Instructions to use HuggingFaceBio/Carbon-500M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HuggingFaceBio/Carbon-500M with Transformers:

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

pipe = pipeline("text-generation", model="HuggingFaceBio/Carbon-500M")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("HuggingFaceBio/Carbon-500M")
model = AutoModelForCausalLM.from_pretrained("HuggingFaceBio/Carbon-500M")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use HuggingFaceBio/Carbon-500M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HuggingFaceBio/Carbon-500M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceBio/Carbon-500M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/HuggingFaceBio/Carbon-500M

SGLang

How to use HuggingFaceBio/Carbon-500M 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-500M" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceBio/Carbon-500M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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-500M" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceBio/Carbon-500M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use HuggingFaceBio/Carbon-500M with Docker Model Runner:
```
docker model run hf.co/HuggingFaceBio/Carbon-500M
```

kashif HF Staff commited on 1 day ago

Commit

47d83fe

1 Parent(s): 19273a6

tokenizer: expose .vocab property for fast-tokenizer-style callers (#3)

Browse files

- tokenizer: expose .vocab property for fast-tokenizer-style callers (54d1dc0a02a2fbce1fc599d4e52d54a1fb49c905)

Files changed (1) hide show

tokenizer.py +8 -0

tokenizer.py CHANGED Viewed

@@ -144,6 +144,14 @@ class HybridDNATokenizer(PreTrainedTokenizer):
     def get_vocab(self) -> Dict[str, int]:
         return self._vocab.copy()
     def __len__(self):
         # Override default (len(get_vocab())) because get_vocab() deduplicates
         # CCCCCC which exists as both BPE (ID 91443) and DNA 6-mer (ID 154402).

     def get_vocab(self) -> Dict[str, int]:
         return self._vocab.copy()
+    @property
+    def vocab(self) -> Dict[str, int]:
+        # Compatibility shim: fast tokenizers (PreTrainedTokenizerFast) expose
+        # `tokenizer.vocab` as a property; slow PreTrainedTokenizer subclasses
+        # like this one only expose `get_vocab()`. Some downstream tools
+        # (e.g. llama.cpp's convert_hf_to_gguf.py) read `.vocab` directly.
+        return self._vocab
     def __len__(self):
         # Override default (len(get_vocab())) because get_vocab() deduplicates
         # CCCCCC which exists as both BPE (ID 91443) and DNA 6-mer (ID 154402).