Apply explicitly selected dataset corrections

README.md

@@ -53,6 +53,17 @@ configs:
   - split: test
     path:
     - data/test/BiTOD.parquet
+- config_name: CCPE-M
+  data_files:
+  - split: train
+    path:
+    - data/train/CCPE-M.parquet
+  - split: validation
+    path:
+    - data/validation/CCPE-M.parquet
+  - split: test
+    path:
+    - data/test/CCPE-M.parquet
 - config_name: CRD3
   data_files:
   - split: train
@@ -64,6 +75,17 @@ configs:
   - split: test
     path:
     - data/test/CRD3.parquet
+- config_name: CaSiNo
+  data_files:
+  - split: train
+    path:
+    - data/train/CaSiNo.parquet
+  - split: validation
+    path:
+    - data/validation/CaSiNo.parquet
+  - split: test
+    path:
+    - data/test/CaSiNo.parquet
 - config_name: CraigslistBargains
   data_files:
   - split: train
@@ -108,6 +130,17 @@ configs:
   - split: test
     path:
     - data/test/ICSI.parquet
+- config_name: INSPIRED
+  data_files:
+  - split: train
+    path:
+    - data/train/INSPIRED.parquet
+  - split: validation
+    path:
+    - data/validation/INSPIRED.parquet
+  - split: test
+    path:
+    - data/test/INSPIRED.parquet
 - config_name: KVRET
   data_files:
   - split: train
@@ -433,7 +466,7 @@ Roles are already oriented for user-simulator training:
 <!-- available-datasets:start -->
 ## Available Datasets
 
-**Dataset count: 35**
+**Dataset count: 38**
 
 Select an individual dataset by its name below, or use `default` for the combined collection.
 
@@ -458,10 +491,10 @@ Datasets are grouped by behavioral category:
     "news-interview", "wizard_of_internet", "wizard_of_wikipedia"
   ],
   "Negotiation, Persuasion & Debate": [
-    "CraigslistBargains", "contextual-abuse", "diplomacy-corpus", "iq2-corpus"
+    "CaSiNo", "CraigslistBargains", "contextual-abuse", "diplomacy-corpus", "iq2-corpus"
   ],
   "Preferences & Recommendations": [
-    "DuRecDial-2.0", "OpenDialKG", "Redial"
+    "CCPE-M", "DuRecDial-2.0", "INSPIRED", "OpenDialKG", "Redial"
   ],
   "Role-Playing & Character Simulation": [
     "CRD3"

SOURCES.md

@@ -5,7 +5,8 @@ perspectives. The changes are role assignment, group-preserving train/validation
 splits, explicit source instruction extraction, and the source-specific filtering
 listed below. Target utterance wording is retained, except that DeliData export
 quoting is decoded: matching outer single quotes are removed and doubled apostrophes
-within those wrapped strings become single apostrophes. Other text is unchanged.
+within those wrapped strings become single apostrophes. INSPIRED also decodes its
+literal QUOTATION_MARK token to a double quote. Other retained text is unchanged.
 Human-authored scripted material
 is included; AI messages can appear as context but are not training targets.
 
@@ -47,3 +48,32 @@ validation/test partitions are combined into validation/test respectively.
 Dialogue wording and order are unchanged. Reading material and ratings are not
 included in training messages; no system messages or generated personas are added.
 Original split identities and annotations are retained locally.
+
+## CCPE-M
+
+Original release: [Filip Radlinski, Krisztian Balog, Bill Byrne and Karthik Krishnamoorthi, Google LLC](https://github.com/google-research-datasets/ccpe), revision
+`2c9cd30f33f3a154b5a27d015333679262ff36f5`.
+Terms: [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). These source-specific terms also apply to the reformatted files.
+
+Changes: User-target role reassignment and deterministic conversation-level splits; original dialogue text and order unchanged; annotations omitted.
+
+## CaSiNo
+
+Original release: [Kushal Chawla, Jaysa Ramirez, Rene Clever, Gale Lucas, Jonathan May and Jonathan Gratch](https://github.com/kushalchawla/CaSiNo), revision
+`2f6ed4a6a55110152a7699fbaa8150d6036314be`.
+Terms: [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). These source-specific terms also apply to the reformatted files.
+
+Changes: Both target-human perspectives; interface actions removed without replacement; own resource priorities and reasons formatted as system context; native splits and remaining dialogue text preserved; other metadata omitted.
+
+## INSPIRED
+
+Original release: [Shirley Anugrah Hayati, Dongyeop Kang, Qingxiaoyang Zhu, Weiyan Shi and Zhou Yu](https://github.com/sweetpeach/Inspired), revision
+`782d7c1845d11f75d2542809c7f2a51b63ef0304`.
+The source repository does not specify a redistribution license. No license or
+independently verified redistribution permission is inferred by this collection.
+
+Changes: Seeker-target role reassignment; literal QUOTATION_MARK decoded to double quotes; other text, utterance boundaries, and native splits unchanged; annotations and surveys omitted.
+
+The released files contain 801 training, 99 development, and 99 test conversations,
+despite the source documentation describing 100 development and 100 test conversations.
+The available conversations are retained without fabricating missing records.

data/test/CCPE-M.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f13631d50e17bfe742c777b5d1e3dbb4d099b45010c7b767bbef76ae017fde0f
+size 17842

data/test/CaSiNo.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93eb8406d97289bc89429091653e2d207977b2bcc5f81ca4694f57368e8befa3
+size 77609

data/test/INSPIRED.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d6d93428dda60c2edcade943a351bef4bfd3fd1b63843eb18d14a9b535676e0
+size 76380

data/train/CCPE-M.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5765d66d837bfee1e37cd017546704c322da67da14ebab6112fba1b1720f5ed
+size 297740

data/train/CaSiNo.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c27c10ade082ed25a190cf0142adf44cb81f7e80f332bc6cdac4e486e080baf2
+size 657349

data/train/INSPIRED.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c5f23a9a16768f2afefc212968b6c910ebbeca8252a43a698673d76a1e414f5
+size 609757

data/validation/CCPE-M.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9096cc8da8719e81e0d92823a44037d4b48a7898c386a142b9ce6c9e7c6986ba
+size 23313

data/validation/CaSiNo.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50e64ab3e42bb58e53a5884db3a7532fe56220f0ec992d4333b52908ff74eaf2
+size 23064

data/validation/INSPIRED.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59545c55084317ecbd068e36582783130f62302de86f8cf614b7f9d27e92d26e
+size 77470