docs: rewrite dataset card in English to avoid encoding issues

README.md

@@ -19,56 +19,56 @@ source_datasets:
 
 # JupiMind datasets
 
-Repo n?y l?u c?c dataset ?? ???c chu?n h?a v? snapshot l?i ?? ph?c v? training cho h? JupiMind. Hi?n t?i asset ch?nh l? mix pretrain d?ng cho `JupiMind-Dense-J1R-Tokenizer-24k`.
+This repository stores normalized dataset snapshots used for JupiMind training workflows. The primary asset at the moment is the pretraining mix built for `aevynt/JupiMind-Dense-J1R-Tokenizer-24k`.
 
-## Hi?n c? trong repo
+## Files currently included
 
 - `pretrain/j1r_24k_7b_mix/j1r_t4x4_24k_mix.jsonl`
 - `pretrain/j1r_24k_7b_mix/j1r_t4x4_24k_mix.summary.json`
 
-## Dataset ch?nh: J1R 24k 7B mix
+## Main dataset: J1R 24k 7B mix
 
-- Lo?i: pretrain JSONL
-- Format: m?i d?ng l? `{"text": "..."}`
-- Tokenizer target: `aevynt/JupiMind-Dense-J1R-Tokenizer-24k`
-- S? sequence: `1,710,000`
-- Context target: `4096`
-- ??c l??ng token budget: x?p x? `~7B` token theo chunk size `4094`
-- Output g?c ???c build t?i: `/home/trainer/jupimind/outputs/pretrain/j1r_t4x4_24k_mix.jsonl`
+- Type: pretraining JSONL
+- Format: one record per line as `{"text": "..."}`
+- Target tokenizer: `aevynt/JupiMind-Dense-J1R-Tokenizer-24k`
+- Generated sequences: `1,710,000`
+- Target context length: `4096`
+- Estimated token budget: about `~7B` tokens using chunk size `4094`
+- Original build output path: `/home/trainer/jupimind/outputs/pretrain/j1r_t4x4_24k_mix.jsonl`
 
-## Th?nh ph?n mix th?c t?
+## Actual source mix after backfill
 
-Do hai ngu?n tech nh? kh?ng ?? volume, pipeline ?? backfill thi?u h?t b?ng `fineweb2_vi`. B?ng d??i l? **t? tr?ng th?c t? sau khi build xong**, kh?ng ph?i t? tr?ng l? t??ng ban ??u.
+Two smaller technical sources did not contain enough material to reach their planned quotas, so the build pipeline backfilled the missing volume with `fineweb2_vi`. The table below reports the **actual final composition**, not the initial intended weights.
 
-| Source key | Upstream dataset | Config | Written sequences | T? tr?ng th?c t? | Target ban ??u | Thi?u h?t ngu?n | Backfill |
+| Source key | Upstream dataset | Config | Written sequences | Actual share | Initial target | Source shortfall | Backfill |
 | --- | --- | --- | ---: | ---: | ---: | ---: | ---: |
-| `fineweb2_vi` | `epfml/FineWeb2-HQ` | `vie_Latn ` | 781,741 | 45.72% | 564,300 | 0 | 217,441 |
-| `vietnamese_curated` | `VTSNLP/vietnamese_curated_dataset` | `- ` | 376,200 | 22.00% | 376,200 | 0 | 0 |
-| `wikipedia_vi` | `wikimedia/wikipedia` | `20231101.vi ` | 171,000 | 10.00% | 171,000 | 0 | 0 |
-| `stackv2_edu` | `common-pile/stackv2_edu_filtered` | `- ` | 256,500 | 15.00% | 256,500 | 0 | 0 |
-| `github_ai_docs` | `meowterspace42/github-ai-project-docs` | `- ` | 3,047 | 0.18% | 136,800 | 133,753 | 0 |
-| `tech_docs` | `saidsef/tech-docs` | `- ` | 1,812 | 0.11% | 85,500 | 83,688 | 0 |
-| `fineweb_edu` | `HuggingFaceFW/fineweb-edu` | `sample-10BT ` | 119,700 | 7.00% | 119,700 | 0 | 0 |
+| `fineweb2_vi` | `epfml/FineWeb2-HQ` | `vie_Latn` | 781,741 | 45.72% | 564,300 | 0 | 217,441 |
+| `vietnamese_curated` | `VTSNLP/vietnamese_curated_dataset` | `-` | 376,200 | 22.00% | 376,200 | 0 | 0 |
+| `wikipedia_vi` | `wikimedia/wikipedia` | `20231101.vi` | 171,000 | 10.00% | 171,000 | 0 | 0 |
+| `stackv2_edu` | `common-pile/stackv2_edu_filtered` | `-` | 256,500 | 15.00% | 256,500 | 0 | 0 |
+| `github_ai_docs` | `meowterspace42/github-ai-project-docs` | `-` | 3,047 | 0.18% | 136,800 | 133,753 | 0 |
+| `tech_docs` | `saidsef/tech-docs` | `-` | 1,812 | 0.11% | 85,500 | 83,688 | 0 |
+| `fineweb_edu` | `HuggingFaceFW/fineweb-edu` | `sample-10BT` | 119,700 | 7.00% | 119,700 | 0 | 0 |
 
-## Provenance v? license
+## Provenance and licensing
 
-Repo n?y **kh?ng g?n m?t blanket license duy nh?t cho to?n b? data content**. M?i th?nh ph?n gi? ?i?u kho?n upstream c?a n?. Xem th?m file [`LICENSE`](./LICENSE) v? [`SOURCES.md`](./SOURCES.md).
+This repository does **not** apply a single blanket content license to the mixed dataset. Each source portion remains governed by its upstream terms. See [`LICENSE`](./LICENSE) and [`SOURCES.md`](./SOURCES.md) for details.
 
-| Upstream dataset | ?i?u kho?n / license upstream | Link |
+| Upstream dataset | Upstream license or terms | Link |
 | --- | --- | --- |
 | `epfml/FineWeb2-HQ` | ODC-By 1.0; also subject to CommonCrawl Terms of Use | [upstream](https://huggingface.co/datasets/epfml/FineWeb2-HQ) |
-| `VTSNLP/vietnamese_curated_dataset` | No explicit Hub license field; derived from multiple open Vietnamese corpora. See upstream card and underlying component datasets. | [upstream](https://huggingface.co/datasets/VTSNLP/vietnamese_curated_dataset) |
+| `VTSNLP/vietnamese_curated_dataset` | No explicit Hub license field; review the upstream dataset card and underlying component datasets. | [upstream](https://huggingface.co/datasets/VTSNLP/vietnamese_curated_dataset) |
 | `wikimedia/wikipedia` | CC-BY-SA 3.0 and GFDL | [upstream](https://huggingface.co/datasets/wikimedia/wikipedia) |
-| `common-pile/stackv2_edu_filtered` | Per-document open-source licenses in metadata; see upstream card and per-row metadata | [upstream](https://huggingface.co/datasets/common-pile/stackv2_edu_filtered) |
-| `meowterspace42/github-ai-project-docs` | Apache-2.0 per Hub metadata; upstream repos may still have individual licenses | [upstream](https://huggingface.co/datasets/meowterspace42/github-ai-project-docs) |
+| `common-pile/stackv2_edu_filtered` | Open-source licenses vary by document; review upstream card and per-row metadata. | [upstream](https://huggingface.co/datasets/common-pile/stackv2_edu_filtered) |
+| `meowterspace42/github-ai-project-docs` | Apache-2.0 per Hub metadata; original source repositories may still carry their own licenses. | [upstream](https://huggingface.co/datasets/meowterspace42/github-ai-project-docs) |
 | `saidsef/tech-docs` | Apache-2.0 | [upstream](https://huggingface.co/datasets/saidsef/tech-docs) |
 | `HuggingFaceFW/fineweb-edu` | ODC-By 1.0; also subject to CommonCrawl Terms of Use | [upstream](https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu) |
 
-## L?u ? s? d?ng
+## Usage notes
 
-- ??y l? m?t **derived training mix** ?? ph?c v? pretraining, kh?ng ph?i b?n ph?n ph?i g?c c?a t?ng upstream dataset.
-- Ng??i d?ng downstream ph?i t? ??nh gi? r?i ro ph?p l?, n?i dung, PII, v? ?i?u kho?n s? d?ng c?a t?ng upstream source tr??c khi hu?n luy?n ho?c ph?n ph?i model.
-- V?i `VTSNLP/vietnamese_curated_dataset` v? `common-pile/stackv2_edu_filtered`, m?t ph?n ?i?u kho?n ph? thu?c v?o component datasets ho?c per-document metadata ? upstream.
+- This is a **derived training mix** assembled for pretraining, not a replacement distribution for the original source datasets.
+- Downstream users are responsible for reviewing legal, safety, privacy, attribution, and redistribution obligations of each upstream source before training or deploying a model.
+- `VTSNLP/vietnamese_curated_dataset` and `common-pile/stackv2_edu_filtered` require extra care because their final content rights depend on upstream components or per-document metadata.
 
 ## Quick load
 

SOURCES.md

@@ -11,13 +11,13 @@
 - Link: https://huggingface.co/datasets/epfml/FineWeb2-HQ
 - Config used: `vie_Latn`
 - Actual written sequences: 781,741
-- Notes: received all backfill when smaller tech sources ran out of data.
+- Notes: received all backfill once smaller technical sources ran out of material.
 - Upstream terms: ODC-By 1.0; also subject to CommonCrawl Terms of Use
 
 ### 2. VTSNLP/vietnamese_curated_dataset
 - Link: https://huggingface.co/datasets/VTSNLP/vietnamese_curated_dataset
 - Actual written sequences: 376,200
-- Upstream terms: No explicit Hub license field; derived from multiple open Vietnamese corpora. See upstream card and underlying component datasets.
+- Upstream terms: No explicit Hub license field; review the upstream dataset card and underlying component datasets.
 
 ### 3. wikimedia/wikipedia
 - Link: https://huggingface.co/datasets/wikimedia/wikipedia
@@ -28,13 +28,13 @@
 ### 4. common-pile/stackv2_edu_filtered
 - Link: https://huggingface.co/datasets/common-pile/stackv2_edu_filtered
 - Actual written sequences: 256,500
-- Upstream terms: Per-document open-source licenses in metadata; see upstream card and per-row metadata
+- Upstream terms: Open-source licenses vary by document; review upstream card and per-row metadata.
 
 ### 5. meowterspace42/github-ai-project-docs
 - Link: https://huggingface.co/datasets/meowterspace42/github-ai-project-docs
 - Actual written sequences: 3,047
 - Shortfall vs target: 133,753
-- Upstream terms: Apache-2.0 per Hub metadata; upstream repos may still have individual licenses
+- Upstream terms: Apache-2.0 per Hub metadata; original source repositories may still carry their own licenses.
 
 ### 6. saidsef/tech-docs
 - Link: https://huggingface.co/datasets/saidsef/tech-docs