Upload folder using huggingface_hub

README.md

@@ -0,0 +1,254 @@
+---
+license: apache-2.0
+task_categories:
+  - text-classification
+language:
+  - multilingual
+  - ar
+  - en
+  - fr
+  - es
+  - de
+tags:
+  - language-identification
+  - lid
+  - benchmark
+  - evaluation
+  - arabic-dialects
+  - multilingual
+pretty_name: LID Benchmark — Language Identification Evaluation Results
+size_categories:
+  - 10K<n<100K
+configs:
+  - config_name: results_per_language
+    data_files:
+      - split: train
+        path: data/results_per_language/train.parquet
+  - config_name: results_aggregate
+    data_files:
+      - split: train
+        path: data/results_aggregate/train.parquet
+  - config_name: results_summary
+    data_files:
+      - split: train
+        path: data/results_summary/train.parquet
+---
+
+# LID Benchmark — Language Identification Evaluation Results
+
+Structured evaluation results for **10 language identification models** across **8 benchmarks** covering **380 languages** — with per-language accuracy, aggregate metrics, and confusion analysis.
+
+Built as part of the [Gherbal](https://www.omneitylabs.com/models/gherbal) evaluation pipeline.
+
+## Quick Start
+
+```python
+from datasets import load_dataset
+
+# Per-language results (26,540 rows)
+per_lang = load_dataset("omneity-labs/lid-benchmark", "results_per_language", split="train")
+
+# Aggregate metrics per model × benchmark × scope (400 rows)
+aggregate = load_dataset("omneity-labs/lid-benchmark", "results_aggregate", split="train")
+
+# Summary — one row per model × benchmark, full scope only (80 rows)
+summary = load_dataset("omneity-labs/lid-benchmark", "results_summary", split="train")
+```
+
+## Leaderboard (Full Scope)
+
+| Model | FLORES+ devtest | MADAR | Gherbal-Multi | ATLASIA-LID |
+|-------|:-:|:-:|:-:|:-:|
+| **GlotLID** | **0.9253** | 0.5648 | 0.7772 | 0.4977 |
+| OpenLID v2 | 0.8748 | 0.6262 | 0.7762 | 0.5735 |
+| OpenLID v3 | 0.8556 | — | 0.6619 | — |
+| **Gherbal v4** | 0.8500 | **0.6298** | **0.8699** | **0.6909** |
+| OpenLID v1 | 0.8425 | 0.5587 | 0.8296 | 0.4845 |
+| NLLB-LID | 0.8331 | 0.1052 | 0.7522 | 0.3348 |
+| FastLID-176 | 0.4006 | 0.1352 | 0.6472 | 0.3899 |
+| Gherbal v3 | 0.3605 | 0.5745 | 0.8966 | 0.6561 |
+| Gherbal v2 | 0.1495 | 0.5811 | 0.7961 | 0.6561 |
+| Gherbal v1 | 0.1374 | 0.2771 | 0.8385 | 0.2718 |
+
+> **Note**: Full-scope FLORES+ accuracy penalizes models that support fewer languages (unsupported languages count as errors). Use `results_aggregate` with `scope=v4` (214 languages) for a fairer narrower comparison. Gherbal v4 achieves **0.9312** accuracy on FLORES+ devtest in the v4 scope.
+
+## Dataset Configs
+
+### `results_per_language` — Per-Language Breakdown
+
+26,540 rows. One row per (model, benchmark, scope, language).
+
+| Column | Type | Description |
+|--------|------|-------------|
+| `model` | string | Model name (e.g. `gherbal-v4`, `glotlid`) |
+| `benchmark` | string | Benchmark name (e.g. `flores-devtest`) |
+| `scope` | string | Language scope: `full`, `v1`, `v2`, `v3`, or `v4` |
+| `language` | string | Language code in `iso639-3_Script` format (e.g. `arb_Arab`) |
+| `n_samples` | int | Number of test samples for this language |
+| `accuracy` | float | Classification accuracy (0–1) |
+| `top_confusion_1` | string | Most confused-with language |
+| `top_confusion_1_count` | int | Count of samples misclassified as this language |
+| `top_confusion_2` | string | 2nd most confused-with language |
+| `top_confusion_2_count` | int | — |
+| `top_confusion_3` | string | 3rd most confused-with language |
+| `top_confusion_3_count` | int | — |
+| `confusions_json` | string | Full confusion map as JSON (all misclassified targets and counts) |
+
+**Example — find the hardest languages for a model:**
+
+```python
+from datasets import load_dataset
+import pandas as pd
+
+ds = load_dataset("omneity-labs/lid-benchmark", "results_per_language", split="train")
+df = ds.to_pandas()
+
+# Worst-performing languages for Gherbal v4 on FLORES
+worst = (
+    df[(df["model"] == "gherbal-v4") &
+       (df["benchmark"] == "flores-devtest") &
+       (df["scope"] == "full") &
+       (df["n_samples"] >= 100)]
+    .sort_values("accuracy")
+    .head(10)
+    [["language", "accuracy", "n_samples", "top_confusion_1"]]
+)
+print(worst)
+```
+
+**Example — compare Arabic dialect accuracy across models:**
+
+```python
+arabic_dialects = [
+    "arz_Arab", "ary_Arab", "arq_Arab", "aeb_Arab",
+    "apc_Arab", "acm_Arab", "ars_Arab", "afb_Arab", 
+    # Add more
+]
+
+arabic_df = df[
+    (df["language"].isin(arabic_dialects)) &
+    (df["benchmark"] == "flores-devtest") &
+    (df["scope"] == "full")
+]
+
+pivot = arabic_df.pivot_table(
+    index="language", columns="model", values="accuracy"
+)
+print(pivot.round(3))
+```
+
+### `results_aggregate` — Aggregate Metrics
+
+400 rows. One row per (model, benchmark, scope).
+
+| Column | Type | Description |
+|--------|------|-------------|
+| `model` | string | Model name |
+| `benchmark` | string | Benchmark name |
+| `scope` | string | Language scope |
+| `accuracy` | float | Overall accuracy |
+| `f1_macro` | float | Macro-averaged F1 |
+| `f1_weighted` | float | Weighted F1 |
+| `precision_macro` | float | Macro-averaged precision |
+| `recall_macro` | float | Macro-averaged recall |
+| `n_samples` | int | Total evaluation samples |
+| `n_classes` | int | Number of unique languages |
+
+**Example — model comparison across scopes:**
+
+```python
+ds = load_dataset("omneity-labs/lid-benchmark", "results_aggregate", split="train")
+df = ds.to_pandas()
+
+comparison = df[
+    (df["benchmark"] == "flores-devtest") &
+    (df["model"].isin(["gherbal-v4", "glotlid", "openlid-v2"]))
+].pivot_table(index="scope", columns="model", values="accuracy")
+print(comparison.round(4))
+```
+
+### `results_summary` — Quick Summary
+
+80 rows. One row per (model, benchmark) — full scope only. Best for quick leaderboard construction.
+
+| Column | Type | Description |
+|--------|------|-------------|
+| `model` | string | Model name |
+| `benchmark` | string | Benchmark name |
+| `accuracy` | float | Overall accuracy (full scope) |
+| `f1_macro` | float | Macro F1 (full scope) |
+| `f1_weighted` | float | Weighted F1 (full scope) |
+| `precision_macro` | float | Macro precision (full scope) |
+| `recall_macro` | float | Macro recall (full scope) |
+| `n_samples` | int | Total samples |
+| `n_classes` | int | Number of classes |
+
+## Models Evaluated
+
+| Model | Type | Languages | Source |
+|-------|------|-----------|--------|
+| [Gherbal v4](https://www.omneitylabs.com/models/gherbal) | FastText | 214 | Omneity Labs |
+| Gherbal v3 | FastText | 106 | Omneity Labs |
+| Gherbal v2 | FastText | 46 | Omneity Labs |
+| Gherbal v1 | FastText | 36 | Omneity Labs |
+| [GlotLID v3](https://huggingface.co/cis-lmu/glotlid) | FastText | 2,102 | LMU Munich |
+| [NLLB-LID](https://huggingface.co/facebook/fasttext-language-identification) | FastText | 218 | Meta |
+| [OpenLID v1](https://huggingface.co/laurievb/OpenLID) | FastText | 201 | Laurie Burchell |
+| [OpenLID v2](https://huggingface.co/laurievb/OpenLID) | FastText | 201 | Laurie Burchell |
+| [OpenLID v3](https://huggingface.co/laurievb/OpenLID) | FastText | 201 | Laurie Burchell |
+| [FastLID-176](https://fasttext.cc/docs/en/language-identification.html) | FastText | 176 | Meta |
+
+## Benchmarks
+
+| Benchmark | Samples | Languages | Description |
+|-----------|---------|-----------|-------------|
+| **FLORES+ devtest** | 222,640 | 214 | [openlanguagedata/flores_plus](https://huggingface.co/datasets/openlanguagedata/flores_plus) devtest split |
+| **FLORES+ dev** | 224,325 | 220 | [openlanguagedata/flores_plus](https://huggingface.co/datasets/openlanguagedata/flores_plus) dev split |
+| **MADAR** | 5,600 | 15 | [sawalni-ai/madar](https://huggingface.co/datasets/sawalni-ai/madar) — Arabic dialect corpus |
+| **Gherbal-Multi** | 185,000+ | 106+ | [sawalni-ai/gherbal-multi](https://huggingface.co/datasets/sawalni-ai/gherbal-multi) — multi-source test set |
+| **ATLASIA-LID** | 234,000+ | 24 | [atlasia/Arabic-LID-Leaderboard](https://huggingface.co/datasets/atlasia/Arabic-LID-Leaderboard) — Arabic country-level dialects |
+| **WiLI-2018** | — | 235 | Wikipedia Language Identification |
+| **CommonLID** | — | — | Common Crawl language ID |
+| **Bouquet** | — | — | Cross-domain evaluation |
+
+## Evaluation Scopes
+
+Results include multiple **scopes** to enable fair comparison between models with different language coverage:
+
+| Scope | Languages | Description |
+|-------|-----------|-------------|
+| `full` | All | All languages in the benchmark (penalizes models with fewer supported languages) |
+| `v1` | 36 | Intersection with Gherbal v1 language set |
+| `v2` | 46 | Intersection with Gherbal v2 language set |
+| `v3` | 106 | Intersection with Gherbal v3 language set |
+| `v4` | 214 | Intersection with Gherbal v4 language set |
+
+Using scoped evaluation ensures models are compared only on languages they were designed to handle. For example, Gherbal v3 supports 106 languages — its `v3` scope accuracy on FLORES+ is much higher than its `full` scope accuracy, because the full scope includes 108+ languages it was never trained on.
+
+## Language Codes
+
+Languages use the `iso639-3_Script` format from FLORES+:
+- `arb_Arab` — Modern Standard Arabic (Arabic script)
+- `arz_Arab` — Egyptian Arabic
+- `ary_Arab` — Moroccan Arabic (Arabic script)
+- `ary_Latn` — Moroccan Arabic (Latin script)
+- `eng_Latn` — English
+- `fra_Latn` — French
+
+Full list of 380 languages available in the `results_per_language` config.
+
+## CSV Downloads
+
+For convenience, CSV versions of all three configs are also included in the `csv/` directory.
+
+## Citation
+
+If you use this benchmark data in your research, please reference:
+
+- **Omneity Labs LID Benchmark**: [https://huggingface.co/datasets/omneity-labs/lid-benchmark]
+- **Gherbal model**: [https://www.omneitylabs.com/models/gherbal](https://www.omneitylabs.com/models/gherbal)
+- **Evaluation benchmarks**: See individual benchmark datasets linked above.
+
+## License
+
+The evaluation results in this dataset are released under [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0). The underlying benchmark datasets retain their original licenses.

csv/results_aggregate.csv

@@ -0,0 +1,401 @@
+model,benchmark,scope,accuracy,f1_macro,f1_weighted,precision_macro,recall_macro,n_samples,n_classes
+gherbal-v2,flores-devtest,full,0.1495,0.0588,0.0616,0.0442,0.1452,222640,214
+gherbal-v2,flores-devtest,v1,0.9312,0.6912,0.9352,0.7094,0.6883,34408,34
+gherbal-v2,flores-devtest,v2,0.889,0.7089,0.8952,0.7361,0.7073,37444,36
+gherbal-v2,flores-devtest,v3,0.3614,0.2188,0.245,0.1798,0.3254,92092,90
+gherbal-v2,flores-devtest,v4,0.162,0.0662,0.0697,0.05,0.1564,205436,198
+gherbal-v2,flores-dev,full,0.147,0.0571,0.0602,0.0429,0.1419,224325,220
+gherbal-v2,flores-dev,v1,0.9303,0.7056,0.9338,0.7244,0.7029,33898,34
+gherbal-v2,flores-dev,v2,0.8939,0.7267,0.8998,0.7524,0.7254,36889,36
+gherbal-v2,flores-dev,v3,0.3634,0.2195,0.2464,0.1802,0.3264,90727,90
+gherbal-v2,flores-dev,v4,0.1637,0.0667,0.0709,0.0504,0.1569,201394,198
+gherbal-v2,madar,full,0.5811,0.1944,0.5625,0.1991,0.2046,5600,15
+gherbal-v2,madar,v1,0.7487,0.1219,0.8069,0.135,0.1146,2077,3
+gherbal-v2,madar,v2,0.6411,0.2388,0.6449,0.2555,0.2402,5076,11
+gherbal-v2,madar,v3,0.6051,0.214,0.5954,0.2231,0.221,5378,13
+gherbal-v2,madar,v4,0.6051,0.214,0.5954,0.2231,0.221,5378,13
+gherbal-v2,gherbal-multi,full,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,gherbal-multi,v1,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,gherbal-multi,v2,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,gherbal-multi,v3,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,gherbal-multi,v4,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,atlasia-lid,full,0.6561,0.1481,0.6199,0.152,0.1612,234327,15
+gherbal-v2,atlasia-lid,v1,0.9532,0.1099,0.9675,0.1156,0.1052,117533,3
+gherbal-v2,atlasia-lid,v2,0.7099,0.1683,0.6888,0.1795,0.1746,216563,13
+gherbal-v2,atlasia-lid,v3,0.6561,0.1481,0.6199,0.152,0.1612,234327,15
+gherbal-v2,atlasia-lid,v4,0.6561,0.1481,0.6199,0.152,0.1612,234327,15
+gherbal-v2,wili-2018,full,0.2374,0.1173,0.1296,0.0909,0.2149,62000,124
+gherbal-v2,wili-2018,v1,0.8921,0.6493,0.8854,0.668,0.6542,16500,33
+gherbal-v2,wili-2018,v2,0.8921,0.6493,0.8854,0.668,0.6542,16500,33
+gherbal-v2,wili-2018,v3,0.4673,0.288,0.3474,0.249,0.3874,31500,63
+gherbal-v2,wili-2018,v4,0.2374,0.1173,0.1296,0.0909,0.2149,62000,124
+gherbal-v2,commonlid,full,0.5934,0.1506,0.5493,0.1377,0.2193,373230,101
+gherbal-v2,commonlid,v1,0.8214,0.4535,0.8319,0.4579,0.5276,269625,31
+gherbal-v2,commonlid,v2,0.8213,0.4609,0.8318,0.4651,0.5436,269667,33
+gherbal-v2,commonlid,v3,0.6819,0.3297,0.6563,0.3168,0.4311,324781,45
+gherbal-v2,commonlid,v4,0.6158,0.1972,0.5777,0.1829,0.2778,359646,77
+gherbal-v2,bouquet,full,0.0986,0.0297,0.0313,0.0209,0.0935,289300,275
+gherbal-v2,bouquet,v1,0.8887,0.618,0.9064,0.6455,0.6059,31560,30
+gherbal-v2,bouquet,v2,0.8743,0.6304,0.8948,0.661,0.616,32612,31
+gherbal-v2,bouquet,v3,0.4302,0.266,0.3293,0.2295,0.3475,66276,63
+gherbal-v2,bouquet,v4,0.1869,0.084,0.0927,0.0643,0.1694,152540,145
+gherbal-v3,flores-devtest,full,0.3605,0.2412,0.2548,0.216,0.3427,222640,214
+gherbal-v3,flores-devtest,v1,0.9596,0.4516,0.9695,0.4583,0.447,34408,34
+gherbal-v3,flores-devtest,v2,0.9245,0.4648,0.9346,0.4745,0.4618,37444,36
+gherbal-v3,flores-devtest,v3,0.8716,0.739,0.8675,0.773,0.7435,92092,90
+gherbal-v3,flores-devtest,v4,0.3907,0.2679,0.2853,0.2429,0.3683,205436,198
+gherbal-v3,flores-dev,full,0.3581,0.2366,0.2514,0.2075,0.3384,224325,220
+gherbal-v3,flores-dev,v1,0.9719,0.4824,0.9789,0.4866,0.4789,33898,34
+gherbal-v3,flores-dev,v2,0.9526,0.5027,0.9595,0.5075,0.5005,36889,36
+gherbal-v3,flores-dev,v3,0.8854,0.7493,0.8813,0.7758,0.7533,90727,90
+gherbal-v3,flores-dev,v4,0.3989,0.273,0.2929,0.2443,0.3732,201394,198
+gherbal-v3,madar,full,0.5745,0.2402,0.5518,0.272,0.2442,5600,15
+gherbal-v3,madar,v1,0.7848,0.1387,0.8243,0.1506,0.1338,2077,3
+gherbal-v3,madar,v2,0.6328,0.2745,0.6315,0.3155,0.2687,5076,11
+gherbal-v3,madar,v3,0.5982,0.2704,0.5845,0.3122,0.2699,5378,13
+gherbal-v3,madar,v4,0.5982,0.2704,0.5845,0.3122,0.2699,5378,13
+gherbal-v3,gherbal-multi,full,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,gherbal-multi,v1,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,gherbal-multi,v2,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,gherbal-multi,v3,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,gherbal-multi,v4,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,atlasia-lid,full,0.6561,0.108,0.6252,0.1132,0.1282,234327,15
+gherbal-v3,atlasia-lid,v1,0.937,0.0829,0.9505,0.0837,0.0821,117533,3
+gherbal-v3,atlasia-lid,v2,0.7098,0.1287,0.6939,0.1362,0.1456,216563,13
+gherbal-v3,atlasia-lid,v3,0.6561,0.108,0.6252,0.1132,0.1282,234327,15
+gherbal-v3,atlasia-lid,v4,0.6561,0.108,0.6252,0.1132,0.1282,234327,15
+gherbal-v3,wili-2018,full,0.4695,0.2834,0.3771,0.2544,0.3529,62000,124
+gherbal-v3,wili-2018,v1,0.9209,0.4191,0.9399,0.4316,0.4107,16500,33
+gherbal-v3,wili-2018,v2,0.9209,0.4191,0.9399,0.4316,0.4107,16500,33
+gherbal-v3,wili-2018,v3,0.9242,0.6842,0.934,0.7014,0.677,31500,63
+gherbal-v3,wili-2018,v4,0.4695,0.2834,0.3771,0.2544,0.3529,62000,124
+gherbal-v3,commonlid,full,0.7441,0.1718,0.7457,0.1667,0.2198,373230,101
+gherbal-v3,commonlid,v1,0.8627,0.2552,0.8926,0.2564,0.2863,269625,31
+gherbal-v3,commonlid,v2,0.8626,0.2584,0.8925,0.2593,0.2937,269667,33
+gherbal-v3,commonlid,v3,0.8551,0.2972,0.8828,0.3046,0.3426,324781,45
+gherbal-v3,commonlid,v4,0.7722,0.2075,0.7808,0.204,0.2589,359646,77
+gherbal-v3,bouquet,full,0.1914,0.0939,0.1086,0.083,0.1655,289300,275
+gherbal-v3,bouquet,v1,0.9479,0.3837,0.9593,0.3907,0.3792,31560,30
+gherbal-v3,bouquet,v2,0.9343,0.3921,0.9485,0.4008,0.3862,32612,31
+gherbal-v3,bouquet,v3,0.8356,0.5098,0.8497,0.5592,0.5014,66276,63
+gherbal-v3,bouquet,v4,0.3631,0.2106,0.2731,0.2004,0.28,152540,145
+gherbal-v4,flores-devtest,full,0.85,0.7693,0.8245,0.7712,0.7943,222640,214
+gherbal-v4,flores-devtest,v1,0.9591,0.3465,0.9682,0.3519,0.3432,34408,34
+gherbal-v4,flores-devtest,v2,0.9213,0.3487,0.9309,0.357,0.3466,37444,36
+gherbal-v4,flores-devtest,v3,0.8914,0.4411,0.8936,0.4612,0.4407,92092,90
+gherbal-v4,flores-devtest,v4,0.9212,0.8505,0.9187,0.8686,0.8537,205436,198
+gherbal-v4,flores-dev,full,0.8334,0.7485,0.801,0.745,0.7798,224325,220
+gherbal-v4,flores-dev,v1,0.9654,0.3558,0.9732,0.3601,0.3529,33898,34
+gherbal-v4,flores-dev,v2,0.9344,0.3559,0.9423,0.3619,0.3543,36889,36
+gherbal-v4,flores-dev,v3,0.9007,0.4625,0.9019,0.4816,0.4625,90727,90
+gherbal-v4,flores-dev,v4,0.9282,0.8565,0.9252,0.873,0.86,201394,198
+gherbal-v4,madar,full,0.6298,0.2608,0.6169,0.316,0.2712,5600,15
+gherbal-v4,madar,v1,0.8411,0.1574,0.8898,0.1672,0.1495,2077,3
+gherbal-v4,madar,v2,0.6629,0.2595,0.6682,0.3354,0.2465,5076,11
+gherbal-v4,madar,v3,0.6558,0.2953,0.6516,0.3682,0.2984,5378,13
+gherbal-v4,madar,v4,0.6558,0.2953,0.6516,0.3682,0.2984,5378,13
+gherbal-v4,gherbal-multi,full,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,gherbal-multi,v1,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,gherbal-multi,v2,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,gherbal-multi,v3,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,gherbal-multi,v4,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,atlasia-lid,full,0.6909,0.109,0.6653,0.1283,0.114,234327,15
+gherbal-v4,atlasia-lid,v1,0.9567,0.0716,0.9687,0.0727,0.0706,117533,3
+gherbal-v4,atlasia-lid,v2,0.7266,0.1154,0.711,0.1335,0.1199,216563,13
+gherbal-v4,atlasia-lid,v3,0.6909,0.109,0.6653,0.1283,0.114,234327,15
+gherbal-v4,atlasia-lid,v4,0.6909,0.109,0.6653,0.1283,0.114,234327,15
+gherbal-v4,wili-2018,full,0.9153,0.6489,0.921,0.6729,0.6448,62000,124
+gherbal-v4,wili-2018,v1,0.9016,0.2944,0.9189,0.3099,0.2889,16500,33
+gherbal-v4,wili-2018,v2,0.9016,0.2944,0.9189,0.3099,0.2889,16500,33
+gherbal-v4,wili-2018,v3,0.9084,0.4457,0.9196,0.4683,0.4402,31500,63
+gherbal-v4,wili-2018,v4,0.9153,0.6489,0.921,0.6729,0.6448,62000,124
+gherbal-v4,commonlid,full,0.798,0.2286,0.8291,0.2276,0.2604,373230,101
+gherbal-v4,commonlid,v1,0.8181,0.118,0.8622,0.1219,0.1325,269625,31
+gherbal-v4,commonlid,v2,0.8181,0.1191,0.8621,0.1228,0.1355,269667,33
+gherbal-v4,commonlid,v3,0.8209,0.1453,0.8599,0.1503,0.1671,324781,45
+gherbal-v4,commonlid,v4,0.8281,0.2592,0.8643,0.261,0.2907,359646,77
+gherbal-v4,bouquet,full,0.4628,0.3074,0.3845,0.2855,0.3699,289300,275
+gherbal-v4,bouquet,v1,0.9403,0.1913,0.9567,0.196,0.1881,31560,30
+gherbal-v4,bouquet,v2,0.9267,0.1942,0.9459,0.1998,0.1903,32612,31
+gherbal-v4,bouquet,v3,0.8622,0.2857,0.8843,0.3088,0.2786,66276,63
+gherbal-v4,bouquet,v4,0.8776,0.6022,0.8888,0.6304,0.5947,152540,145
+nllb-lid,flores-devtest,full,0.8331,0.7481,0.8088,0.746,0.7715,222640,214
+nllb-lid,flores-devtest,v1,0.9346,0.6145,0.9217,0.6125,0.6231,34408,34
+nllb-lid,flores-devtest,v2,0.8588,0.5872,0.8412,0.5863,0.5996,37444,36
+nllb-lid,flores-devtest,v3,0.8565,0.5589,0.8538,0.5674,0.5607,92092,90
+nllb-lid,flores-devtest,v4,0.8885,0.8144,0.8846,0.8252,0.8187,205436,198
+nllb-lid,flores-dev,full,0.8129,0.708,0.7853,0.7046,0.7338,224325,220
+nllb-lid,flores-dev,v1,0.9336,0.614,0.921,0.6123,0.6224,33898,34
+nllb-lid,flores-dev,v2,0.8579,0.5868,0.8405,0.586,0.5989,36889,36
+nllb-lid,flores-dev,v3,0.8549,0.5276,0.8523,0.5361,0.5292,90727,90
+nllb-lid,flores-dev,v4,0.8864,0.8061,0.8826,0.8179,0.8104,201394,198
+nllb-lid,madar,full,0.1052,0.0087,0.0202,0.0048,0.0454,5600,15
+nllb-lid,madar,v1,0.2836,0.0738,0.1258,0.0474,0.1664,2077,3
+nllb-lid,madar,v2,0.116,0.0116,0.0243,0.0065,0.0555,5076,11
+nllb-lid,madar,v3,0.1095,0.0099,0.0218,0.0055,0.0499,5378,13
+nllb-lid,madar,v4,0.1095,0.0099,0.0218,0.0055,0.0499,5378,13
+nllb-lid,gherbal-multi,full,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,gherbal-multi,v1,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,gherbal-multi,v2,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,gherbal-multi,v3,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,gherbal-multi,v4,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,atlasia-lid,full,0.3348,0.0091,0.1686,0.0061,0.0182,234327,15
+nllb-lid,atlasia-lid,v1,0.6675,0.0422,0.5349,0.0352,0.0526,117533,3
+nllb-lid,atlasia-lid,v2,0.3623,0.0103,0.1933,0.007,0.0192,216563,13
+nllb-lid,atlasia-lid,v3,0.3348,0.0091,0.1686,0.0061,0.0182,234327,15
+nllb-lid,atlasia-lid,v4,0.3348,0.0091,0.1686,0.0061,0.0182,234327,15
+nllb-lid,wili-2018,full,0.927,0.7721,0.9278,0.7828,0.7715,62000,124
+nllb-lid,wili-2018,v1,0.9103,0.556,0.9099,0.5713,0.5563,16500,33
+nllb-lid,wili-2018,v2,0.9103,0.556,0.9099,0.5713,0.5563,16500,33
+nllb-lid,wili-2018,v3,0.9123,0.5786,0.9093,0.5887,0.5806,31500,63
+nllb-lid,wili-2018,v4,0.927,0.7721,0.9278,0.7828,0.7715,62000,124
+nllb-lid,commonlid,full,0.8609,0.2541,0.8711,0.2561,0.2757,373230,101
+nllb-lid,commonlid,v1,0.8861,0.1381,0.905,0.1427,0.1524,269625,31
+nllb-lid,commonlid,v2,0.886,0.1365,0.9048,0.141,0.1507,269667,33
+nllb-lid,commonlid,v3,0.8892,0.1755,0.9077,0.1834,0.1879,324781,45
+nllb-lid,commonlid,v4,0.8935,0.2945,0.9102,0.3011,0.3124,359646,77
+nllb-lid,bouquet,full,0.5023,0.3398,0.4214,0.3145,0.4051,289300,275
+nllb-lid,bouquet,v1,0.9281,0.2361,0.9364,0.2389,0.234,31560,30
+nllb-lid,bouquet,v2,0.8982,0.2322,0.9062,0.2349,0.2301,32612,31
+nllb-lid,bouquet,v3,0.8758,0.3128,0.8938,0.3239,0.3065,66276,63
+nllb-lid,bouquet,v4,0.893,0.6206,0.8987,0.6347,0.6166,152540,145
+openlid-v1,flores-devtest,full,0.8425,0.7907,0.8167,0.7913,0.8166,222640,214
+openlid-v1,flores-devtest,v1,0.9679,0.6118,0.9718,0.6154,0.6094,34408,34
+openlid-v1,flores-devtest,v2,0.9024,0.6111,0.8981,0.6308,0.6159,37444,36
+openlid-v1,flores-devtest,v3,0.8837,0.6766,0.8798,0.7007,0.6804,92092,90
+openlid-v1,flores-devtest,v4,0.9091,0.8714,0.9048,0.8872,0.8762,205436,198
+openlid-v1,flores-dev,full,0.8187,0.7628,0.7871,0.7553,0.7944,224325,220
+openlid-v1,flores-dev,v1,0.9677,0.5417,0.9719,0.5447,0.5394,33898,34
+openlid-v1,flores-dev,v2,0.9032,0.5246,0.8998,0.5403,0.5283,36889,36
+openlid-v1,flores-dev,v3,0.8826,0.6277,0.8786,0.6435,0.6313,90727,90
+openlid-v1,flores-dev,v4,0.908,0.8706,0.9036,0.8822,0.8754,201394,198
+openlid-v1,madar,full,0.5587,0.205,0.5456,0.207,0.2425,5600,15
+openlid-v1,madar,v1,0.7925,0.1929,0.8508,0.2233,0.1778,2077,3
+openlid-v1,madar,v2,0.569,0.1882,0.5764,0.1998,0.1874,5076,11
+openlid-v1,madar,v3,0.5818,0.2348,0.5766,0.2431,0.2667,5378,13
+openlid-v1,madar,v4,0.5818,0.2348,0.5766,0.2431,0.2667,5378,13
+openlid-v1,gherbal-multi,full,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,gherbal-multi,v1,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,gherbal-multi,v2,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,gherbal-multi,v3,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,gherbal-multi,v4,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,atlasia-lid,full,0.4845,0.0591,0.468,0.0612,0.0768,234327,15
+openlid-v1,atlasia-lid,v1,0.7238,0.0758,0.7476,0.0745,0.0983,117533,3
+openlid-v1,atlasia-lid,v2,0.4823,0.0489,0.4819,0.0482,0.0641,216563,13
+openlid-v1,atlasia-lid,v3,0.4845,0.0591,0.468,0.0612,0.0768,234327,15
+openlid-v1,atlasia-lid,v4,0.4845,0.0591,0.468,0.0612,0.0768,234327,15
+openlid-v1,wili-2018,full,0.943,0.7669,0.9462,0.7743,0.7643,62000,124
+openlid-v1,wili-2018,v1,0.9309,0.3912,0.9484,0.4025,0.384,16500,33
+openlid-v1,wili-2018,v2,0.9309,0.3912,0.9484,0.4025,0.384,16500,33
+openlid-v1,wili-2018,v3,0.9387,0.56,0.9423,0.5671,0.5579,31500,63
+openlid-v1,wili-2018,v4,0.943,0.7669,0.9462,0.7743,0.7643,62000,124
+openlid-v1,commonlid,full,0.8405,0.2561,0.8668,0.2474,0.3004,373230,101
+openlid-v1,commonlid,v1,0.8509,0.1339,0.8935,0.1352,0.1573,269625,31
+openlid-v1,commonlid,v2,0.8508,0.1332,0.8933,0.1345,0.1575,269667,33
+openlid-v1,commonlid,v3,0.8647,0.1738,0.9014,0.1734,0.2019,324781,45
+openlid-v1,commonlid,v4,0.8722,0.297,0.9043,0.2907,0.3405,359646,77
+openlid-v1,bouquet,full,0.4856,0.3299,0.3983,0.3023,0.4023,289300,275
+openlid-v1,bouquet,v1,0.9683,0.264,0.9769,0.267,0.2617,31560,30
+openlid-v1,bouquet,v2,0.9423,0.2605,0.9497,0.2673,0.2585,32612,31
+openlid-v1,bouquet,v3,0.9107,0.345,0.9201,0.3581,0.3415,66276,63
+openlid-v1,bouquet,v4,0.9095,0.6572,0.9155,0.6736,0.6529,152540,145
+openlid-v2,flores-devtest,full,0.8748,0.8507,0.8471,0.8509,0.8798,222640,214
+openlid-v2,flores-devtest,v1,0.9738,0.6155,0.9775,0.6181,0.6131,34408,34
+openlid-v2,flores-devtest,v2,0.9176,0.6208,0.9167,0.6343,0.6241,37444,36
+openlid-v2,flores-devtest,v3,0.9096,0.6902,0.9068,0.7121,0.6935,92092,90
+openlid-v2,flores-devtest,v4,0.9394,0.9275,0.9351,0.9436,0.9326,205436,198
+openlid-v2,flores-dev,full,0.8508,0.8203,0.8169,0.8157,0.8556,224325,220
+openlid-v2,flores-dev,v1,0.9725,0.5629,0.9768,0.5656,0.5604,33898,34
+openlid-v2,flores-dev,v2,0.9183,0.5689,0.9186,0.5807,0.5713,36889,36
+openlid-v2,flores-dev,v3,0.9085,0.6217,0.9061,0.6404,0.6242,90727,90
+openlid-v2,flores-dev,v4,0.9389,0.9273,0.9347,0.9417,0.9323,201394,198
+openlid-v2,madar,full,0.6262,0.2403,0.6004,0.2358,0.289,5600,15
+openlid-v2,madar,v1,0.7882,0.2276,0.8482,0.2624,0.2095,2077,3
+openlid-v2,madar,v2,0.6395,0.2267,0.6368,0.2334,0.2269,5076,11
+openlid-v2,madar,v3,0.6521,0.2762,0.6326,0.2745,0.3211,5378,13
+openlid-v2,madar,v4,0.6521,0.2762,0.6326,0.2745,0.3211,5378,13
+openlid-v2,gherbal-multi,full,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,gherbal-multi,v1,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,gherbal-multi,v2,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,gherbal-multi,v3,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,gherbal-multi,v4,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,atlasia-lid,full,0.5735,0.055,0.5322,0.0525,0.0724,234327,15
+openlid-v2,atlasia-lid,v1,0.7945,0.078,0.8207,0.0752,0.0983,117533,3
+openlid-v2,atlasia-lid,v2,0.5759,0.0494,0.5542,0.0451,0.0644,216563,13
+openlid-v2,atlasia-lid,v3,0.5735,0.055,0.5322,0.0525,0.0724,234327,15
+openlid-v2,atlasia-lid,v4,0.5735,0.055,0.5322,0.0525,0.0724,234327,15
+openlid-v2,wili-2018,full,0.9368,0.7881,0.9407,0.7954,0.7849,62000,124
+openlid-v2,wili-2018,v1,0.9238,0.3711,0.9333,0.3762,0.3673,16500,33
+openlid-v2,wili-2018,v2,0.9238,0.3711,0.9333,0.3762,0.3673,16500,33
+openlid-v2,wili-2018,v3,0.9186,0.5314,0.9195,0.5345,0.5309,31500,63
+openlid-v2,wili-2018,v4,0.9368,0.7881,0.9407,0.7954,0.7849,62000,124
+openlid-v2,commonlid,full,0.8221,0.2588,0.8466,0.2509,0.2962,373230,101
+openlid-v2,commonlid,v1,0.8331,0.1303,0.8699,0.1308,0.1476,269625,31
+openlid-v2,commonlid,v2,0.833,0.1297,0.8698,0.1301,0.1486,269667,33
+openlid-v2,commonlid,v3,0.8422,0.1682,0.875,0.1677,0.1905,324781,45
+openlid-v2,commonlid,v4,0.8531,0.2979,0.8818,0.2914,0.3342,359646,77
+openlid-v2,bouquet,full,0.4941,0.3344,0.4001,0.305,0.413,289300,275
+openlid-v2,bouquet,v1,0.9563,0.2326,0.9691,0.2365,0.2295,31560,30
+openlid-v2,bouquet,v2,0.9414,0.237,0.9557,0.2419,0.2335,32612,31
+openlid-v2,bouquet,v3,0.9075,0.3396,0.9218,0.3525,0.3343,66276,63
+openlid-v2,bouquet,v4,0.9257,0.6834,0.9332,0.6998,0.6779,152540,145
+fastlid-176,flores-devtest,full,0.4006,0.2516,0.3302,0.2426,0.3059,222640,214
+fastlid-176,flores-devtest,v1,0.745,0.2195,0.7038,0.2285,0.2324,34408,34
+fastlid-176,flores-devtest,v2,0.6846,0.2136,0.6409,0.2228,0.2282,37444,36
+fastlid-176,flores-devtest,v3,0.4226,0.1761,0.356,0.1765,0.209,92092,90
+fastlid-176,flores-devtest,v4,0.4341,0.2804,0.366,0.2738,0.3333,205436,198
+fastlid-176,flores-dev,full,0.3866,0.243,0.3166,0.2332,0.2979,224325,220
+fastlid-176,flores-dev,v1,0.7425,0.209,0.7007,0.2169,0.2214,33898,34
+fastlid-176,flores-dev,v2,0.6823,0.2035,0.6381,0.2095,0.2176,36889,36
+fastlid-176,flores-dev,v3,0.4214,0.1736,0.3549,0.1744,0.2062,90727,90
+fastlid-176,flores-dev,v4,0.4306,0.2772,0.3629,0.2709,0.3302,201394,198
+fastlid-176,madar,full,0.1352,0.0228,0.0637,0.028,0.055,5600,15
+fastlid-176,madar,v1,0.3645,0.1172,0.2667,0.1695,0.1728,2077,3
+fastlid-176,madar,v2,0.1491,0.0293,0.0735,0.037,0.0672,5076,11
+fastlid-176,madar,v3,0.1408,0.0256,0.0677,0.032,0.0605,5378,13
+fastlid-176,madar,v4,0.1408,0.0256,0.0677,0.032,0.0605,5378,13
+fastlid-176,gherbal-multi,full,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,gherbal-multi,v1,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,gherbal-multi,v2,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,gherbal-multi,v3,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,gherbal-multi,v4,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,atlasia-lid,full,0.3899,0.0174,0.2532,0.0187,0.0233,234327,15
+fastlid-176,atlasia-lid,v1,0.7773,0.0766,0.7352,0.0968,0.0752,117533,3
+fastlid-176,atlasia-lid,v2,0.4218,0.0187,0.2866,0.0202,0.0242,216563,13
+fastlid-176,atlasia-lid,v3,0.3899,0.0174,0.2532,0.0187,0.0233,234327,15
+fastlid-176,atlasia-lid,v4,0.3899,0.0174,0.2532,0.0187,0.0233,234327,15
+fastlid-176,wili-2018,full,0.7152,0.5118,0.6769,0.5197,0.5408,62000,124
+fastlid-176,wili-2018,v1,0.8084,0.3042,0.7928,0.3127,0.3102,16500,33
+fastlid-176,wili-2018,v2,0.8084,0.3042,0.7928,0.3127,0.3102,16500,33
+fastlid-176,wili-2018,v3,0.6784,0.3282,0.6355,0.3348,0.3503,31500,63
+fastlid-176,wili-2018,v4,0.7152,0.5118,0.6769,0.5197,0.5408,62000,124
+fastlid-176,commonlid,full,0.7702,0.1947,0.7424,0.2009,0.2226,373230,101
+fastlid-176,commonlid,v1,0.817,0.1345,0.7914,0.1387,0.1528,269625,31
+fastlid-176,commonlid,v2,0.8169,0.1328,0.7912,0.1368,0.1508,269667,33
+fastlid-176,commonlid,v3,0.8032,0.1454,0.783,0.1516,0.1665,324781,45
+fastlid-176,commonlid,v4,0.7987,0.2163,0.7766,0.2215,0.2443,359646,77
+fastlid-176,bouquet,full,0.2724,0.1558,0.2034,0.143,0.2087,289300,275
+fastlid-176,bouquet,v1,0.7172,0.1594,0.6854,0.1692,0.1668,31560,30
+fastlid-176,bouquet,v2,0.694,0.1581,0.6631,0.1653,0.1655,32612,31
+fastlid-176,bouquet,v3,0.5278,0.1729,0.4802,0.1793,0.19,66276,63
+fastlid-176,bouquet,v4,0.5166,0.2998,0.4549,0.2942,0.3405,152540,145
+glotlid,flores-devtest,full,0.9253,0.5562,0.9259,0.5635,0.5566,222640,214
+glotlid,flores-devtest,v1,0.9468,0.6461,0.9502,0.6497,0.6438,34408,34
+glotlid,flores-devtest,v2,0.8957,0.6569,0.8972,0.6701,0.6585,37444,36
+glotlid,flores-devtest,v3,0.91,0.5647,0.9098,0.5711,0.5657,92092,90
+glotlid,flores-devtest,v4,0.94,0.5723,0.9405,0.5796,0.5725,205436,198
+glotlid,flores-dev,full,0.931,0.5972,0.9317,0.6025,0.5976,224325,220
+glotlid,flores-dev,v1,0.9409,0.6085,0.9486,0.6149,0.6036,33898,34
+glotlid,flores-dev,v2,0.8993,0.6154,0.9097,0.6277,0.6111,36889,36
+glotlid,flores-dev,v3,0.9247,0.6591,0.9245,0.6626,0.6604,90727,90
+glotlid,flores-dev,v4,0.9477,0.6532,0.9481,0.6593,0.6534,201394,198
+glotlid,madar,full,0.5648,0.162,0.5626,0.1606,0.1912,5600,15
+glotlid,madar,v1,0.8353,0.2041,0.8852,0.2168,0.1931,2077,3
+glotlid,madar,v2,0.577,0.155,0.5977,0.1608,0.1544,5076,11
+glotlid,madar,v3,0.5881,0.1861,0.5931,0.1877,0.2125,5378,13
+glotlid,madar,v4,0.5881,0.1861,0.5931,0.1877,0.2125,5378,13
+glotlid,gherbal-multi,full,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,gherbal-multi,v1,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,gherbal-multi,v2,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,gherbal-multi,v3,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,gherbal-multi,v4,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,atlasia-lid,full,0.4977,0.0229,0.4963,0.0256,0.0289,234327,15
+glotlid,atlasia-lid,v1,0.7559,0.0351,0.8139,0.0364,0.0451,117533,3
+glotlid,atlasia-lid,v2,0.4964,0.0208,0.5132,0.0231,0.026,216563,13
+glotlid,atlasia-lid,v3,0.4977,0.0229,0.4963,0.0256,0.0289,234327,15
+glotlid,atlasia-lid,v4,0.4977,0.0229,0.4963,0.0256,0.0289,234327,15
+glotlid,wili-2018,full,0.9398,0.614,0.9457,0.6207,0.6101,62000,124
+glotlid,wili-2018,v1,0.9041,0.3572,0.9093,0.3605,0.3552,16500,33
+glotlid,wili-2018,v2,0.9041,0.3572,0.9093,0.3605,0.3552,16500,33
+glotlid,wili-2018,v3,0.922,0.4527,0.927,0.4566,0.4503,31500,63
+glotlid,wili-2018,v4,0.9398,0.614,0.9457,0.6207,0.6101,62000,124
+glotlid,commonlid,full,0.7888,0.0747,0.8046,0.0734,0.0824,373230,101
+glotlid,commonlid,v1,0.7858,0.0366,0.8079,0.0365,0.0413,269625,31
+glotlid,commonlid,v2,0.7857,0.0366,0.8077,0.0365,0.0416,269667,33
+glotlid,commonlid,v3,0.8039,0.0453,0.8242,0.0454,0.0502,324781,45
+glotlid,commonlid,v4,0.8186,0.0796,0.8364,0.0785,0.0872,359646,77
+glotlid,bouquet,full,0.6659,0.0998,0.659,0.105,0.1008,289300,275
+glotlid,bouquet,v1,0.9223,0.1732,0.9352,0.1768,0.1708,31560,30
+glotlid,bouquet,v2,0.9011,0.1644,0.9172,0.17,0.1615,32612,31
+glotlid,bouquet,v3,0.8944,0.1519,0.9137,0.158,0.1487,66276,63
+glotlid,bouquet,v4,0.9123,0.2513,0.9255,0.2595,0.2477,152540,145
+openlid-v3,flores-devtest,full,0.8556,0.8158,0.8336,0.8059,0.8379,222640,214
+openlid-v3,flores-devtest,v1,0.8777,0.524,0.8785,0.5246,0.5235,34408,34
+openlid-v3,flores-devtest,v2,0.8065,0.5063,0.8073,0.5068,0.5058,37444,36
+openlid-v3,flores-devtest,v3,0.8588,0.6285,0.8633,0.634,0.6252,92092,90
+openlid-v3,flores-devtest,v4,0.9186,0.8896,0.9178,0.8933,0.8906,205436,198
+openlid-v3,flores-dev,full,0.8318,0.7876,0.8042,0.775,0.8153,224325,220
+openlid-v3,flores-dev,v1,0.8767,0.4974,0.8778,0.4981,0.4968,33898,34
+openlid-v3,flores-dev,v2,0.8057,0.4814,0.8067,0.482,0.4808,36889,36
+openlid-v3,flores-dev,v3,0.8576,0.5728,0.8623,0.5783,0.5696,90727,90
+openlid-v3,flores-dev,v4,0.9178,0.8893,0.917,0.8928,0.8903,201394,198
+openlid-v3,madar,full,0.0,0.0,0.0,0.0,0.0,5600,15
+openlid-v3,madar,v1,0.0,0.0,0.0,0.0,0.0,2077,3
+openlid-v3,madar,v2,0.0,0.0,0.0,0.0,0.0,5076,11
+openlid-v3,madar,v3,0.0,0.0,0.0,0.0,0.0,5378,13
+openlid-v3,madar,v4,0.0,0.0,0.0,0.0,0.0,5378,13
+openlid-v3,gherbal-multi,full,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,gherbal-multi,v1,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,gherbal-multi,v2,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,gherbal-multi,v3,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,gherbal-multi,v4,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,atlasia-lid,full,0.0,0.0,0.0,0.0,0.0,234327,15
+openlid-v3,atlasia-lid,v1,0.0,0.0,0.0,0.0,0.0,117533,3
+openlid-v3,atlasia-lid,v2,0.0,0.0,0.0,0.0,0.0,216563,13
+openlid-v3,atlasia-lid,v3,0.0,0.0,0.0,0.0,0.0,234327,15
+openlid-v3,atlasia-lid,v4,0.0,0.0,0.0,0.0,0.0,234327,15
+openlid-v3,wili-2018,full,0.9185,0.7615,0.9211,0.767,0.7593,62000,124
+openlid-v3,wili-2018,v1,0.8407,0.3189,0.8503,0.3236,0.3152,16500,33
+openlid-v3,wili-2018,v2,0.8407,0.3189,0.8503,0.3236,0.3152,16500,33
+openlid-v3,wili-2018,v3,0.8751,0.4929,0.8762,0.4959,0.4922,31500,63
+openlid-v3,wili-2018,v4,0.9185,0.7615,0.9211,0.767,0.7593,62000,124
+openlid-v3,commonlid,full,0.6662,0.2443,0.6888,0.2394,0.272,373230,101
+openlid-v3,commonlid,v1,0.6194,0.1182,0.6516,0.1202,0.1294,269625,31
+openlid-v3,commonlid,v2,0.6193,0.1169,0.6515,0.1189,0.128,269667,33
+openlid-v3,commonlid,v3,0.6639,0.1544,0.6938,0.1564,0.167,324781,45
+openlid-v3,commonlid,v4,0.6914,0.2815,0.718,0.2782,0.3064,359646,77
+openlid-v3,bouquet,full,0.4883,0.3414,0.4059,0.3145,0.4106,289300,275
+openlid-v3,bouquet,v1,0.9022,0.2555,0.9112,0.2584,0.253,31560,30
+openlid-v3,bouquet,v2,0.8731,0.2531,0.8817,0.256,0.2506,32612,31
+openlid-v3,bouquet,v3,0.8767,0.3523,0.889,0.3645,0.3474,66276,63
+openlid-v3,bouquet,v4,0.9146,0.6856,0.922,0.7008,0.6801,152540,145
+gherbal-v1,flores-devtest,full,0.1374,0.0582,0.0571,0.0426,0.1399,222640,214
+gherbal-v1,flores-devtest,v1,0.8889,0.8258,0.8744,0.87,0.8395,34408,34
+gherbal-v1,flores-devtest,v2,0.8168,0.7656,0.7863,0.7974,0.7953,37444,36
+gherbal-v1,flores-devtest,v3,0.3321,0.226,0.2285,0.1948,0.3285,92092,90
+gherbal-v1,flores-devtest,v4,0.1489,0.0649,0.064,0.0479,0.1511,205436,198
+gherbal-v1,flores-dev,full,0.1345,0.0564,0.0557,0.0413,0.1363,224325,220
+gherbal-v1,flores-dev,v1,0.8899,0.8266,0.8752,0.8706,0.8405,33898,34
+gherbal-v1,flores-dev,v2,0.8178,0.7657,0.7864,0.7975,0.7962,36889,36
+gherbal-v1,flores-dev,v3,0.3325,0.2258,0.2282,0.1949,0.3289,90727,90
+gherbal-v1,flores-dev,v4,0.1498,0.0654,0.0647,0.0485,0.1513,201394,198
+gherbal-v1,madar,full,0.2771,0.0618,0.1621,0.0491,0.1103,5600,15
+gherbal-v1,madar,v1,0.7472,0.3197,0.7368,0.346,0.331,2077,3
+gherbal-v1,madar,v2,0.3058,0.081,0.1892,0.066,0.1363,5076,11
+gherbal-v1,madar,v3,0.2886,0.0701,0.1731,0.0565,0.122,5378,13
+gherbal-v1,madar,v4,0.2886,0.0701,0.1731,0.0565,0.122,5378,13
+gherbal-v1,gherbal-multi,full,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,gherbal-multi,v1,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,gherbal-multi,v2,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,gherbal-multi,v3,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,gherbal-multi,v4,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,atlasia-lid,full,0.2718,0.0246,0.1878,0.0194,0.0596,234327,15
+gherbal-v1,atlasia-lid,v1,0.542,0.1075,0.5271,0.1186,0.1298,117533,3
+gherbal-v1,atlasia-lid,v2,0.2941,0.0282,0.2135,0.0232,0.0649,216563,13
+gherbal-v1,atlasia-lid,v3,0.2718,0.0246,0.1878,0.0194,0.0596,234327,15
+gherbal-v1,atlasia-lid,v4,0.2718,0.0246,0.1878,0.0194,0.0596,234327,15
+gherbal-v1,wili-2018,full,0.2249,0.1189,0.1218,0.0949,0.2196,62000,124
+gherbal-v1,wili-2018,v1,0.845,0.7522,0.8206,0.7985,0.7746,16500,33
+gherbal-v1,wili-2018,v2,0.845,0.7522,0.8206,0.7985,0.7746,16500,33
+gherbal-v1,wili-2018,v3,0.4426,0.3218,0.3371,0.2973,0.4225,31500,63
+gherbal-v1,wili-2018,v4,0.2249,0.1189,0.1218,0.0949,0.2196,62000,124
+gherbal-v1,commonlid,full,0.6074,0.1645,0.5658,0.1471,0.2279,373230,101
+gherbal-v1,commonlid,v1,0.8407,0.5956,0.8516,0.5977,0.671,269625,31
+gherbal-v1,commonlid,v2,0.8406,0.5642,0.8514,0.5661,0.6357,269667,33
+gherbal-v1,commonlid,v3,0.698,0.3942,0.679,0.3792,0.4831,324781,45
+gherbal-v1,commonlid,v4,0.6303,0.2201,0.5959,0.2007,0.2946,359646,77
+gherbal-v1,bouquet,full,0.0975,0.035,0.0358,0.0247,0.0954,289300,275
+gherbal-v1,bouquet,v1,0.8935,0.7422,0.8907,0.7822,0.7446,31560,30
+gherbal-v1,bouquet,v2,0.8647,0.719,0.8582,0.7507,0.7244,32612,31
+gherbal-v1,bouquet,v3,0.4255,0.3034,0.3323,0.279,0.3885,66276,63
+gherbal-v1,bouquet,v4,0.1849,0.0935,0.0973,0.0731,0.1775,152540,145

csv/results_summary.csv

@@ -0,0 +1,81 @@
+model,benchmark,accuracy,f1_macro,f1_weighted,precision_macro,recall_macro,n_samples,n_classes
+gherbal-v2,flores-devtest,0.1495,0.0588,0.0616,0.0442,0.1452,222640,214
+gherbal-v2,flores-dev,0.147,0.0571,0.0602,0.0429,0.1419,224325,220
+gherbal-v2,madar,0.5811,0.1944,0.5625,0.1991,0.2046,5600,15
+gherbal-v2,gherbal-multi,0.7961,0.6377,0.8132,0.6733,0.6217,184994,36
+gherbal-v2,atlasia-lid,0.6561,0.1481,0.6199,0.152,0.1612,234327,15
+gherbal-v2,wili-2018,0.2374,0.1173,0.1296,0.0909,0.2149,62000,124
+gherbal-v2,commonlid,0.5934,0.1506,0.5493,0.1377,0.2193,373230,101
+gherbal-v2,bouquet,0.0986,0.0297,0.0313,0.0209,0.0935,289300,275
+gherbal-v3,flores-devtest,0.3605,0.2412,0.2548,0.216,0.3427,222640,214
+gherbal-v3,flores-dev,0.3581,0.2366,0.2514,0.2075,0.3384,224325,220
+gherbal-v3,madar,0.5745,0.2402,0.5518,0.272,0.2442,5600,15
+gherbal-v3,gherbal-multi,0.8966,0.3534,0.9028,0.3562,0.3514,184994,36
+gherbal-v3,atlasia-lid,0.6561,0.108,0.6252,0.1132,0.1282,234327,15
+gherbal-v3,wili-2018,0.4695,0.2834,0.3771,0.2544,0.3529,62000,124
+gherbal-v3,commonlid,0.7441,0.1718,0.7457,0.1667,0.2198,373230,101
+gherbal-v3,bouquet,0.1914,0.0939,0.1086,0.083,0.1655,289300,275
+gherbal-v4,flores-devtest,0.85,0.7693,0.8245,0.7712,0.7943,222640,214
+gherbal-v4,flores-dev,0.8334,0.7485,0.801,0.745,0.7798,224325,220
+gherbal-v4,madar,0.6298,0.2608,0.6169,0.316,0.2712,5600,15
+gherbal-v4,gherbal-multi,0.8699,0.163,0.8964,0.1684,0.1583,184994,36
+gherbal-v4,atlasia-lid,0.6909,0.109,0.6653,0.1283,0.114,234327,15
+gherbal-v4,wili-2018,0.9153,0.6489,0.921,0.6729,0.6448,62000,124
+gherbal-v4,commonlid,0.798,0.2286,0.8291,0.2276,0.2604,373230,101
+gherbal-v4,bouquet,0.4628,0.3074,0.3845,0.2855,0.3699,289300,275
+nllb-lid,flores-devtest,0.8331,0.7481,0.8088,0.746,0.7715,222640,214
+nllb-lid,flores-dev,0.8129,0.708,0.7853,0.7046,0.7338,224325,220
+nllb-lid,madar,0.1052,0.0087,0.0202,0.0048,0.0454,5600,15
+nllb-lid,gherbal-multi,0.7522,0.1447,0.752,0.1549,0.1429,184994,36
+nllb-lid,atlasia-lid,0.3348,0.0091,0.1686,0.0061,0.0182,234327,15
+nllb-lid,wili-2018,0.927,0.7721,0.9278,0.7828,0.7715,62000,124
+nllb-lid,commonlid,0.8609,0.2541,0.8711,0.2561,0.2757,373230,101
+nllb-lid,bouquet,0.5023,0.3398,0.4214,0.3145,0.4051,289300,275
+openlid-v1,flores-devtest,0.8425,0.7907,0.8167,0.7913,0.8166,222640,214
+openlid-v1,flores-dev,0.8187,0.7628,0.7871,0.7553,0.7944,224325,220
+openlid-v1,madar,0.5587,0.205,0.5456,0.207,0.2425,5600,15
+openlid-v1,gherbal-multi,0.8296,0.1717,0.8525,0.1794,0.1675,184994,36
+openlid-v1,atlasia-lid,0.4845,0.0591,0.468,0.0612,0.0768,234327,15
+openlid-v1,wili-2018,0.943,0.7669,0.9462,0.7743,0.7643,62000,124
+openlid-v1,commonlid,0.8405,0.2561,0.8668,0.2474,0.3004,373230,101
+openlid-v1,bouquet,0.4856,0.3299,0.3983,0.3023,0.4023,289300,275
+openlid-v2,flores-devtest,0.8748,0.8507,0.8471,0.8509,0.8798,222640,214
+openlid-v2,flores-dev,0.8508,0.8203,0.8169,0.8157,0.8556,224325,220
+openlid-v2,madar,0.6262,0.2403,0.6004,0.2358,0.289,5600,15
+openlid-v2,gherbal-multi,0.7762,0.1547,0.8126,0.1655,0.148,184994,36
+openlid-v2,atlasia-lid,0.5735,0.055,0.5322,0.0525,0.0724,234327,15
+openlid-v2,wili-2018,0.9368,0.7881,0.9407,0.7954,0.7849,62000,124
+openlid-v2,commonlid,0.8221,0.2588,0.8466,0.2509,0.2962,373230,101
+openlid-v2,bouquet,0.4941,0.3344,0.4001,0.305,0.413,289300,275
+fastlid-176,flores-devtest,0.4006,0.2516,0.3302,0.2426,0.3059,222640,214
+fastlid-176,flores-dev,0.3866,0.243,0.3166,0.2332,0.2979,224325,220
+fastlid-176,madar,0.1352,0.0228,0.0637,0.028,0.055,5600,15
+fastlid-176,gherbal-multi,0.6472,0.1282,0.6143,0.1357,0.1337,184994,36
+fastlid-176,atlasia-lid,0.3899,0.0174,0.2532,0.0187,0.0233,234327,15
+fastlid-176,wili-2018,0.7152,0.5118,0.6769,0.5197,0.5408,62000,124
+fastlid-176,commonlid,0.7702,0.1947,0.7424,0.2009,0.2226,373230,101
+fastlid-176,bouquet,0.2724,0.1558,0.2034,0.143,0.2087,289300,275
+glotlid,flores-devtest,0.9253,0.5562,0.9259,0.5635,0.5566,222640,214
+glotlid,flores-dev,0.931,0.5972,0.9317,0.6025,0.5976,224325,220
+glotlid,madar,0.5648,0.162,0.5626,0.1606,0.1912,5600,15
+glotlid,gherbal-multi,0.7772,0.0372,0.801,0.0396,0.0362,184994,36
+glotlid,atlasia-lid,0.4977,0.0229,0.4963,0.0256,0.0289,234327,15
+glotlid,wili-2018,0.9398,0.614,0.9457,0.6207,0.6101,62000,124
+glotlid,commonlid,0.7888,0.0747,0.8046,0.0734,0.0824,373230,101
+glotlid,bouquet,0.6659,0.0998,0.659,0.105,0.1008,289300,275
+openlid-v3,flores-devtest,0.8556,0.8158,0.8336,0.8059,0.8379,222640,214
+openlid-v3,flores-dev,0.8318,0.7876,0.8042,0.775,0.8153,224325,220
+openlid-v3,madar,0.0,0.0,0.0,0.0,0.0,5600,15
+openlid-v3,gherbal-multi,0.6619,0.1438,0.6995,0.1557,0.1361,184994,36
+openlid-v3,atlasia-lid,0.0,0.0,0.0,0.0,0.0,234327,15
+openlid-v3,wili-2018,0.9185,0.7615,0.9211,0.767,0.7593,62000,124
+openlid-v3,commonlid,0.6662,0.2443,0.6888,0.2394,0.272,373230,101
+openlid-v3,bouquet,0.4883,0.3414,0.4059,0.3145,0.4106,289300,275
+gherbal-v1,flores-devtest,0.1374,0.0582,0.0571,0.0426,0.1399,222640,214
+gherbal-v1,flores-dev,0.1345,0.0564,0.0557,0.0413,0.1363,224325,220
+gherbal-v1,madar,0.2771,0.0618,0.1621,0.0491,0.1103,5600,15
+gherbal-v1,gherbal-multi,0.8385,0.8311,0.8339,0.8667,0.8366,184994,36
+gherbal-v1,atlasia-lid,0.2718,0.0246,0.1878,0.0194,0.0596,234327,15
+gherbal-v1,wili-2018,0.2249,0.1189,0.1218,0.0949,0.2196,62000,124
+gherbal-v1,commonlid,0.6074,0.1645,0.5658,0.1471,0.2279,373230,101
+gherbal-v1,bouquet,0.0975,0.035,0.0358,0.0247,0.0954,289300,275

data/results_aggregate/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0148a0715641b21feb960da3c547686abb014cc1d282ca680da9937de7714e47
+size 20155

data/results_per_language/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7ef522eecf849f4b30b4e46ae62da111c67797dc4b12e1ae50426a9f0917803
+size 436793

data/results_summary/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3ff21564f0b891c5474273283e980a9ba2de9b2a7139e865c60faed3afd720b
+size 9576

space/README.md

@@ -0,0 +1,18 @@
+---
+title: LID Benchmark — Language Identification Leaderboard
+emoji: 🌍
+colorFrom: yellow
+colorTo: gray
+sdk: gradio
+sdk_version: 5.12.0
+app_file: app.py
+pinned: true
+license: apache-2.0
+short_description: Compare 10 LID models across 8 benchmarks and 214 languages
+tags:
+  - language-identification
+  - benchmark
+  - leaderboard
+  - arabic-dialects
+  - multilingual
+---

space/app.py

@@ -0,0 +1,477 @@
+"""
+LID Benchmark — Language Identification Leaderboard
+Built by Omneity Labs · https://www.omneitylabs.com
+"""
+
+import gradio as gr
+import pandas as pd
+import plotly.graph_objects as go
+from datasets import load_dataset
+
+# ── Brand colors (Omneity gold palette) ──────────────────────────────────────
+GOLD = "#C4962C"
+GOLD_LIGHT = "#E8C96A"
+GOLD_BG = "#FDF8ED"
+DARK = "#2A2520"
+MUTED = "#8A8078"
+TEAL = "#4A8C7A"
+NAVY = "#3A5A8C"
+WARM_ORANGE = "#C47A2C"
+TERRA = "#A05A3C"
+
+CHART_COLORS = [GOLD, TEAL, NAVY, WARM_ORANGE, TERRA,
+                "#7A6AAC", "#5A9A5A", "#C44A4A", "#4AC4C4", "#8A8A4A"]
+
+# ── Load data ─────────────────────────────────────────────────────────────────
+DATASET_ID = "omneity-labs/lid-benchmark"
+
+try:
+    df_summary = load_dataset(DATASET_ID, "results_summary", split="train").to_pandas()
+    df_agg = load_dataset(DATASET_ID, "results_aggregate", split="train").to_pandas()
+    df_lang = load_dataset(DATASET_ID, "results_per_language", split="train").to_pandas()
+except Exception:
+    # Fallback: load from local parquet (for dev)
+    df_summary = pd.read_parquet("../data/results_summary/train.parquet")
+    df_agg = pd.read_parquet("../data/results_aggregate/train.parquet")
+    df_lang = pd.read_parquet("../data/results_per_language/train.parquet")
+
+ALL_MODELS = sorted(df_summary["model"].unique())
+ALL_BENCHMARKS = sorted(df_summary["benchmark"].unique())
+ALL_SCOPES = sorted(df_agg["scope"].unique())
+ALL_LANGUAGES = sorted(df_lang["language"].unique())
+
+BENCHMARK_DISPLAY = {
+    "flores-devtest": "FLORES+ devtest",
+    "flores-dev": "FLORES+ dev",
+    "madar": "MADAR",
+    "gherbal-multi": "Gherbal-Multi",
+    "atlasia-lid": "ATLASIA-LID",
+    "wili-2018": "WiLI-2018",
+    "commonlid": "CommonLID",
+    "bouquet": "Bouquet",
+}
+
+MODEL_URLS = {
+    "gherbal-v4": "https://www.omneitylabs.com/models/gherbal",
+    "gherbal-v3": "https://www.omneitylabs.com/models/gherbal",
+    "gherbal-v2": "https://www.omneitylabs.com/models/gherbal",
+    "gherbal-v1": "https://www.omneitylabs.com/models/gherbal",
+    "glotlid": "https://huggingface.co/cis-lmu/glotlid",
+    "nllb-lid": "https://huggingface.co/facebook/fasttext-language-identification",
+    "openlid-v1": "https://huggingface.co/laurievb/OpenLID",
+    "openlid-v2": "https://huggingface.co/laurievb/OpenLID",
+    "openlid-v3": "https://huggingface.co/laurievb/OpenLID",
+    "fastlid-176": "https://fasttext.cc/docs/en/language-identification.html",
+}
+
+
+# ── Tab 1: Leaderboard ───────────────────────────────────────────────────────
+
+def build_leaderboard(benchmark, scope, metric):
+    if scope == "full":
+        src = df_summary[df_summary["benchmark"] == benchmark].copy()
+    else:
+        src = df_agg[(df_agg["benchmark"] == benchmark) & (df_agg["scope"] == scope)].copy()
+
+    if src.empty:
+        return pd.DataFrame({"Info": ["No data for this combination."]})
+
+    src = src.sort_values(metric, ascending=False).reset_index(drop=True)
+    src.index = src.index + 1
+    src.index.name = "Rank"
+
+    cols = ["model", metric, "f1_weighted", "n_samples", "n_classes"]
+    cols = [c for c in cols if c in src.columns]
+    display = src[cols].copy()
+    display.columns = ["Model", metric.replace("_", " ").title(),
+                        "F1 Weighted", "Samples", "Classes"][:len(cols)]
+
+    for col in display.columns[1:]:
+        if display[col].dtype == float:
+            display[col] = display[col].apply(lambda x: f"{x:.4f}" if pd.notna(x) else "—")
+
+    return display
+
+
+def leaderboard_chart(benchmark, scope, metric):
+    if scope == "full":
+        src = df_summary[df_summary["benchmark"] == benchmark].copy()
+    else:
+        src = df_agg[(df_agg["benchmark"] == benchmark) & (df_agg["scope"] == scope)].copy()
+
+    src = src.sort_values(metric, ascending=True)
+
+    fig = go.Figure()
+    fig.add_trace(go.Bar(
+        y=src["model"], x=src[metric],
+        orientation="h",
+        marker_color=[GOLD if "gherbal" in m else MUTED for m in src["model"]],
+        text=src[metric].apply(lambda x: f"{x:.4f}"),
+        textposition="outside",
+    ))
+    fig.update_layout(
+        title=f"{BENCHMARK_DISPLAY.get(benchmark, benchmark)} — {metric.replace('_', ' ').title()}",
+        xaxis_title=metric.replace("_", " ").title(),
+        yaxis_title="",
+        plot_bgcolor="white",
+        paper_bgcolor="white",
+        font=dict(family="Inter, system-ui, sans-serif", size=13, color=DARK),
+        margin=dict(l=120, r=40, t=50, b=40),
+        height=max(350, len(src) * 40 + 80),
+        xaxis=dict(range=[0, min(1.05, src[metric].max() * 1.15)],
+                   gridcolor="#f0ece4"),
+    )
+    return fig
+
+
+# ── Tab 2: Per-Language Explorer ──────────────────────────────────────────────
+
+def per_language_table(models, benchmark, scope, min_samples):
+    mask = (
+        (df_lang["benchmark"] == benchmark) &
+        (df_lang["scope"] == scope) &
+        (df_lang["model"].isin(models)) &
+        (df_lang["n_samples"] >= min_samples)
+    )
+    subset = df_lang[mask].copy()
+
+    if subset.empty:
+        return pd.DataFrame({"Info": ["No data for this combination."]})
+
+    pivot = subset.pivot_table(
+        index="language", columns="model", values="accuracy"
+    ).reset_index()
+
+    # Add sample count
+    sample_counts = subset.groupby("language")["n_samples"].first()
+    pivot = pivot.merge(sample_counts.reset_index(), on="language", how="left")
+
+    # Sort by first model's accuracy descending
+    sort_col = models[0] if models[0] in pivot.columns else pivot.columns[1]
+    pivot = pivot.sort_values(sort_col, ascending=False, na_position="last")
+
+    # Rename
+    pivot = pivot.rename(columns={"language": "Language", "n_samples": "Samples"})
+
+    # Format floats
+    for col in pivot.columns:
+        if col not in ("Language", "Samples") and pivot[col].dtype == float:
+            pivot[col] = pivot[col].apply(lambda x: f"{x:.4f}" if pd.notna(x) else "—")
+
+    return pivot
+
+
+def per_language_chart(models, benchmark, scope, min_samples, top_n):
+    mask = (
+        (df_lang["benchmark"] == benchmark) &
+        (df_lang["scope"] == scope) &
+        (df_lang["model"].isin(models)) &
+        (df_lang["n_samples"] >= min_samples)
+    )
+    subset = df_lang[mask].copy()
+
+    if subset.empty:
+        return go.Figure()
+
+    # Get top_n languages by average accuracy across selected models
+    avg_acc = subset.groupby("language")["accuracy"].mean().sort_values(ascending=False)
+    top_langs = avg_acc.head(top_n).index.tolist()
+    subset = subset[subset["language"].isin(top_langs)]
+
+    fig = go.Figure()
+    for i, model in enumerate(models):
+        model_data = subset[subset["model"] == model].set_index("language")
+        model_data = model_data.reindex(top_langs)
+        fig.add_trace(go.Bar(
+            name=model,
+            x=model_data.index,
+            y=model_data["accuracy"],
+            marker_color=CHART_COLORS[i % len(CHART_COLORS)],
+        ))
+
+    fig.update_layout(
+        barmode="group",
+        title=f"Per-Language Accuracy — {BENCHMARK_DISPLAY.get(benchmark, benchmark)}",
+        yaxis_title="Accuracy",
+        xaxis_tickangle=-45,
+        plot_bgcolor="white",
+        paper_bgcolor="white",
+        font=dict(family="Inter, system-ui, sans-serif", size=12, color=DARK),
+        legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1),
+        margin=dict(l=50, r=20, t=70, b=100),
+        height=500,
+        yaxis=dict(range=[0, 1.05], gridcolor="#f0ece4"),
+    )
+    return fig
+
+
+# ── Tab 3: Confusion Explorer ────────────────────────────────────────────────
+
+def confusion_table(model, benchmark, scope, min_samples):
+    mask = (
+        (df_lang["model"] == model) &
+        (df_lang["benchmark"] == benchmark) &
+        (df_lang["scope"] == scope) &
+        (df_lang["n_samples"] >= min_samples) &
+        (df_lang["top_confusion_1"].notna())
+    )
+    subset = df_lang[mask][
+        ["language", "accuracy", "n_samples",
+         "top_confusion_1", "top_confusion_1_count",
+         "top_confusion_2", "top_confusion_2_count",
+         "top_confusion_3", "top_confusion_3_count"]
+    ].copy()
+
+    subset = subset.sort_values("accuracy", ascending=True)
+    subset["accuracy"] = subset["accuracy"].apply(lambda x: f"{x:.4f}")
+
+    subset.columns = [
+        "Language", "Accuracy", "Samples",
+        "Top Confusion", "Count",
+        "2nd Confusion", "Count ",
+        "3rd Confusion", "Count  ",
+    ]
+    return subset
+
+
+# ── Tab 4: Model Comparison (Radar) ──────────────────────────────────────────
+
+def model_radar(models, scope):
+    benchmarks = ["flores-devtest", "madar", "gherbal-multi", "atlasia-lid"]
+    bench_labels = [BENCHMARK_DISPLAY.get(b, b) for b in benchmarks]
+
+    fig = go.Figure()
+    for i, model in enumerate(models):
+        values = []
+        for b in benchmarks:
+            if scope == "full":
+                row = df_summary[(df_summary["model"] == model) & (df_summary["benchmark"] == b)]
+            else:
+                row = df_agg[(df_agg["model"] == model) & (df_agg["benchmark"] == b) & (df_agg["scope"] == scope)]
+            val = row["accuracy"].values[0] if len(row) > 0 else 0
+            values.append(val)
+        values.append(values[0])  # close the polygon
+
+        fig.add_trace(go.Scatterpolar(
+            r=values,
+            theta=bench_labels + [bench_labels[0]],
+            fill="toself",
+            name=model,
+            line_color=CHART_COLORS[i % len(CHART_COLORS)],
+            fillcolor=CHART_COLORS[i % len(CHART_COLORS)],
+            opacity=0.3,
+        ))
+
+    fig.update_layout(
+        polar=dict(
+            radialaxis=dict(visible=True, range=[0, 1], gridcolor="#f0ece4"),
+            bgcolor="white",
+        ),
+        plot_bgcolor="white",
+        paper_bgcolor="white",
+        font=dict(family="Inter, system-ui, sans-serif", size=13, color=DARK),
+        title="Model Comparison",
+        margin=dict(l=60, r=60, t=60, b=40),
+        height=500,
+        legend=dict(orientation="h", yanchor="bottom", y=-0.15, xanchor="center", x=0.5),
+    )
+    return fig
+
+
+# ── Custom CSS ────────────────────────────────────────────────────────────────
+
+CUSTOM_CSS = """
+.gradio-container {
+    max-width: 1200px !important;
+    font-family: 'Inter', system-ui, -apple-system, sans-serif !important;
+}
+.header-banner {
+    background: linear-gradient(135deg, #2A2520 0%, #3A3530 100%);
+    border-radius: 12px;
+    padding: 28px 32px;
+    margin-bottom: 16px;
+    color: white;
+}
+.header-banner h1 {
+    color: #E8C96A !important;
+    font-size: 1.6em !important;
+    margin-bottom: 4px !important;
+    font-weight: 700 !important;
+}
+.header-banner p {
+    color: #B8B0A8 !important;
+    font-size: 0.95em !important;
+    margin: 0 !important;
+}
+.header-banner a {
+    color: #E8C96A !important;
+    text-decoration: none;
+}
+.header-banner a:hover {
+    text-decoration: underline;
+}
+footer { display: none !important; }
+"""
+
+
+# ── Build the Gradio app ─────────────────────────────────────────────────────
+
+theme = gr.themes.Soft(
+    primary_hue=gr.themes.Color(
+        c50="#FDF8ED", c100="#F8EDCC", c200="#F0DCA0",
+        c300="#E8C96A", c400="#D4AD3C", c500="#C4962C",
+        c600="#A07A20", c700="#7C5E18", c800="#584210", c900="#3A2C0A",
+        c950="#2A2008",
+    ),
+    neutral_hue=gr.themes.Color(
+        c50="#FAF8F5", c100="#F0ECE4", c200="#E0D8CC",
+        c300="#C8BEB0", c400="#A89E90", c500="#8A8078",
+        c600="#6A6258", c700="#4A4440", c800="#2A2520",
+        c900="#1A1815", c950="#0A0908",
+    ),
+    font=["Inter", "system-ui", "sans-serif"],
+)
+
+with gr.Blocks(theme=theme, css=CUSTOM_CSS, title="LID Benchmark") as demo:
+
+    gr.HTML("""
+    <div class="header-banner">
+        <h1>🌍 LID Benchmark</h1>
+        <p>
+            10 models · 8 benchmarks · 380 languages ·
+            Built by <a href="https://www.omneitylabs.com" target="_blank">Omneity Labs</a> ·
+            Model: <a href="https://www.omneitylabs.com/models/gherbal" target="_blank">Gherbal</a>
+        </p>
+    </div>
+    """)
+
+    with gr.Tabs():
+
+        # ── TAB 1: Leaderboard ────────────────────────────────────────────
+        with gr.Tab("Leaderboard"):
+            gr.Markdown("Compare all models on a benchmark. Select scope to control which languages are included.")
+            with gr.Row():
+                lb_bench = gr.Dropdown(
+                    choices=ALL_BENCHMARKS, value="flores-devtest",
+                    label="Benchmark", scale=2)
+                lb_scope = gr.Dropdown(
+                    choices=ALL_SCOPES, value="full",
+                    label="Scope", scale=1)
+                lb_metric = gr.Dropdown(
+                    choices=["accuracy", "f1_macro", "f1_weighted",
+                             "precision_macro", "recall_macro"],
+                    value="accuracy", label="Metric", scale=1)
+
+            lb_chart = gr.Plot()
+            lb_table = gr.Dataframe(label="Results")
+
+            def update_leaderboard(bench, scope, metric):
+                return leaderboard_chart(bench, scope, metric), build_leaderboard(bench, scope, metric)
+
+            for inp in [lb_bench, lb_scope, lb_metric]:
+                inp.change(update_leaderboard, [lb_bench, lb_scope, lb_metric], [lb_chart, lb_table])
+
+            demo.load(update_leaderboard, [lb_bench, lb_scope, lb_metric], [lb_chart, lb_table])
+
+        # ── TAB 2: Per-Language ───────────────────────────────────────────
+        with gr.Tab("Per-Language"):
+            gr.Markdown("Explore accuracy for individual languages. Compare how different models handle specific languages.")
+            with gr.Row():
+                pl_models = gr.Dropdown(
+                    choices=ALL_MODELS,
+                    value=["gherbal-v4", "glotlid", "openlid-v2"],
+                    multiselect=True, label="Models", scale=3)
+                pl_bench = gr.Dropdown(
+                    choices=ALL_BENCHMARKS, value="flores-devtest",
+                    label="Benchmark", scale=2)
+            with gr.Row():
+                pl_scope = gr.Dropdown(
+                    choices=ALL_SCOPES, value="full",
+                    label="Scope", scale=1)
+                pl_min_samples = gr.Slider(
+                    minimum=1, maximum=1000, value=50, step=10,
+                    label="Min samples", scale=1)
+                pl_top_n = gr.Slider(
+                    minimum=5, maximum=50, value=20, step=5,
+                    label="Top N (chart)", scale=1)
+
+            pl_chart = gr.Plot()
+            pl_table = gr.Dataframe(label="Per-Language Accuracy")
+
+            def update_per_lang(models, bench, scope, min_s, top_n):
+                if not models:
+                    return go.Figure(), pd.DataFrame({"Info": ["Select at least one model."]})
+                return (per_language_chart(models, bench, scope, min_s, top_n),
+                        per_language_table(models, bench, scope, min_s))
+
+            for inp in [pl_models, pl_bench, pl_scope, pl_min_samples, pl_top_n]:
+                inp.change(update_per_lang,
+                           [pl_models, pl_bench, pl_scope, pl_min_samples, pl_top_n],
+                           [pl_chart, pl_table])
+
+            demo.load(update_per_lang,
+                      [pl_models, pl_bench, pl_scope, pl_min_samples, pl_top_n],
+                      [pl_chart, pl_table])
+
+        # ── TAB 3: Confusion Analysis ─────────────────────────────────────
+        with gr.Tab("Confusions"):
+            gr.Markdown("See which languages are most frequently confused with each other. Sorted worst-first.")
+            with gr.Row():
+                cf_model = gr.Dropdown(
+                    choices=ALL_MODELS, value="gherbal-v4",
+                    label="Model", scale=2)
+                cf_bench = gr.Dropdown(
+                    choices=ALL_BENCHMARKS, value="flores-devtest",
+                    label="Benchmark", scale=2)
+                cf_scope = gr.Dropdown(
+                    choices=ALL_SCOPES, value="full",
+                    label="Scope", scale=1)
+                cf_min = gr.Slider(
+                    minimum=1, maximum=500, value=50, step=10,
+                    label="Min samples", scale=1)
+
+            cf_table = gr.Dataframe(label="Confusion Pairs (worst first)")
+
+            for inp in [cf_model, cf_bench, cf_scope, cf_min]:
+                inp.change(confusion_table, [cf_model, cf_bench, cf_scope, cf_min], cf_table)
+
+            demo.load(confusion_table, [cf_model, cf_bench, cf_scope, cf_min], cf_table)
+
+        # ── TAB 4: Model Comparison ───────────────────────────────────────
+        with gr.Tab("Compare"):
+            gr.Markdown("Radar chart comparing selected models across core benchmarks.")
+            with gr.Row():
+                mc_models = gr.Dropdown(
+                    choices=ALL_MODELS,
+                    value=["gherbal-v4", "glotlid", "nllb-lid"],
+                    multiselect=True, label="Models", scale=3)
+                mc_scope = gr.Dropdown(
+                    choices=ALL_SCOPES, value="full",
+                    label="Scope", scale=1)
+
+            mc_chart = gr.Plot()
+
+            def update_radar(models, scope):
+                if not models:
+                    return go.Figure()
+                return model_radar(models, scope)
+
+            for inp in [mc_models, mc_scope]:
+                inp.change(update_radar, [mc_models, mc_scope], mc_chart)
+
+            demo.load(update_radar, [mc_models, mc_scope], mc_chart)
+
+    gr.HTML("""
+    <div style="text-align: center; padding: 16px 0 8px; color: #8A8078; font-size: 0.85em;">
+        Data: <a href="https://huggingface.co/datasets/omneity-labs/lid-benchmark"
+                style="color: #C4962C;">omneity-labs/lid-benchmark</a> ·
+        Model: <a href="https://www.omneitylabs.com/models/gherbal"
+                 style="color: #C4962C;">Gherbal</a> ·
+        <a href="https://omarkamali.com" style="color: #C4962C;">Omar Kamali</a> ·
+        <a href="https://www.omneitylabs.com" style="color: #C4962C;">Omneity Labs</a>
+    </div>
+    """)
+
+
+if __name__ == "__main__":
+    demo.launch()

space/requirements.txt

@@ -0,0 +1,5 @@
+datasets>=2.0.0
+gradio==5.12.0
+pandas>=1.5.0
+plotly>=5.0.0
+pyaudioop>=0.1.0