Instructions to use HuggingFaceBio/Carbon-8B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HuggingFaceBio/Carbon-8B with Transformers:

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

pipe = pipeline("text-generation", model="HuggingFaceBio/Carbon-8B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("HuggingFaceBio/Carbon-8B")
model = AutoModelForCausalLM.from_pretrained("HuggingFaceBio/Carbon-8B")
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

vLLM

How to use HuggingFaceBio/Carbon-8B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HuggingFaceBio/Carbon-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": "HuggingFaceBio/Carbon-8B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/HuggingFaceBio/Carbon-8B

SGLang

How to use HuggingFaceBio/Carbon-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 "HuggingFaceBio/Carbon-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": "HuggingFaceBio/Carbon-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 "HuggingFaceBio/Carbon-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": "HuggingFaceBio/Carbon-8B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use HuggingFaceBio/Carbon-8B with Docker Model Runner:
```
docker model run hf.co/HuggingFaceBio/Carbon-8B
```

loubnabnl HF Staff commited on about 20 hours ago

Commit

6fac784

verified ·

1 Parent(s): 578c911

update tata and syn

Browse files

Files changed (1) hide show

README.md +4 -4

README.md CHANGED Viewed

@@ -63,12 +63,12 @@ All evaluations are zero-shot and use the [public Carbon evaluation pipeline](ht
 | | ClinVar coding (24 kb) | 92.89 | **93.11** | +0.22 |
 | | ClinVar non-coding (24 kb) | 91.14 | **91.63** | +0.49 |
 | Perturbation | Nucleotide triplet-expansion | 85.20 | **89.05** | +3.85 |
-| | Synonymous codon | 88.89 | **91.46** | +2.57 |
-| Long-context retrieval | Genome-NIAH @ 393 kbp | 79.00 | **86.00** | +7.00 |
-### Genome-NIAH (long-context retrieval)
-Genome-NIAH measures how well a DNA model actually *uses* its long context. See the [`hf-carbon/genome-niah` dataset card](https://huggingface.co/datasets/hf-carbon/genome-niah) for the benchmark design.
 | Context length         | Carbon 3B (native / YaRN 4×) | Carbon 8B (native / YaRN 4×) | Evo2 7B |
 |------------------------|------------------------------|------------------------------|---------|

 | | ClinVar coding (24 kb) | 92.89 | **93.11** | +0.22 |
 | | ClinVar non-coding (24 kb) | 91.14 | **91.63** | +0.49 |
 | Perturbation | Nucleotide triplet-expansion | 85.20 | **89.05** | +3.85 |
+| | Synonymous codon replacement | 88.89 | **91.46** | +2.57 |
+| Long-context retrieval | Genomic-NIAH @ 393 kbp | 79.00 | **86.00** | +7.00 |
+### Genomic-NIAH (long-context retrieval)
+Genomic-NIAH measures how well a DNA model actually *uses* its long context. See the [`HuggingFaceBio/genomic-niah` dataset card](https://huggingface.co/datasets/HuggingFaceBio/genomic-niah) for the benchmark design.
 | Context length         | Carbon 3B (native / YaRN 4×) | Carbon 8B (native / YaRN 4×) | Evo2 7B |
 |------------------------|------------------------------|------------------------------|---------|