Instructions to use nyu-visionx/cambrian-8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nyu-visionx/cambrian-8b with Transformers:

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

pipe = pipeline("text-generation", model="nyu-visionx/cambrian-8b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("nyu-visionx/cambrian-8b", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use nyu-visionx/cambrian-8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nyu-visionx/cambrian-8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nyu-visionx/cambrian-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nyu-visionx/cambrian-8b

SGLang

How to use nyu-visionx/cambrian-8b 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 "nyu-visionx/cambrian-8b" \
    --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": "nyu-visionx/cambrian-8b",
		"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 "nyu-visionx/cambrian-8b" \
        --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": "nyu-visionx/cambrian-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nyu-visionx/cambrian-8b with Docker Model Runner:
```
docker model run hf.co/nyu-visionx/cambrian-8b
```

cambrian-8b / config.json

craigwu

fix name

86557ae verified almost 2 years ago

raw

history blame contribute delete

1.97 kB

	{
	"_name_or_path": "./Meta-Llama-3-8B-Instruct",
	"architectures": [
	"CambrianLlamaForCausalLM"
	],
	"attention_bias": false,
	"attention_dropout": 0.0,
	"bos_token_id": 128000,
	"connector_depth": 3,
	"connector_only": false,
	"eos_token_id": 128001,
	"freeze_mm_mlp_adapter": false,
	"hidden_act": "silu",
	"hidden_size": 4096,
	"image_aspect_ratio": "pad",
	"image_position": 91,
	"image_token_len": 576,
	"initializer_range": 0.02,
	"intermediate_size": 14336,
	"max_position_embeddings": 8192,
	"mm_patch_merge_type": "flat",
	"mm_projector_lr": null,
	"mm_projector_type": "sva",
	"mm_use_im_patch_token": false,
	"mm_use_im_start_end": false,
	"mm_vision_sampler_lr": null,
	"mm_vision_select_feature": "patch",
	"mm_vision_select_layer": -2,
	"mm_vision_tower_aux_list": [
	"siglip/CLIP-ViT-SO400M-14-384",
	"openai/clip-vit-large-patch14-336",
	"facebook/dinov2-giant-res378",
	"clip-convnext-XXL-multi-stage"
	],
	"mm_vision_tower_aux_token_len_list": [
	576,
	576,
	576,
	9216
	],
	"mm_vision_tower_lr": null,
	"model_type": "cambrian_llama",
	"num_attention_heads": 32,
	"num_hidden_layers": 32,
	"num_key_value_heads": 8,
	"num_of_vision_sampler_layers": 10,
	"num_query_group": 1,
	"pretraining_tp": 1,
	"query_num_list": [
	576
	],
	"rms_norm_eps": 1e-05,
	"rope_scaling": null,
	"rope_theta": 500000.0,
	"spmd_debug": null,
	"spmd_fsdp_sharding": null,
	"spmd_mesh": null,
	"start_of_vision_sampler_layers": 0,
	"stride_of_vision_sampler_layers": 3,
	"tie_word_embeddings": false,
	"tokenizer_model_max_length": 2048,
	"tokenizer_padding_side": "right",
	"torch_dtype": "float32",
	"transformers_version": "4.40.1",
	"tune_mm_mlp_adapter": false,
	"unfreeze_mm_vision_tower": false,
	"use_cache": true,
	"use_mm_proj": true,
	"vision_hidden_size": 1024,
	"vision_tower_aux_token_len_list": [
	576,
	576,
	576,
	9216
	],
	"vocab_size": 128256
	}