Ravert commited on
Commit
15551d1
·
verified ·
1 Parent(s): 7ede2c9

Regenerate from Nowy_Dataset with exam-ready Alpaca prompts

Browse files
Files changed (4) hide show
  1. README.md +30 -38
  2. data/batch_00000.jsonl +0 -0
  3. data/batch_00000.parquet +2 -2
  4. metadata.json +13 -34
README.md CHANGED
@@ -1,39 +1,31 @@
1
- ---
2
- language:
3
- - pl
4
- tags:
5
- - unsloth
6
- - alpaca
7
- - machine-learning
8
- - python
9
- - exam-prep
10
- configs:
11
- - config_name: data
12
- data_files: data/*.parquet
13
- default: true
14
- ---
15
-
16
- # Test2 - Exam Ready
17
-
18
- Poprawiona wersja datasetu pod kolokwium z uczenia maszynowego w Pythonie.
19
-
20
- Zamiast prostego QA z cytatem dataset uczy model:
21
- - rozpoznawać typ problemu po opisie datasetu,
22
- - wykonać EDA i preprocessing,
23
- - dobrać model albo metodę,
24
- - napisać kod Python,
25
- - wskazać metryki i typowe pułapki.
26
-
27
- ## Format
28
-
29
- Alpaca / Unsloth:
30
- - `instruction`
31
- - `input`
32
- - `output`
33
-
34
- Metadata:
35
- - `topic`
36
- - `task_type`
37
- - `source_file`
38
-
39
  Records: 1000
 
1
+ ---
2
+ language:
3
+ - pl
4
+ tags:
5
+ - unsloth
6
+ - alpaca
7
+ - gemma
8
+ - machine-learning
9
+ - python
10
+ - exam-prep
11
+ configs:
12
+ - config_name: data
13
+ data_files: data/*.parquet
14
+ default: true
15
+ ---
16
+
17
+ # ML Kolokwium Dataset
18
+
19
+ Dataset Alpaca/Unsloth do fine-tuningu Gemma 4 E4B-IT pod kolokwium z uczenia maszynowego w Pythonie.
20
+
21
+ Model ma uczyć się: dostaję URL/dataset, rozpoznaję typ problemu, wykonuję EDA, preprocessing, model/metodę i ocenę.
22
+
23
+ Kolumny:
24
+ - instruction
25
+ - input
26
+ - output
27
+ - source_file
28
+ - topic
29
+ - task_type
30
+
 
 
 
 
 
 
 
 
31
  Records: 1000
data/batch_00000.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
data/batch_00000.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:121ffbc7c2cc343b45d9e8b030f919a58c11dd3865233c36562522b62a79dedf
3
- size 156860
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb09763c404e4752cc30b116fd54b30fbc969006a3b6cf20588838ec69ebd487
3
+ size 99916
metadata.json CHANGED
@@ -1,34 +1,13 @@
1
- {
2
- "records": 1000,
3
- "schema": {
4
- "instruction": "string",
5
- "input": "string",
6
- "output": "string",
7
- "topic": "string",
8
- "task_type": "string",
9
- "source_file": "string"
10
- },
11
- "topic_counts": {
12
- "pozyskiwanie danych": 138,
13
- "braki danych i outliery": 77,
14
- "normalizacja i kodowanie": 65,
15
- "podział danych i walidacja": 128,
16
- "wizualizacja danych": 120,
17
- "KNN": 92,
18
- "drzewa decyzyjne": 99,
19
- "regresja liniowa": 49,
20
- "regresja logistyczna": 73,
21
- "klasteryzacja": 45,
22
- "PCA": 53,
23
- "klasyfikacja": 61
24
- },
25
- "task_type_counts": {
26
- "original_lab_prompt": 123,
27
- "code_repair": 30,
28
- "lecture_concept_application": 40,
29
- "md_exercise_solution": 124,
30
- "exam_scenario": 50,
31
- "exam_dataset_scenario": 633
32
- },
33
- "goal": "Fine-tuning pod kolokwium: student dostaje dataset i stosuje wiedzę z labów 1-12."
34
- }
 
1
+ {
2
+ "records": 1000,
3
+ "schema": {
4
+ "output": "string",
5
+ "task_type": "string",
6
+ "source_file": "string",
7
+ "topic": "string",
8
+ "input": "string",
9
+ "instruction": "string"
10
+ },
11
+ "goal": "Gemma 4 E4B-IT fine-tuning pod kolokwium ML: dataset URL -\u003e EDA -\u003e preprocessing -\u003e model/metoda -\u003e ocena.",
12
+ "source_dir": "C:\\Users\\rober\\Downloads\\ML\\Nowy_Dataset"
13
+ }