)
[rank3]:                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank3]:   File "lib/python3.11/site-packages/deepspeed/runtime/zero/linear.py", line 118, in zero3_linear_wrap
[rank3]:     return LinearFunctionForZeroStage3.apply(input, weight, bias)
[rank3]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank3]:   File "lib/python3.11/site-packages/torch/autograd/function.py", line 574, in apply
[rank3]:     return super().apply(*args, **kwargs)  # type: ignore[misc]
[rank3]:            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank3]:   File "lib/python3.11/site-packages/torch/amp/autocast_mode.py", line 455, in decorate_fwd
[rank3]:     return fwd(*args, **kwargs)
[rank3]:            ^^^^^^^^^^^^^^^^^^^^
[rank3]:   File "lib/python3.11/site-packages/deepspeed/runtime/zero/linear.py", line 62, in forward
[rank3]:     ret = torch.addmm(bias, input, weight.t())
[rank3]: RuntimeError: mat2 must be a matrix, got 1-D tensor

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment