Add OpenGPT-X

data/datasets.json

@@ -74,6 +74,18 @@
         "parallel": true,
         "base": "MMLU"
     },
+    {
+        "name": "MMLU-X",
+        "author": "OpenGPT-X",
+        "author_url": null,
+        "url": "https://huggingface.co/datasets/openGPT-X/mmlux",
+        "n_languages": 20,
+        "tasks": [
+            "question_answering"
+        ],
+        "parallel": true,
+        "base": "MMLU"
+    },
     {
         "name": "Global MMLU",
         "author": "Cohere",
@@ -110,6 +122,18 @@
         "parallel": true,
         "base": "MGSM"
     },
+    {
+        "name": "GSM8K-X",
+        "author": "OpenGPT-X",
+        "author_url": null,
+        "url": "https://huggingface.co/datasets/openGPT-X/gsm8kx",
+        "n_languages": 20,
+        "tasks": [
+            "math"
+        ],
+        "parallel": true,
+        "base": "MGSM"
+    },
     {
         "name": "Okapi ARC Challenge",
         "author": "Academic",
@@ -134,6 +158,18 @@
         "parallel": true,
         "base": "AI2 ARC"
     },
+    {
+        "name": "Arc-X",
+        "author": "OpenGPT-X",
+        "author_url": null,
+        "url": "https://huggingface.co/datasets/openGPT-X/arcx",
+        "n_languages": 20,
+        "tasks": [
+            "question_answering"
+        ],
+        "parallel": true,
+        "base": "AI2 ARC"
+    },
     {
         "name": "Okapi TruthfulQA",
         "author": "Academic",
@@ -158,6 +194,18 @@
         "parallel": true,
         "base": "TruthfulQA"
     },
+    {
+        "name": "TruthfulQA-X",
+        "author": "OpenGPT-X",
+        "author_url": null,
+        "url": "https://huggingface.co/datasets/openGPT-X/truthfulqax",
+        "n_languages": 20,
+        "tasks": [
+            "question_answering"
+        ],
+        "parallel": true,
+        "base": "TruthfulQA"
+    },
     {
         "name": "XNLI",
         "author": "Meta",
@@ -194,6 +242,18 @@
         "parallel": true,
         "base": "HellaSwag"
     },
+    {
+        "name": "HellaSwag-X",
+        "author": "OpenGPT-X",
+        "author_url": null,
+        "url": "https://huggingface.co/datasets/openGPT-X/hellaswagx",
+        "n_languages": 20,
+        "tasks": [
+            "question_answering"
+        ],
+        "parallel": true,
+        "base": "HellaSwag"
+    },
     {
         "name": "WikiANN / PAN-X",
         "author": "Academic",
@@ -208,7 +268,7 @@
     {
         "name": "MSVAMP",
         "author": "Microsoft",
-        "author_url": "https://microsoft.com",
+        "author_url": "https://microsoft.ai",
         "url": "https://huggingface.co/datasets/Mathoctopus/MSVAMP",
         "n_languages": 10,
         "tasks": [
@@ -256,7 +316,7 @@
     {
         "name": "XGLUE",
         "author": "Microsoft",
-        "author_url": "https://microsoft.com",
+        "author_url": "https://microsoft.ai",
         "url": "https://huggingface.co/datasets/microsoft/xglue",
         "n_languages": 18,
         "tasks": [

evals/models.py

@@ -28,6 +28,7 @@ models = [
     # "microsoft/phi-4",  # 0.07$/M tokens; only 16k tokens context
     "microsoft/phi-4-multimodal-instruct",
     "amazon/nova-micro-v1",  # 0.09$/M tokens
+    # "openGPT-X/Teuken-7B-instruct-research-v0.4",  # not on OpenRouter
 ]
 model_fast = "meta-llama/llama-3.3-70b-instruct"
 

frontend/public/results.json

@@ -9756,6 +9756,19 @@
       "base": "MMLU",
       "implemented": null
     },
+    {
+      "name": "MMLU-X",
+      "author": "OpenGPT-X",
+      "author_url": null,
+      "url": "https://huggingface.co/datasets/openGPT-X/mmlux",
+      "n_languages": 20,
+      "tasks": [
+        "question_answering"
+      ],
+      "parallel": 1.0,
+      "base": "MMLU",
+      "implemented": null
+    },
     {
       "name": "Global MMLU",
       "author": "Cohere",
@@ -9795,6 +9808,19 @@
       "base": "MGSM",
       "implemented": null
     },
+    {
+      "name": "GSM8K-X",
+      "author": "OpenGPT-X",
+      "author_url": null,
+      "url": "https://huggingface.co/datasets/openGPT-X/gsm8kx",
+      "n_languages": 20,
+      "tasks": [
+        "math"
+      ],
+      "parallel": 1.0,
+      "base": "MGSM",
+      "implemented": null
+    },
     {
       "name": "Okapi ARC Challenge",
       "author": "Academic",
@@ -9821,6 +9847,19 @@
       "base": "AI2 ARC",
       "implemented": null
     },
+    {
+      "name": "Arc-X",
+      "author": "OpenGPT-X",
+      "author_url": null,
+      "url": "https://huggingface.co/datasets/openGPT-X/arcx",
+      "n_languages": 20,
+      "tasks": [
+        "question_answering"
+      ],
+      "parallel": 1.0,
+      "base": "AI2 ARC",
+      "implemented": null
+    },
     {
       "name": "Okapi TruthfulQA",
       "author": "Academic",
@@ -9847,6 +9886,19 @@
       "base": "TruthfulQA",
       "implemented": null
     },
+    {
+      "name": "TruthfulQA-X",
+      "author": "OpenGPT-X",
+      "author_url": null,
+      "url": "https://huggingface.co/datasets/openGPT-X/truthfulqax",
+      "n_languages": 20,
+      "tasks": [
+        "question_answering"
+      ],
+      "parallel": 1.0,
+      "base": "TruthfulQA",
+      "implemented": null
+    },
     {
       "name": "XNLI",
       "author": "Meta",
@@ -9886,6 +9938,19 @@
       "base": "HellaSwag",
       "implemented": null
     },
+    {
+      "name": "HellaSwag-X",
+      "author": "OpenGPT-X",
+      "author_url": null,
+      "url": "https://huggingface.co/datasets/openGPT-X/hellaswagx",
+      "n_languages": 20,
+      "tasks": [
+        "question_answering"
+      ],
+      "parallel": 1.0,
+      "base": "HellaSwag",
+      "implemented": null
+    },
     {
       "name": "WikiANN / PAN-X",
       "author": "Academic",
@@ -9902,7 +9967,7 @@
     {
       "name": "MSVAMP",
       "author": "Microsoft",
-      "author_url": "https://microsoft.com",
+      "author_url": "https://microsoft.ai",
       "url": "https://huggingface.co/datasets/Mathoctopus/MSVAMP",
       "n_languages": 10,
       "tasks": [
@@ -9957,7 +10022,7 @@
     {
       "name": "XGLUE",
       "author": "Microsoft",
-      "author_url": "https://microsoft.com",
+      "author_url": "https://microsoft.ai",
       "url": "https://huggingface.co/datasets/microsoft/xglue",
       "n_languages": 18,
       "tasks": [