Carbon-3B GGUF
Available files
| Quant | File | Size |
|---|---|---|
| BF16 | carbon-3b-bf16.gguf |
6.4 GB |
| Q8_0 | carbon-3b-q8_0.gguf |
3.4 GB |
| Q6_K | carbon-3b-q6_k.gguf |
2.8 GB |
| Q5_K_M | carbon-3b-q5_k_m.gguf |
2.4 GB |
| Q4_K_M | carbon-3b-q4_k_m.gguf |
2.1 GB |
K-quants are produced with --token-embedding-type Q8_0 --output-tensor-type Q8_0
so the sensitive embedding/output tensors stay at Q8_0 precision (important for
this model's tight DNA-vs-text logit margins).
How to use
Click Use this model > llama.cpp on this page for a ready-made snippet. To install llama.cpp, follow llama.app or download a prebuilt binary from the releases page.
Carbon-3B is a base DNA model, not a chat model, so use text completion rather than the chat UI. Start the prompt
with <dna>:
llama completion -hf HuggingFaceBio/Carbon-3B-GGUF:Q8_0 -p '<dna>ATGCGCTAGCTACGATCGATCGTAGCTAGCTAGCTAGCTACG' -n 128 --temp 0
llama serve -hf HuggingFaceBio/Carbon-3B-GGUF:Q8_0 serves the same model over HTTP; send raw prompts to its /completion endpoint.
- Metadata conditioning: prefix the prompt with tags, e.g.
<vertebrate_mammalian><protein_coding_region><dna>ATGCGCTAG. - Speculative decoding: add
-hfd HuggingFaceBio/Carbon-500M-GGUF:Q8_0to use Carbon-500M, which shares the vocabulary, as a draft model (about 1.35x faster at temperature 0). - Long context: for up to 65k tokens (about 393 kbp), add
-c 65536 --rope-scaling yarn --rope-scale 4 --yarn-orig-ctx 32768. - Likelihoods:
llama perplexity -hf HuggingFaceBio/Carbon-3B-GGUF:Q8_0 -f dna.txtgives corpus perplexity. For per-sequence log-likelihoods, use the Pythonscore()from the source card.
Tokenization parity
llama.cpp produces byte-for-byte identical token IDs to the Python HybridDNATokenizer (loaded with trust_remote_code=True) on the standard <dna>/metadata/edge-case fixtures shipped in ggml-org/vocabs.
See also
- Source weights: HuggingFaceBio/Carbon-3B
- Other GGUF variants: 500M · 3B · 8B
License
Apache-2.0, inherited from the source model.