Upload README.md with huggingface_hub

README.md

@@ -1,78 +1,21 @@
 ---
-dataset_info:
-- config_name: corpus
-  features:
-  - name: _id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: text
-    dtype: string
-  - name: corpus_language
-    dtype: string
-  - name: base_id
-    dtype: string
-  - name: depth
-    dtype: int64
-  - name: mode_added
-    dtype: string
-  - name: source_publication_number
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 231228
-    num_examples: 180
-  download_size: 42133
-  dataset_size: 231228
-- config_name: qrels
-  features:
-  - name: query-id
-    dtype: string
-  - name: corpus-id
-    dtype: string
-  - name: score
-    dtype: float64
-  - name: depth
-    dtype: int64
-  splits:
-  - name: train
-    num_bytes: 56808
-    num_examples: 804
-  download_size: 5190
-  dataset_size: 56808
-- config_name: queries
-  features:
-  - name: _id
-    dtype: string
-  - name: text
-    dtype: string
-  - name: query_language
-    dtype: string
-  - name: base_id
-    dtype: string
-  - name: strategy
-    dtype: string
-  - name: concept_chebi_id
-    dtype: string
-  - name: term_used
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 40078
-    num_examples: 134
-  download_size: 23865
-  dataset_size: 40078
 configs:
 - config_name: corpus
   data_files:
   - split: train
-    path: data/corpus/train-*
-- config_name: qrels
+    path: data/corpus/*.parquet
+- config_name: queries
   data_files:
   - split: train
-    path: data/qrels/train-*
-- config_name: queries
+    path: data/queries/*.parquet
+- config_name: qrels
   data_files:
   - split: train
-    path: data/queries/train-*
+    path: data/qrels/*.parquet
 ---
+
+# MehdiAstaraki/progressive-code-switch
+
+**Progressive code-switching** retrieval-decay benchmark. Each base patent yields a cumulative ladder of documents (`base__r0` clean → `base__rN`), where each step swaps one more chemistry term into another language / spelling / ChEBI form. One fixed question per base (about the step-1 term) is reused for every depth; the `qrels` carry the `depth` so you can measure how retrieval decays as more terms are code-switched.
+
+Configs: `corpus` (ladder variant documents, with `depth` + `mode_added`), `queries` (one per base), `qrels` (query→variant, score 1, with `depth`). The retrieval haystack is this `corpus` PLUS a shared patent corpus passed at eval time.