---
license: apache-2.0
task_categories:
  - text-generation
tags:
  - dna
  - long-context
  - retrieval
  - benchmark
  - niah
configs:
  - config_name: niah_4k
    data_files:
      - split: test
        path: data/niah_4k/test-*.parquet
  - config_name: niah_8k
    data_files:
      - split: test
        path: data/niah_8k/test-*.parquet
  - config_name: niah_16k
    data_files:
      - split: test
        path: data/niah_16k/test-*.parquet
  - config_name: niah_32k
    data_files:
      - split: test
        path: data/niah_32k/test-*.parquet
  - config_name: niah_64k
    data_files:
      - split: test
        path: data/niah_64k/test-*.parquet
  - config_name: niah_128k
    data_files:
      - split: test
        path: data/niah_128k/test-*.parquet
  - config_name: niah_neardup_d4_4k
    data_files:
      - split: test
        path: data/niah_neardup_d4_4k/test-*.parquet
  - config_name: niah_neardup_d4_8k
    data_files:
      - split: test
        path: data/niah_neardup_d4_8k/test-*.parquet
  - config_name: niah_neardup_d4_16k
    data_files:
      - split: test
        path: data/niah_neardup_d4_16k/test-*.parquet
  - config_name: niah_neardup_d4_32k
    data_files:
      - split: test
        path: data/niah_neardup_d4_32k/test-*.parquet
  - config_name: niah_neardup_d4_64k
    data_files:
      - split: test
        path: data/niah_neardup_d4_64k/test-*.parquet
  - config_name: niah_neardup_d4_128k
    data_files:
      - split: test
        path: data/niah_neardup_d4_128k/test-*.parquet
  - config_name: niah_neardup_d2_4k
    data_files:
      - split: test
        path: data/niah_neardup_d2_4k/test-*.parquet
  - config_name: niah_neardup_d2_8k
    data_files:
      - split: test
        path: data/niah_neardup_d2_8k/test-*.parquet
  - config_name: niah_neardup_d2_16k
    data_files:
      - split: test
        path: data/niah_neardup_d2_16k/test-*.parquet
  - config_name: niah_neardup_d2_32k
    data_files:
      - split: test
        path: data/niah_neardup_d2_32k/test-*.parquet
  - config_name: niah_neardup_d2_64k
    data_files:
      - split: test
        path: data/niah_neardup_d2_64k/test-*.parquet
  - config_name: niah_neardup_d2_128k
    data_files:
      - split: test
        path: data/niah_neardup_d2_128k/test-*.parquet
  - config_name: niah_neardup_d1_4k
    data_files:
      - split: test
        path: data/niah_neardup_d1_4k/test-*.parquet
  - config_name: niah_neardup_d1_8k
    data_files:
      - split: test
        path: data/niah_neardup_d1_8k/test-*.parquet
  - config_name: niah_neardup_d1_16k
    data_files:
      - split: test
        path: data/niah_neardup_d1_16k/test-*.parquet
  - config_name: niah_neardup_d1_32k
    data_files:
      - split: test
        path: data/niah_neardup_d1_32k/test-*.parquet
  - config_name: niah_neardup_d1_64k
    data_files:
      - split: test
        path: data/niah_neardup_d1_64k/test-*.parquet
  - config_name: niah_neardup_d1_128k
    data_files:
      - split: test
        path: data/niah_neardup_d1_128k/test-*.parquet
---

# Genome-NIAH

A NIAH-style needle-in-a-haystack benchmark for genomic language models, with
a DNA-specific **near-duplicate** distractor extension that's not in RULER.

## Tasks

| Task | What it tests |
|---|---|
| `niah` | Plain NIAH: insert (KEY, VALUE) at a random depth, query KEY → expect VALUE. The cliff diagnostic. |
| `niah_neardup_d4` | NIAH + 8 distractor (KEY', VALUE') pairs where KEY' differs from KEY by 4 bp (83 % key identity). |
| `niah_neardup_d2` | Same but Δ = 2 bp (92 % key identity). |
| `niah_neardup_d1` | Same but Δ = 1 bp (96 % key identity) — the discrimination test. |

## Context lengths

Six lengths in 6-mer-tokens (Carbon convention) → bp:

| ctx (tokens) | ctx (bp) | mode |
|---:|---:|---|
|   4 k |   24 kbp | contiguous (single ACGT run from one OG2 record) |
|   8 k |   49 kbp | contiguous |
|  16 k |   98 kbp | contiguous |
|  32 k |  197 kbp | **stitched-within-record** |
|  64 k |  393 kbp | stitched |
| 128 k |  786 kbp | stitched |

**Why two modes?** OpenGenome2 records are pre-chunked at 128 KiB max
contiguous ACGT (gaps from N's, soft-masked repeats, and metadata break the
sequence). For ctx ≤ 16 k tokens we use single contiguous runs from one OG2
record. For ctx ≥ 32 k tokens we **stitch all ACGT runs from a single record
together** (gaps removed within the same chromosome) — this matches how DNA
language models are trained and preserves chromosomal locality.

The `haystack_mode` column distinguishes `"real"` (contiguous) from
`"real_stitched"` (stitched).

## Subset structure (24 configs total)

Each config = one (task, ctx) pair. Layout matches RULER's HF convention:

```
data/<task>_<ctx>/test-00000-of-00001.parquet
```

```python
from datasets import load_dataset
ds = load_dataset("hf-carbon/genome-niah", "niah_16k")        # 500 rows, 5 depths × 100/cell
ds = load_dataset("hf-carbon/genome-niah", "niah_neardup_d1_64k")
```

## Sampling design (per config)

- **n = 500 retrieval rows** per (task, ctx) — matches RULER's per-cell scale
- **5 needle depths** (10 / 25 / 50 / 75 / 90 %), 100 rows each
- **Stratified across 4 species** (animalia / fungi / plantae / protista)
  matching Carbon's training distribution; 25 rows per (depth, species)
- Wilson 95 % CI half-widths: ± 9.8 % per-cell, ± 4.4 % per-(task, ctx)

## Source

OpenGenome2 NCBI eukaryotic test + valid splits across all 8 batches:
[arcinstitute/opengenome2 → midtraining_specific/ncbi_eukaryotic_genomes/](https://huggingface.co/datasets/arcinstitute/opengenome2/tree/main/json/midtraining_specific/ncbi_eukaryotic_genomes).
Chromista is dropped (not in Carbon's training mix; also has the worst
fragmentation under the [ACGT]+ filter).

## Schema

| column | dtype | description |
|---|---|---|
| `uid` | string | deterministic SHA1 row id |
| `haystack_mode` | string | `real` (contiguous) or `real_stitched` |
| `source` | string | `animalia` / `fungi` / `plantae` / `protista` |
| `context_length_tokens` | int | 6-mer tokens (× 6 = bp) |
| `context_length_bp` | int | length of `positive_sequence` |
| `depth` | int | needle insertion depth as percentage |
| `condition` | string | `retrieval` (only — no `no_memory` in this release) |
| `key`, `value` | string | random 24 bp DNA strings |
| `negative_value` | string | `value` with all 24 bp mutated to a different base |
| `prompt` | string | `<dna>` + haystack with embedded (KEY, VALUE) + trailing KEY |
| `positive_sequence` | string | `prompt` + `value` |
| `negative_sequence` | string | `prompt` + `negative_value` |
| `target_insert_start_bp` / `target_insert_end_bp` | int | needle position (`niah` only) |
| `distractor_keys`, `distractor_values` | string | comma-separated lists (`niah_neardup_*` only) |

## Evaluation metrics

- `gen_exact_match` — strict: greedy decode (DNA-token-restricted) matches `value` char-for-char.
- `ll_correct` — lenient: `LL(value | prompt) > LL(negative_value | prompt)`.

## Citation

If you use this benchmark, please cite the Carbon DNA model project (Hugging Face).

Built with the `build_dna_long_context_retrieval.py` /
`build_dna_lcr_harder.py` pipelines in the carbon-long-context repo.