From 0dc8ac8c2fdeed606a4711e87dfdfd2dc5824fe4 Mon Sep 17 00:00:00 2001 From: Imene Kerboua Date: Thu, 24 Oct 2024 18:49:01 +0200 Subject: [PATCH 1/2] fix dataset upload --- .../Clustering/multilingual/MLSUMClusteringP2P.py | 14 ++++++-------- .../Clustering/multilingual/MLSUMClusteringS2S.py | 14 ++++++-------- 2 files changed, 12 insertions(+), 16 deletions(-) diff --git a/mteb/tasks/Clustering/multilingual/MLSUMClusteringP2P.py b/mteb/tasks/Clustering/multilingual/MLSUMClusteringP2P.py index 97158dc144..9f823e774f 100644 --- a/mteb/tasks/Clustering/multilingual/MLSUMClusteringP2P.py +++ b/mteb/tasks/Clustering/multilingual/MLSUMClusteringP2P.py @@ -26,11 +26,10 @@ class MLSUMClusteringP2P(AbsTaskClustering, MultilingualTask): metadata = TaskMetadata( name="MLSUMClusteringP2P", description="Clustering of newspaper article contents and titles from MLSUM dataset. Clustering of 10 sets on the newpaper article topics.", - reference="https://huggingface.co/datasets/reciTAL/mlsum", + reference="https://huggingface.co/datasets/mteb/mlsum", dataset={ - "path": "reciTAL/mlsum", - "revision": "b5d54f8f3b61ae17845046286940f03c6bc79bc7", - "trust_remote_code": True, + "path": "mteb/mlsum", + "revision": "b4efe498c4d0b9d7bdd2905f6fff4e22ae251d00", }, type="Clustering", category="p2p", @@ -101,11 +100,10 @@ class MLSUMClusteringP2PFast(AbsTaskClusteringFast, MultilingualTask): metadata = TaskMetadata( name="MLSUMClusteringP2P.v2", description="Clustering of newspaper article contents and titles from MLSUM dataset. Clustering of 10 sets on the newpaper article topics.", - reference="https://huggingface.co/datasets/mlsum", + reference="https://huggingface.co/datasets/mteb/mlsum", dataset={ - "path": "reciTAL/mlsum", - "revision": "b5d54f8f3b61ae17845046286940f03c6bc79bc7", - "trust_remote_code": True, + "path": "mteb/mlsum", + "revision": "b4efe498c4d0b9d7bdd2905f6fff4e22ae251d00", }, type="Clustering", category="p2p", diff --git a/mteb/tasks/Clustering/multilingual/MLSUMClusteringS2S.py b/mteb/tasks/Clustering/multilingual/MLSUMClusteringS2S.py index 0a7eadb328..9e3978ff30 100644 --- a/mteb/tasks/Clustering/multilingual/MLSUMClusteringS2S.py +++ b/mteb/tasks/Clustering/multilingual/MLSUMClusteringS2S.py @@ -26,11 +26,10 @@ class MLSUMClusteringS2S(AbsTaskClustering, MultilingualTask): metadata = TaskMetadata( name="MLSUMClusteringS2S", description="Clustering of newspaper article contents and titles from MLSUM dataset. Clustering of 10 sets on the newpaper article topics.", - reference="https://huggingface.co/datasets/reciTAL/mlsum", + reference="https://huggingface.co/datasets/mteb/mlsum", dataset={ - "path": "reciTAL/mlsum", - "revision": "b5d54f8f3b61ae17845046286940f03c6bc79bc7", - "trust_remote_code": True, + "path": "mteb/mlsum", + "revision": "b4efe498c4d0b9d7bdd2905f6fff4e22ae251d00", }, type="Clustering", category="s2s", @@ -96,11 +95,10 @@ class MLSUMClusteringS2SFast(AbsTaskClusteringFast, MultilingualTask): metadata = TaskMetadata( name="MLSUMClusteringS2S.v2", description="Clustering of newspaper article contents and titles from MLSUM dataset. Clustering of 10 sets on the newpaper article topics.", - reference="https://huggingface.co/datasets/mlsum", + reference="https://huggingface.co/datasets/mteb/mlsum", dataset={ - "path": "reciTAL/mlsum", - "revision": "b5d54f8f3b61ae17845046286940f03c6bc79bc7", - "trust_remote_code": True, + "path": "mteb/mlsum", + "revision": "b4efe498c4d0b9d7bdd2905f6fff4e22ae251d00", }, type="Clustering", category="s2s", From 1504791740a6f7481cdaabccebbf8ea521e67d20 Mon Sep 17 00:00:00 2001 From: Imene Kerboua Date: Thu, 24 Oct 2024 18:50:40 +0200 Subject: [PATCH 2/2] add linting --- mteb/benchmarks/benchmarks.py | 1 + 1 file changed, 1 insertion(+) diff --git a/mteb/benchmarks/benchmarks.py b/mteb/benchmarks/benchmarks.py index 01a5784757..cc075ec270 100644 --- a/mteb/benchmarks/benchmarks.py +++ b/mteb/benchmarks/benchmarks.py @@ -63,6 +63,7 @@ def load_results( base_results = load_results() return base_results.select_tasks(self.tasks) + MTEB_MAIN_EN = Benchmark( name="MTEB(eng)", tasks=get_tasks(