umutkkgz commited on
Commit
1e544cc
·
verified ·
1 Parent(s): f5054f6

Upload KAIRA master fine-tune dataset

Browse files
.gitattributes CHANGED
@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
61
+ KAIRA_MASTER_PHASE_2_INTELLIGENCE.jsonl filter=lfs diff=lfs merge=lfs -text
KAIRA_MASTER_PHASE_1_STABILIZATION.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
KAIRA_MASTER_PHASE_2_INTELLIGENCE.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5eaef63bd890be2f1cf1305cbc927fd567c8b3978c277626d516786184f57668
3
+ size 83703593
KAIRA_MASTER_PHASE_3_CONSCIOUS.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
README.md ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - tr
4
+ license: other
5
+ task_categories:
6
+ - text-generation
7
+ - question-answering
8
+ - summarization
9
+ tags:
10
+ - turkish
11
+ - sft
12
+ - instruction-tuning
13
+ - reasoning
14
+ - kaira
15
+ pretty_name: KAIRA Master Fine Tune
16
+ size_categories:
17
+ - 10K<n<100K
18
+ ---
19
+
20
+ # KAIRA Master Fine Tune
21
+
22
+ Turkish instruction-tuning dataset prepared for KAIRA fine-tuning.
23
+
24
+ ## Files
25
+
26
+ | File | Rows | Purpose |
27
+ | --- | ---: | --- |
28
+ | `KAIRA_MASTER_PHASE_1_STABILIZATION.jsonl` | 28,225 | Stabilization, Turkish basics, summaries, definitions, translation, daily language |
29
+ | `KAIRA_MASTER_PHASE_2_INTELLIGENCE.jsonl` | 50,532 | Longer analytical answers, planning, review, technical and scientific explanations |
30
+ | `KAIRA_MASTER_PHASE_3_CONSCIOUS.jsonl` | 6,998 | Self-correction, intent tracking, plan/review style responses |
31
+
32
+ Total rows: 85,755.
33
+
34
+ ## Format
35
+
36
+ Each row is JSONL with OpenAI-style chat messages:
37
+
38
+ ```json
39
+ {"messages":[{"role":"user","content":"..."},{"role":"assistant","content":"..."}]}
40
+ ```
41
+
42
+ Some assistant responses include structured tags such as `<think>`, `<plan>`,
43
+ `<review>`, and `<intent>` to support reasoning, planning, and self-correction
44
+ behavior during supervised fine-tuning.
45
+
46
+ ## Checksums
47
+
48
+ ```text
49
+ 57c900f1ea2e92207ef492dbb03c8cda64ea541baf1fca464107a5469f2038e6 KAIRA_MASTER_PHASE_1_STABILIZATION.jsonl
50
+ 5eaef63bd890be2f1cf1305cbc927fd567c8b3978c277626d516786184f57668 KAIRA_MASTER_PHASE_2_INTELLIGENCE.jsonl
51
+ c6561eb21f098113ae1771bbdaf48df8ae48b7748d19a983fefdc5e6b8df2c18 KAIRA_MASTER_PHASE_3_CONSCIOUS.jsonl
52
+ ```