Update README.md

README.md

@@ -392,7 +392,7 @@ size_categories:
 
 ## Dataset Summary
 A corpus of GitHub issues and pull requests from the
-[huggingface/datasets](https://github.com/huggingface/datasets) repository.
+[JPEG XL support](https://github.com/huggingface/datasets/issues/8537) repository.
 Created as a practice exercise following AT1-Task01 and Hugging Face LLM Course ("Creating your own dataset").
 
 ## How the data was collected
@@ -403,6 +403,19 @@ Created as a practice exercise following AT1-Task01 and Hugging Face LLM Course
 - This "raw" version keeps the original API fields with minimal processing
   and does not include issue comments.
 
+## Data Collection Process
+1. **Authentication:** A read-only GitHub personal access token was stored in Kaggle Secret Key
+   , and loaded at runtime. It was never written into the notebook.
+2. **Collection:** Issues and pull requests were requested from the GitHub REST API
+   using Python `requests`, page by page, staying within GitHub's rate limit of 5,000 requests per hour.
+4. **Raw storage:** Responses were saved as a JSON Lines file.
+5. **Loading:** The file was loaded with `datasets.load_dataset.
+6. **Cleaning:** 100 duplicate records (caused by a pagination error) were removed using
+   `number` as a unique key.
+7. **Augmentation:** Derived fields were added.
+8. **Publishing:** The dataset was pushed to the Hugging Face Hub with `push_to_hub()`,
+   authenticated with a Hugging Face write token stored in Secret Keys.
+
 ## Dataset Structure
 Each row is one issue or pull request. 
 Key fields:
@@ -420,22 +433,18 @@ Key fields:
 - Multilabel classification of issue labels
 - Semantic search over issues
 
-## Data Collection Process
-1. **Authentication:** A read-only GitHub personal access token was stored in Kaggle Secret Key
-   , and loaded at runtime. It was never written into the notebook.
-2. **Collection:** Issues and pull requests were requested from the GitHub REST API
-   using Python `requests`, page by page, staying within GitHub's rate limit of 5,000 requests per hour.
-4. **Raw storage:** Responses were saved as a JSON Lines file (`datasets-issues.jsonl`).
-5. **Loading:** The file was loaded with `datasets.load_dataset("json", ...)`.
-6. **Cleaning:** 100 duplicate records (caused by a pagination error) were removed using
-   `number` as a unique key.
-7. **Augmentation:** Derived fields were added (`is_pull_request`, [`days_to_close`, `label_names`]).
-8. **Publishing:** The dataset was pushed to the Hugging Face Hub with `push_to_hub()`,
-   authenticated with a Hugging Face write token stored in Secret Keys.
+## Storage, Locations and Versioning
+| Item | Location | Versioning |
+| Published dataset | https://huggingface.co/datasets/Color0106/github-raw | Hugging Face commit history (Files and versions → History) |
+| Raw data | Collection date: [2026-10-05] |
+| Notebook 1 – Collection & augmentation | [Kaggle Notebook] | [Ver 1 / 2026-10-05] |
+| Notebook 2 – Transformers pipeline | [Colab Notebook] | [Ver 2 / 2026/10/18] |
+| Notebook 3 – Semantic search (FAISS) | [Colab Notebook] | [Ver 1 / 2026/10/18] |
+| Access tokens | Kaggle Secrets / Colab Secrets — not stored in any file | Expire on [2026/12/05] |
 
 ## Limitations
 - A snapshot in time; newer issues are not included.
-- (If you limited `num_issues`) Only a subset of all issues was collected.
+- Only a subset of all issues was collected.
 - Text was written by GitHub users and may contain usernames, code, and links.
 
 ## Legal and Policy Compliance
@@ -470,4 +479,4 @@ Shared here for educational purposes.
 - Office of the Australian Information Commissioner. (n.d.). *Australian Privacy Principles*.
   https://www.oaic.gov.au/privacy/australian-privacy-principles
 - *Copyright Act 1968* (Cth).
-- North Metropolitan TAFE. [Title of the data storage/management policy on the LMS].
+