Upload README.md
Browse files- data/README.md +13 -0
data/README.md
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Keep your dataset small and focused.
|
| 2 |
+
|
| 3 |
+
### Pretrain (`corpus_raw.txt`)
|
| 4 |
+
Short concept notes you wrote yourself (or public domain). Aim 0.5–5 MB:
|
| 5 |
+
- definitions (inertia, momentum, energy…)
|
| 6 |
+
- short derivations (one paragraph)
|
| 7 |
+
- worked mini examples (1–3 lines)
|
| 8 |
+
|
| 9 |
+
### SFT (`sft_train.jsonl`)
|
| 10 |
+
Each line: {"instruction": "...", "output": "..."}
|
| 11 |
+
Keep answers short and factual. 200–1000 examples is fine.
|
| 12 |
+
|
| 13 |
+
**Do not include copyrighted textbook dumps.**
|