Update README.md
Browse files
README.md
CHANGED
|
@@ -43,7 +43,7 @@ Carbon-3B is the **flagship** model of the Carbon family. We also release [**Car
|
|
| 43 |
- **Native context: 32,768 tokens ≈ 197 kbp.** Extendable to 65,536 tokens (≈ 393 kbp) at inference time using YaRN.
|
| 44 |
- **Trained with a Cross-Entropy → Factorised Nucleotide Supervision (FNS) objective schedule** to bridge coarse tokenization and single-nucleotide resolution (see the Carbon technical report).
|
| 45 |
- **Metadata-conditioned**: optional species-type and gene-type metadata tokens enable conditional generation.
|
| 46 |
-
- **Efficient inference**:
|
| 47 |
|
| 48 |
Across our zero-shot evaluation suite — sequence recovery, four variant-effect-prediction (VEP) benchmarks (ClinVar coding, ClinVar non-coding, BRCA2, TraitGym Mendelian), and two sequence-level perturbation tasks (TATA-box and synonymous codon) — Carbon-3B is competitive with Evo2-7B. It additionally works well on long context and retrieves needles reliably from up to ≈ 393 kbp of distal context on the Genome-NIAH long-context benchmark, while remaining several times faster than Evo2-7B.
|
| 49 |
|
|
|
|
| 43 |
- **Native context: 32,768 tokens ≈ 197 kbp.** Extendable to 65,536 tokens (≈ 393 kbp) at inference time using YaRN.
|
| 44 |
- **Trained with a Cross-Entropy → Factorised Nucleotide Supervision (FNS) objective schedule** to bridge coarse tokenization and single-nucleotide resolution (see the Carbon technical report).
|
| 45 |
- **Metadata-conditioned**: optional species-type and gene-type metadata tokens enable conditional generation.
|
| 46 |
+
- **Efficient inference**: compatible with vLLM and other inference engines. Can generate over 100,000 base-pairs per second on a single H100 GPU.
|
| 47 |
|
| 48 |
Across our zero-shot evaluation suite — sequence recovery, four variant-effect-prediction (VEP) benchmarks (ClinVar coding, ClinVar non-coding, BRCA2, TraitGym Mendelian), and two sequence-level perturbation tasks (TATA-box and synonymous codon) — Carbon-3B is competitive with Evo2-7B. It additionally works well on long context and retrieves needles reliably from up to ≈ 393 kbp of distal context on the Genome-NIAH long-context benchmark, while remaining several times faster than Evo2-7B.
|
| 49 |
|