sanjeevafk commited on
Commit
22d52f3
·
verified ·
1 Parent(s): 287f8fa

Refresh dataset from local Supabase low-memory exporter

Browse files
LICENSE_SUMMARY.md ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # License Summary
2
+
3
+ This dataset is mixed-license. Downstream consumers must respect the upstream license attached to each chunk.
4
+
5
+ | Upstream License | Documents |
6
+ | --- | ---: |
7
+ | unknown | 100 |
8
+
9
+ ## Policy
10
+
11
+ - No dataset-wide MIT claim is made for upstream source content.
12
+ - Per-chunk `upstream_license`, `source`, and `source_url` fields are authoritative.
13
+ - Unknown licenses remain in the export and should be filtered for conservative enterprise use.
README.md CHANGED
@@ -1,80 +1,46 @@
1
- ---
2
- language:
3
- - en
4
- license: mit
5
- tags:
6
- - rag
7
- - technical-docs
8
- - programming
9
- - cs
10
- - programming-books
11
- size_categories:
12
- - 100K<n<1M
13
- dataset_info:
14
- features:
15
- - name: chunk_id
16
- dtype: large_string
17
- - name: doc_id
18
- dtype: large_string
19
- - name: collection_id
20
- dtype: large_string
21
- - name: collection_name
22
- dtype: large_string
23
- - name: namespace
24
- dtype: large_string
25
- - name: source_name
26
- dtype: large_string
27
- - name: source_url
28
- dtype: large_string
29
- - name: raw_text
30
- dtype: large_string
31
- - name: cleaned_text
32
- dtype: large_string
33
- - name: tags
34
- dtype: large_string
35
- - name: chunk_order
36
- dtype: int64
37
- splits:
38
- - name: postmortems_backfill
39
- num_bytes: 310519841
40
- num_examples: 248982
41
- download_size: 153166348
42
- dataset_size: 310519841
43
- configs:
44
- - config_name: default
45
- data_files:
46
- - split: postmortems_backfill
47
- path: data/postmortems_backfill-*
48
- ---
49
 
50
- # DepthAPI Technical Corpus
51
 
52
- This dataset contains a comprehensive technical corpus optimized for Retrieval-Augmented Generation (RAG). It includes ~240k semantic chunks of high-quality, trusted technical documentation and books.
53
 
54
- ## Dataset Contents
55
-
56
- - MDN Web Docs
57
- - Kubernetes Documentation
58
- - CPython Documentation
59
- - Node.js API Docs
60
- - React.dev Content
61
- - Various Notes for Professionals (Java, Python, SQL, JS, TS, etc.)
62
- - Algorithms and System Design Primers
63
 
 
64
 
 
 
 
65
 
66
  ## Schema
67
 
68
- - `chunk_id`: Unique identifier for the chunk.
69
- - `doc_id`: Identifier for the source document.
70
- - `namespace`: Category/namespace (e.g., `trusted`).
71
- - `source_name`: Human-readable name of the source (e.g., "CPython Docs").
72
- - `source_url`: URL or source locator.
73
- - `raw_text`: The raw text content of the chunk.
74
- - `cleaned_text`: The cleaned, parsed markdown/text content ready for embedding.
75
- - `tags`: Comma-separated list of tags (e.g., "python, stdlib, P0").
76
- - `chunk_order`: Integer representing the sequential order of the chunk in the document.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
77
 
78
- ## Intended Use
79
 
80
- This dataset is ideal for training and evaluating Large Language Models (LLMs) on technical coding tasks, as well as serving as a high-quality knowledge base for hybrid search / RAG pipelines.
 
1
+ # depthapi_technical_corpus
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
+ ## Summary
4
 
5
+ `depthapi_technical_corpus` is a mixed-license retrieval corpus built from technical documentation, books, engineering writeups, and repository-derived reference material. It is intended for open-source RAG systems, embedding benchmarks, reranker evaluation, and enterprise retrieval experiments.
6
 
7
+ ## Ingestion Process
 
 
 
 
 
 
 
 
8
 
9
+ The corpus was assembled from multiple upstream source types and normalized through a reproducible local pipeline backed by Supabase.
10
 
11
+ - `Scrapling` was used for live technical documentation crawling and structured HTML extraction from documentation websites.
12
+ - `opendataloader-pdf` was used for PDF extraction when ingesting book-like and document-style technical sources into normalized markdown/text blocks.
13
+ - Source material was then normalized, rechunked deterministically, deduplicated, validated, and exported to parquet for Hugging Face datasets compatibility.
14
 
15
  ## Schema
16
 
17
+ Each chunk includes:
18
+
19
+ - `chunk_id`
20
+ - `source`
21
+ - `source_url`
22
+ - `upstream_license`
23
+ - `document_id`
24
+ - `chunk_index`
25
+ - `retrieved_at`
26
+ - `chunker_version`
27
+ - `content_hash`
28
+ - `content`
29
+
30
+ ## Usage
31
+
32
+ ```python
33
+ from datasets import load_dataset
34
+
35
+ dataset = load_dataset("sanjeevafk/depthapi_technical_corpus", split="train", streaming=True)
36
+ for row in dataset.take(3):
37
+ print(row["chunk_id"], row["source"], row["upstream_license"])
38
+ ```
39
+
40
+ ## Retrieval Benchmark Example
41
+
42
+ Use `data/research_corpus/benchmarks/queries.jsonl` with your retriever and score against `qrels.jsonl`.
43
 
44
+ ## Licensing
45
 
46
+ This is a mixed-license dataset. Downstream users must inspect `upstream_license` and `SOURCES_MANIFEST.yaml` before redistribution or commercial use.
SOURCES_MANIFEST.yaml ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ version: 1
2
+ sources:
3
+ - source: BashNotesForProfessionals
4
+ source_url: https://goalkicker.com/books#bash
5
+ upstream_license: unknown
6
+ retrieved_at: unknown
7
+ documents: 1
8
+ - source: CPython Docs
9
+ source_url: file://datasets/cpython/Doc/reference/executionmodel.rst
10
+ upstream_license: unknown
11
+ retrieved_at: unknown
12
+ documents: 8
13
+ - source: Cracking the coding interview 6th edition
14
+ source_url: https://goalkicker.com/books#cracking the coding interview 6th edition
15
+ upstream_license: unknown
16
+ retrieved_at: unknown
17
+ documents: 2
18
+ - source: Deep Learning with Python
19
+ source_url: https://www.manning.com/books/deep-learning-with-python-third-edition
20
+ upstream_license: unknown
21
+ retrieved_at: unknown
22
+ documents: 1
23
+ - source: Kubernetes Website EN
24
+ source_url: file://datasets/website/content/en/docs/concepts/workloads/controllers/job.md
25
+ upstream_license: unknown
26
+ retrieved_at: unknown
27
+ documents: 20
28
+ - source: MDN Content EN-US
29
+ source_url: file://datasets/content/files/en-us/web/api/paymentresponse/retry/index.md
30
+ upstream_license: unknown
31
+ retrieved_at: unknown
32
+ documents: 55
33
+ - source: Node.js API Docs
34
+ source_url: file://datasets/node/doc/api/worker_threads.md
35
+ upstream_license: unknown
36
+ retrieved_at: unknown
37
+ documents: 5
38
+ - source: NodeJSNotesForProfessionals
39
+ source_url: https://goalkicker.com/books#nodejs
40
+ upstream_license: unknown
41
+ retrieved_at: unknown
42
+ documents: 1
43
+ - source: React.dev Content
44
+ source_url: file://datasets/react.dev/src/content/reference/dev-tools/react-performance-tracks.md
45
+ upstream_license: unknown
46
+ retrieved_at: unknown
47
+ documents: 5
48
+ - source: Tech Postmortems Full
49
+ source_url: https://www.elastic.co/blog/elastic-cloud-january-18-2019-incident-report
50
+ upstream_license: unknown
51
+ retrieved_at: unknown
52
+ documents: 2
data/postmortems_backfill-00000-of-00001.parquet → train-00000.parquet RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:123af62de807df5775e0c1fca43a22eb0a0581921ccb6b72950c30bb671ac1f1
3
- size 153166348
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7bcc9b9f7f912c64b4a35365874906e4596d881ec73cba7c15f531f25c0a5734
3
+ size 27848
data/train-00000-of-00001.parquet → train-00001.parquet RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e7677d72e3f7e3f47d71723ca6bd846ab6049884daaa983f390a2f29f64b5a3e
3
- size 145041587
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671925d9442b900d238fa3abc207b58cba6d7a683f8491293e55f828ff87d52
3
+ size 25619
train-00002.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95c71e75197408f02a61f98219ac1161cb26c6cae5b8c3ffe79f5d5f553a3d28
3
+ size 31204
train-00003.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f0f5670dc5b5702b1be560f9046a694747e3cc4f32170b4e456963f4fc5b0fc
3
+ size 29247