Add DenseOn and LateOn results

# Add DenseOn and LateOn HAKARI-Bench results

## Summary

This PR adds complete standard-scope results for LightOn's paired English
retrieval models:

| Model | Method | Revision | Files | Grouped Overall nDCG@10 |
| --- | --- | --- | ---: | ---: |
| `lightonai/DenseOn` | dense | `cb9947ebccb33862d24e3c7ca2edb25e51acd887` | 563 | 0.4232 |
| `lightonai/LateOn` | late interaction / exact MaxSim | `c01907b70557ee5c7753680d4819a5cce1674b83` | 563 | 0.4451 |

Both submissions contain all 563 standard evaluation tasks. The Overall values
above use 381 grouped score units derived from the 550 non-overlapping Overall
task results.

## NanoBEIR-en validation gate

The official Hugging Face model cards and checkpoint configs were reviewed
before evaluation. Both models were first evaluated on NanoBEIR-en and only
advanced to `--all` after the scores were consistent with the official 13-task
BEIR profiles.

| Model | NanoBEIR-en mean nDCG@10 | Pearson vs. official BEIR tasks | Spearman vs. official BEIR tasks |
| --- | ---: | ---: | ---: |
| `lightonai/DenseOn` | 0.677988 | 0.914556 | 0.901099 |
| `lightonai/LateOn` | 0.686811 | 0.911242 | 0.857143 |

The official model-card BEIR averages are 56.20 for DenseOn and 57.22 for
LateOn. Nano sampling changes the absolute aggregate, so the per-task
correlations were used as the primary reproduction check.

## Runtime and official settings

DenseOn uses the checkpoint's SentenceTransformers configuration: CLS pooling,
768 dimensions, cosine similarity, a 512-token maximum, and the registered
`query` / `document` prompts (`query: ` and `document: `). The dense submission
contains `base`, `int8`, `binary`, `int8_rescore`, and `binary_rescore` for every
task.

LateOn uses PyLate exact MaxSim with 128-dimensional token embeddings, query
length 32, document length 300 (checkpoint model maximum 299), `[Q] ` / `[D] `
prefixes, and query expansion disabled. It contains one exact-MaxSim base result
per task.

Both checkpoints were pinned by SHA and evaluated in fp32 with explicit SDPA.
No task failed and no batch-size or sequence-length adjustment was made.

Environment: Python 3.12.12, PyTorch 2.9.0, Transformers 5.12.1,
SentenceTransformers 5.4.1, Datasets 4.8.4, CUDA 12.8, and two NVIDIA GeForce
RTX 5090 GPUs (one independent worker per model).

## Commands

```bash
CUDA_VISIBLE_DEVICES=0 UV_PROJECT_ENVIRONMENT=tmp/.venv_eval_denseon \
uv run hakari-bench evaluate from-model-card \
--model-card config/model_cards/lightonai__DenseOn.yaml \
--all --batch-size 64 --device cuda:0

CUDA_VISIBLE_DEVICES=1 UV_PROJECT_ENVIRONMENT=tmp/.venv_eval_lateon \
uv run --group pylate hakari-bench evaluate from-model-card \
--model-card config/model_cards/lightonai__LateOn.yaml \
--all --batch-size 32 --device cuda:0
```

## Coverage and plausibility audit

- 1,126/1,126 expected `.json.xz` files are present and readable.
- Each model has 563 unique task keys, with no missing, extra, or duplicate task.
- All scores are finite and within `[0, 1]`; all resolved dataset revisions are
present and internally consistent.
- DenseOn has all five planned variants on every task; LateOn has the planned
exact-MaxSim base result on every task.
- In the merged leaderboard database, both models have 550/550 Overall tasks.
Base retrieval means are 0.417095 for DenseOn and 0.446855 for LateOn.
- Relative placement is plausible: DenseOn ranks above `BAAI/bge-small-en-v1.5`;
LateOn falls between LightOn's ColBERT-Zero/GTE-ModernColBERT models and the
smaller AnswerAI ColBERT baseline on Overall.

## Checklist

- [x] Results are under `hakari-results/lightonai__DenseOn/` and
`hakari-results/lightonai__LateOn/`.
- [x] Only compressed result JSON is included; no caches, DuckDB, HTML, or
scratch artifacts are submitted.
- [x] Result JSON records model/dataset revisions, runtime configuration, and
package/CUDA versions.
- [x] Official prompts, sequence lengths, attention choice, dtype, variants,
and late-interaction settings are documented above.

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/arguana.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7b19a890bd2f480a52aa424921d227600dc4108b479779752b0219ffb99e2cc
+size 165800

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/climatefever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:835de7a446417b21425d8bcf65698eed7494f8c2655d507a978037e8a1c00af7
+size 157400

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/dbpedia.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59a93692957efc50e0512fc8ad2377c59a62e4bcef7ec408ab5856a613df3cf8
+size 191120

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/fever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:046a2703b4c86077e7d191b7b0322f28a121b6ebec855987299c2c0776789ff9
+size 167760

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/fiqa2018.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:868aee77eb637a76431e876a933822f46cd74b3763fffb207ca2564fa2af1584
+size 120396

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/hotpotqa.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76d1d433fe5e5318b6cc19a20c056f168b9099c1cc06afbed257601ea13a664b
+size 131228

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/msmarco.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3434b7390757cf55299c9caf57b8d97f495c78872594bbe9c6c1e41b463ab2b2
+size 128464

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/nfcorpus.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f054ee898b4d5639ad47b5d1b2fc838d432169911e815709cb29f01d90fd73ed
+size 107788

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/nq.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:965d19f5703870d5d7840a4dac3cb4becf9dc9bd458097c77e79aa2943bc978c
+size 128076

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/quoraretrieval.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11978d360895010f3b9ae087cda3e8cef0de64542546e9ba594e597fa59f0549
+size 117912

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/scidocs.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79399acf5a4e9648d9e566aaf682e501f220b492bbb15e7f8408c0ce299a4fe2
+size 158496

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/scifact.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23fa3408522051f2765db83855b570670d06a28b720e07c53c84e2e637339079
+size 126296

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-ar/touche2020.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50ca4a3ce37943e9a58e4aedbcf4f22bd120919de20509d34bcad068ecb82447
+size 167508

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/arguana.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7c4023fd3da26c1e28acb7d008c6fd32301f03a9763b0f962f2bf35657345b4
+size 150768

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/climatefever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a59d57a8015272d834a46c96a9371e4160f271d3543f4dab9c82ac7316e5e293
+size 146856

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/dbpedia.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be9cc4a9cf08b03832e0e65ce8c5491b9af2937d3fc87f5389628532438ef302
+size 156500

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/fever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79dacd8e3ca75d17b79a63e722e92e1e26f46778b6ce89a208b6e4fc156f9b7b
+size 146732

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/fiqa2018.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83a6e67603f9d048d87733e1e896287b440c876240f49e832c9046988746e21c
+size 118708

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/hotpotqa.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf72f3ecf10b4bccce70ace4a398c65f9c3d653621c8563bed165cf6e14d61e0
+size 113212

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/msmarco.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2732ca2a9ed222f4ba614eee97e7d496be7b129283f7bda618ddb30848406f0
+size 101212

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/nfcorpus.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:361b390580b667f40d68bfe9102b2eafe0e41767a7dabd5528683f55f512f6a5
+size 102984

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/nq.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:018d435845dd7661d9c92ce4dfd2a4128da00d5d483165e7adf13acad3c5de10
+size 111020

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/quoraretrieval.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b812ed171966c6df330dcd000a53294bf07fedaaa990eb49901480536a7a2eb0
+size 107312

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/scidocs.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4de96d079663cc3c1390ecb671cbb8f500191697ceb98b98f45d7ef117ae6fd7
+size 164180

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/scifact.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c9bed91965065f0d749e337ea17075ab95fb6ed54bbfc242e8c52fd4788e921
+size 120460

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-de/touche2020.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02225d58a22dad06b36bff931adddaf723fb252194e0b1a34bbc7263fe8db5e8
+size 146060

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/arguana.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f35461299506a581a8ecdcf7ac9e8417793988c5d2a35e6d3e0d3dcd41181381
+size 135336

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/climatefever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:783bbd5c06f9161f00b6985f070f38ca4802083d4501998ed1a10e71b2c12a5b
+size 138028

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/dbpedia.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2cf2280eb1c5e55d8faf81e2bb542d036245115c74fef0e7088cc084b96d02c
+size 142524

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/fever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c07afcf07ea4c9aeef8c0b3c43cdd82c9ee65d9f8b364b0f05a5d407dc6d1283
+size 139304

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/fiqa2018.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0029608e144c0d94080c5461c531ddcce27093082e74ac8eb061b4c9b7c5a008
+size 105300

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/hotpotqa.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1660b4dbaaa46ac5ffc08dd4173ef5e72fbd75ffd7b0ddbc7684862c3b5013a7
+size 103772

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/msmarco.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c9802c0876e0f3969e7f6edf6299c844e5589dae049c97ca83247129d377718
+size 77820

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/nfcorpus.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27e1f6244b1d591a379ecc000f8eef59636e4e0c83a66a8e98363426cf90a77b
+size 95200

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/nq.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a7df540e54535419e414d1af4446e7f484ec50036b28d7659851a8202dbab32
+size 93104

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/quoraretrieval.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:697e06974d23b23f9b1629f9cbe0a4b74d8e0a59195e59e1cc2d76a44324ad5f
+size 95320

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/scidocs.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb2e21a7d9c6248eb4282e7f6f42aa007b6690b94f5a8ef33362079d345433c4
+size 147024

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/scifact.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baa1d51d527da2108c1eb5975bc3f3b816a679ba0ec255f6d570e61ab565e342
+size 109944

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-en/touche2020.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0648c7962ee02fbb2759e52326044353b201999a5db88bd3d513335839b719ac
+size 126052

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/arguana.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25bb5980f0e8f9168c4b25e4bfe3039ba52435972799201735df870b811b063b
+size 143016

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/climatefever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ad938fdda72ef4ef2bf5acee52b236a08254514969f39586cb800347d759ca1
+size 138268

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/dbpedia.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40142082147d2e445903a98ec7e05a52b2ca54090c6b01631c6ce392b794684f
+size 151880

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/fever.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90bf1936d1094c26f0954336f00a1a1bb97f74028597cdf624596c13b2533897
+size 145216

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/fiqa2018.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:308e1628705a525471e7e4f348ddc642728c498187dba58adf35ff66c4199ff3
+size 111596

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/hotpotqa.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9bfe5ba3e06b2661275520dd068c8625ba7d5f6fadd891f85b7c3f27ab40180
+size 114052

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/msmarco.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:507ffe4a8a3a680c57e6bd212707ae9a7db715aeede6a71279561088cae636ec
+size 91776

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/nfcorpus.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:991ee08a2b5ba3d4c16f959a8f784f02ff32249032b5f7ade438383c044735fa
+size 100512

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/nq.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb6d5b8b762311a6a2226c91cb906fc683c365987d386d292ea2bf5090a27916
+size 105844

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/quoraretrieval.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f39730cba8eeb8be99b40e17cea8d7203971e86525b848844c1481cbb80f8fed
+size 102272

hakari-results/lightonai__DenseOn/hakari-bench__NanoBEIR-es/scidocs.json.xz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80403818877b70bd18ebe11b8587d40b8350c90a5e28389db49816bf5da6b39a
+size 160736