Initial Genome-NIAH release: 4 tasks × 6 ctxs = 24 configs, 12000 rows

README.md

@@ -0,0 +1,205 @@
+---
+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.

data/niah_128k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:190cda5de33621edae66a99580a193508a862daa2266e0741aa6d3c3bd825670
+size 540661475

data/niah_16k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ded1a097f2b4c1db1e46c64fa55a5650bfdf3c6fab25b6ae98f1146463e7a382
+size 67360514

data/niah_32k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f08a4fc2be562a4dbc04a174de224435e3d78105503fb92e5140cf0bb314399
+size 134536280

data/niah_4k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c61a397b374c0951a28cb4229fe9c61c87c1303140affcfd40ff4ebd0f8de786
+size 16866230

data/niah_64k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:682dca7649b6756af5d1cab2dac7091a2f4ab4863904d85127c9efafdf5dbce5
+size 269845940

data/niah_8k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b49af7c70d4e75a8c68dfc320791769a2343242021bb874b9a98cb30cda4cae0
+size 33668189

data/niah_neardup_d1_128k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da2e741548b9d4efcd74b39af1eb4312fa5a712e266c418cdbd9a05d75bab7c6
+size 538234039

data/niah_neardup_d1_16k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ead7f98ed9547db6bf8e0033c9cc1439765f6c93e481b4e69cbdda197d2946db
+size 67411573

data/niah_neardup_d1_32k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9322fd4b52c2538e9b5c8670e647a2542af6e2cb5a402f83877aba451223734a
+size 135042453

data/niah_neardup_d1_4k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f45e2a037aa9f3b6635bf8e0cd2d4d17d97d4cb9dc17ed9980986a6ae948d175
+size 16939613

data/niah_neardup_d1_64k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e370efe8f2c1e3520920c046498117d5d29203e1ab666abd256014c1a4546d29
+size 270192674

data/niah_neardup_d1_8k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38b03fc2212df0d478ef70e6aa2833a6a9445e87eadf277532f6c7d345a0432b
+size 33792639

data/niah_neardup_d2_128k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32c4d94c43bdf31da22e559b6873a75f3c317343a7f203ed8200fea75b5a0d04
+size 538947077

data/niah_neardup_d2_16k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0a0ebaf522164c5cf177fe562eb7e5c0b0fa60159fc294600e10b176f61f8f0
+size 67524883

data/niah_neardup_d2_32k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:210ce66420363a84e12b5efa1818b707b7439d8cffdaf35eddbc381b0d54daa0
+size 135136189

data/niah_neardup_d2_4k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9f49fee9787b34ce21cb7e1b9c87839de5ab3e620a18aa71e8ffc1dd82bc892
+size 17019833

data/niah_neardup_d2_64k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:606ba07d69befd3e6a380dfe41c44855d9a35c50a47419c02dd4b7fc0ce3ea16
+size 270104983

data/niah_neardup_d2_8k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:934ca547d8df3d603c851ea40b3dba7df53f1058a038c90f893dc8cbb9863c53
+size 33954610

data/niah_neardup_d4_128k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f187c83a90d60fda8f78ec529ec92b0f3cbb9cdd6609d8be46f59af65bd8aee
+size 539391268

data/niah_neardup_d4_16k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6229d97134c61ddbe4b135ec22aaf00ba90a5f434873ca4818664d1b6ae84c7d
+size 67449262

data/niah_neardup_d4_32k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fcc9e6916ed5bcb5f82f74159750f7d3a95ad9eab49b1e304c1ff530515215c
+size 135105382

data/niah_neardup_d4_4k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4eae58cde7984ea9dd9395de2dc762a2b2e26da2d8997bbd61cd649a247e6816
+size 17012405

data/niah_neardup_d4_64k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a08abadf74695ec5dbd895d0ecf6a10affd8b2c27dc2a9886245306d2b692389
+size 269891351

data/niah_neardup_d4_8k/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e352441afa4db6e5901a3e064ae0998daea767e79d2abc8c0fad810e9803d8be
+size 33835668