Add TSV files with column headers and update dataset card

.gitattributes

@@ -63,3 +63,10 @@ BaatCheet_Syn_I/eng_tam/tam_tgt.txt filter=lfs diff=lfs merge=lfs -text
 BaatCheet_Syn_I/eng_tel/tel_tgt.txt filter=lfs diff=lfs merge=lfs -text
 BaatCheet_Syn_I/hin_tam/tam_tgt.txt filter=lfs diff=lfs merge=lfs -text
 BaatCheet_Syn_I/hin_tel/tel_tgt.txt filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/eng_hin/eng_hin.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/eng_hin_tel_tam.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/eng_tam/eng_tam.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/eng_tel/eng_tel.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/hin_tam/hin_tam.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/hin_tel/hin_tel.tsv filter=lfs diff=lfs merge=lfs -text
+BaatCheet_Syn_I/hin_tel_tam.tsv filter=lfs diff=lfs merge=lfs -text

BaatCheet_Syn_I/eng_hin/eng_hin.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaf9f93ff987b212621c1daa43a0f67d659db3e1cc233a42265857756e99af4a
+size 16708141

BaatCheet_Syn_I/eng_hin_tel_tam.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c20d2a41bedb73c3c7c5d9ccab65cdcb41af39e8fc0bb1244c0905f87fe4f41d
+size 44434292

BaatCheet_Syn_I/eng_tam/eng_tam.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00ef560e148f227aebfaaafad632854d57d919e8d77985cb65b40c7dc66e7cad
+size 19548125

BaatCheet_Syn_I/eng_tel/eng_tel.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5bf6154a856f219803e0df19f12d65368c8f0c6a88ce3fb8f381f3de5ac3245
+size 17434220

BaatCheet_Syn_I/hin_tam/hin_tam.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:399aae53d5c777fc7ad606d48b3e38244aaeeb74a614aece84b7de07ee477356
+size 21053922

BaatCheet_Syn_I/hin_tel/hin_tel.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26ac787f385b024f11bab9afb5d80a90c944946c2c2b6193a34d0e25ec996181
+size 19389420

BaatCheet_Syn_I/hin_tel_tam.tsv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c5336980eb1397151bdd48d4b6f16c1f814167f16aa3bc50e7ceb26fb3af2df
+size 31620368

README.md

@@ -20,115 +20,137 @@ source_datasets:
 task_categories:
 - translation
 pretty_name: BaatCheet Corpus
+tags:
+- dialogue
+- indian-languages
+- low-resource
 ---
 
 # BaatCheet Corpus
 
-**BaatCheet Corpus** is a benchmark dialogue translation dataset for Indian languages across 5 language pairs:
-- **English – Hindi** (`eng_hin`)
-- **English – Tamil** (`eng_tam`)
-- **English – Telugu** (`eng_tel`)
-- **Hindi – Tamil** (`hin_tam`)
-- **Hindi – Telugu** (`hin_tel`)
+**BaatCheet Corpus** is a multilingual dialogue translation dataset for Indian languages. It covers **5 language pairs** across 4 subsets — two synthetic training sets, one human-annotated training set, and one benchmark test set.
 
 ---
 
-## 📊 Language Pair Matrix & Columns
+## 🌐 Language Pairs
 
-The dataset is organized in tabular TSV format with column headers for source and target languages:
+| Source Language | Target Languages |
+|---|---|
+| English (`eng`) | Hindi (`hin`), Telugu (`tel`), Tamil (`tam`) |
+| Hindi (`hin`) | Telugu (`tel`), Tamil (`tam`) |
 
-### 1. English Source Datasets (`eng` $\rightarrow$ `hin`, `tel`, `tam`)
-| Source Column | Target Column 1 | Target Column 2 | Target Column 3 |
-|---|---|---|---|
-| `eng` | `hin` | `tel` | `tam` |
+---
+
+## 📁 Dataset Subsets
+
+| Subset | Description | Size |
+|---|---|---|
+| `BaatCheet_Syn_I` | Synthetic Train Set I — COMET quality-filtered, utterance-level best MT output | ~79K eng / ~52K hin utterances |
+| `BaatCheet_Syn_II` | Synthetic Train Set II | ~23K eng / ~16K hin utterances |
+| `BaatCheet_Gold/BaatCheet_Human_Train` | Human-annotated Gold Training Set | ~1.8K utterances per pair |
+| `BaatCheet_Gold/BaatCheet_Test` | Gold Benchmark Test Set (dailytalk, mutual, news, hinglishgit) | ~2.1K eng / ~1.6K hin utterances |
+
+---
+
+## 📄 File Format
+
+All files are **tab-separated (TSV)** with the first row as a **column header** containing language codes.
 
-**Multi-column TSV Header Format**:
-```tsv
+### English Source Files (`eng` → `hin`, `tel`, `tam`)
+
+**Multi-target combined file** (`eng_hin_tel_tam.tsv`) — only in `BaatCheet_Syn_I`:
+
+```
 eng	hin	tel	tam
+Speaker 1 : What would you like to drink?	वक्ता 1: आप क्या पीना पसंद करेंगे?	స్పీకర్ 1: మీరు ఏమి తాగాలనుకుంటున్నారు?	பேச்சாளர் 1: நீங்கள் என்ன குடிக்க விரும்புகிறீர்கள்?
+Speaker 2 : I feel like a cup of black tea.	वक्ता 2: मुझे एक कप काली चाय पीने का मन है।	స్పీకర్ 2: నాకు ఒక కప్పు బ్లాక్ టీ లాగా అనిపిస్తుంది.	Speaker 2: எனக்கு ஒரு கப் கருப்பு தேநீர் குடிக்க வேண்டும் போலிருக்கிறது.
 ```
 
-### 2. Hindi Source Datasets (`hin` $\rightarrow$ `tel`, `tam`)
-| Source Column | Target Column 1 | Target Column 2 |
-|---|---|---|
-| `hin` | `tel` | `tam` |
+**Per-pair file** (`eng_hin/eng_hin.tsv`, `eng_tel/eng_tel.tsv`, `eng_tam/eng_tam.tsv`):
 
-**Multi-column TSV Header Format**:
-```tsv
-hin	tel	tam
+```
+eng	hin
+Speaker 1 : What would you like to drink?	वक्ता 1: आप क्या पीना पसंद करेंगे?
+Speaker 2 : I feel like a cup of black tea.	वक्ता 2: मुझे एक कप काली चाय पीने का मन है।
+	
+Speaker 1 : Hey! How are you?	वक्ता 1: हे! कैसे हो?
 ```
 
----
+### Hindi Source Files (`hin` → `tel`, `tam`)
 
-## 📁 Repository Structure
+**Multi-target combined file** (`hin_tel_tam.tsv`) — only in `BaatCheet_Syn_I`:
 
-```text
-Baatcheet_Corpus/
-├── BaatCheet_Syn_I/                  # Synthetic Trainset I (COMET Quality Filtered)
-│   ├── eng_hin_tel_tam.tsv           # Multi-target TSV: (eng \t hin \t tel \t tam)
-│   ├── hin_tel_tam.tsv               # Multi-target TSV: (hin \t tel \t tam)
-│   ├── eng_hin/ (eng_hin.tsv)
-│   ├── eng_tam/ (eng_tam.tsv)
-│   ├── eng_tel/ (eng_tel.tsv)
-│   ├── hin_tam/ (hin_tam.tsv)
-│   └── hin_tel/ (hin_tel.tsv)
-├── BaatCheet_Syn_II/                 # Synthetic Trainset II
-│   ├── eng_hin/ (eng_hin.tsv)
-│   ├── eng_tam/ (eng_tam.tsv)
-│   ├── eng_tel/ (eng_tel.tsv)
-│   ├── hin_tam/ (hin_tam.tsv)
-│   └── hin_tel/ (hin_tel.tsv)
-└── BaatCheet_Gold/
-    ├── BaatCheet_Human_Train/        # Human-Annotated Gold Training Set
-    │   ├── eng_hin/ (eng_hin.tsv)
-    │   ├── eng_tam/ (eng_tam.tsv)
-    │   ├── eng_tel/ (eng_tel.tsv)
-    │   ├── hin_tam/ (hin_tam.tsv)
-    │   └── hin_tel/ (hin_tel.tsv)
-    └── BaatCheet_Test/               # Gold Benchmark Test Set
-        ├── eng_hin/ (eng_hin.tsv)
-        ├── eng_tam/ (eng_tam.tsv)
-        ├── eng_tel/ (eng_tel.tsv)
-        ├── hin_tam/ (hin_tam.tsv)
-        └── hin_tel/ (hin_tel.tsv)
 ```
+hin	tel	tam
+राहुल : हे पल्लवी! कैसी हो?	రాహుల్: హే పల్లవి! ఎలా ఉన్నావు?	ராகுல்: ஹே பல்லவி! எப்படி இருக்கீங்க?
+पल्लवी : हे राहुल! ठीक हूँ।	పల్లవి: హే రాహుల్! బాగానే ఉంది.	பல்லவி: ஹே ராகுல்! நான் நலம்.
+```
+
+**Per-pair file** (`hin_tel/hin_tel.tsv`, `hin_tam/hin_tam.tsv`):
+
+```
+hin	tel
+राहुल : हे पल्लवी! कैसी हो?	రాహుల్: హే పల్లవి! ఎలా ఉన్నావు?
+पल्लवी : हे राहुल! ठीक हूँ।	పల్లవి: హే రాహుల్! బాగానే ఉంది.
+```
+
+> **Note:** Dialogues are separated by an empty line (blank `\t` blank row).
 
 ---
 
-## 📄 File & Column Format
+## 📁 Repository Structure
 
-### Single-Pair TSV Format (`<src>_<tgt>.tsv`)
-Each pair directory contains a TSV file with explicit language column headers:
-```tsv
-eng	hin
-Speaker 1 : Wow, this is amazing.	स्पीकर 1: वाह, यह अद्भुत है।
-Speaker 2 : Sorry, they have the machine.	स्पीकर 2: सॉरी, इनके पास मशीन है।
 ```
-
-### Multi-Target TSV Format (`eng_hin_tel_tam.tsv` and `hin_tel_tam.tsv`)
-Aligned synthetic training sets contain multi-column TSV files:
-```tsv
-eng	hin	tel	tam
-Speaker 1 : How are you?	स्पीकर 1: आप कैसे हैं?	స్పీకర్ 1: మీరు ఎలా ఉన్నారు?	சபாநாயகர் 1: எப்படி இருக்கிறீர்கள்?
+BaatCheet_Corpus/
+├── BaatCheet_Syn_I/
+│   ├── eng_hin_tel_tam.tsv     ← eng | hin | tel | tam (multi-target)
+│   ├── hin_tel_tam.tsv         ← hin | tel | tam (multi-target)
+│   ├── eng_hin/ (eng_hin.tsv)  ← eng | hin
+│   ├── eng_tam/ (eng_tam.tsv)  ← eng | tam
+│   ├── eng_tel/ (eng_tel.tsv)  ← eng | tel
+│   ├── hin_tam/ (hin_tam.tsv)  ← hin | tam
+│   └── hin_tel/ (hin_tel.tsv)  ← hin | tel
+│
+├── BaatCheet_Syn_II/
+│   ├── eng_hin/ (eng_hin.tsv)  ← eng | hin
+│   ├── eng_tam/ (eng_tam.tsv)  ← eng | tam
+│   ├── eng_tel/ (eng_tel.tsv)  ← eng | tel
+│   ├── hin_tam/ (hin_tam.tsv)  ← hin | tam
+│   └── hin_tel/ (hin_tel.tsv)  ← hin | tel
+│
+└── BaatCheet_Gold/
+    ├── BaatCheet_Human_Train/
+    │   ├── eng_hin/ (eng_hin.tsv)  ← eng | hin
+    │   ├── eng_tam/ (eng_tam.tsv)  ← eng | tam
+    │   ├── eng_tel/ (eng_tel.tsv)  ← eng | tel
+    │   ├── hin_tam/ (hin_tam.tsv)  ← hin | tam
+    │   └── hin_tel/ (hin_tel.tsv)  ← hin | tel
+    └── BaatCheet_Test/
+        ├── eng_hin/ (eng_hin.tsv)  ← eng | hin
+        ├── eng_tam/ (eng_tam.tsv)  ← eng | tam
+        ├── eng_tel/ (eng_tel.tsv)  ← eng | tel
+        ├── hin_tam/ (hin_tam.tsv)  ← hin | tam
+        └── hin_tel/ (hin_tel.tsv)  ← hin | tel
 ```
 
 ---
 
-## 💻 Loading Data in Python (Pandas / Datasets)
+## 💻 Loading in Python
 
-### Using Pandas:
 ```python
 import pandas as pd
 
-# Load 4-column English-source translation table
-df_eng = pd.read_csv("BaatCheet_Syn_I/eng_hin_tel_tam.tsv", sep="\t")
-print(df_eng.columns) # ['eng', 'hin', 'tel', 'tam']
+# Load 4-column English source translations (BaatCheet_Syn_I only)
+df = pd.read_csv("BaatCheet_Syn_I/eng_hin_tel_tam.tsv", sep="\t", keep_default_na=False)
+print(df.columns.tolist())  # ['eng', 'hin', 'tel', 'tam']
+print(df.head(3))
 
-# Load 3-column Hindi-source translation table
-df_hin = pd.read_csv("BaatCheet_Syn_I/hin_tel_tam.tsv", sep="\t")
-print(df_hin.columns) # ['hin', 'tel', 'tam']
+# Load a single language pair
+df_pair = pd.read_csv("BaatCheet_Gold/BaatCheet_Test/eng_hin/eng_hin.tsv", sep="\t", keep_default_na=False)
+print(df_pair.columns.tolist())  # ['eng', 'hin']
 ```
 
 ---
 
 ## 📜 License
-This dataset is distributed under the [Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/) License.
+[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/)