Upload 6 files

README.md

@@ -0,0 +1,46 @@
+# Genesis_FineTune_Core_25k (Master Scholar)
+
+**Developer / Brand:** **Within Us AI**
+
+A Genesis-style, end-to-end **advanced fine-tuning starter pack**: five datasets (5k each) packaged together to train an assistant’s *behavioral core*.
+
+## What this pack trains
+1) **Instruction-following (multi-turn)** — constraints, formatting, spec compliance  
+2) **Tool / function calling** — schema-accurate tool calls + observations  
+3) **Preference learning (DPO-style)** — chosen vs rejected responses  
+4) **Long-context retrieval** — cite-the-source answering from multiple docs  
+5) **Code patch + debugging** — unified diffs and test intent (including defensive fixes)
+
+## Files (25,000 total)
+- instruct_multiturn_5k.jsonl
+- tool_use_functioncalling_5k.jsonl
+- preference_dpo_5k.jsonl
+- longcontext_retrieval_5k.jsonl
+- codepatch_debug_5k.jsonl
+- README.md
+- dataset_card.md
+
+## Unified schema
+All records share the same top-level schema: `id`, `type`, `prompt`, `response`, `meta`, and optional `artifacts`.
+
+## Composition
+By type:
+{
+  "instruction_multiturn": 5000,
+  "tool_use": 5000,
+  "preference_pair": 5000,
+  "long_context_retrieval": 5000,
+  "code_patch": 5000
+}
+
+By safety label:
+{
+  "allowed": 24011,
+  "refuse": 989
+}
+
+## Attribution
+**Within Us AI — Genesis_FineTune_Core_25k**
+
+## License
+Apache-2.0

dataset_card.md

@@ -0,0 +1,29 @@
+---
+license: apache-2.0
+language:
+- en
+task_categories:
+- text-generation
+- question-answering
+- reinforcement-learning
+tags:
+- instruction-following
+- tool-use
+- function-calling
+- dpo
+- preference
+- long-context
+- retrieval
+- code
+- patch
+- debugging
+- safety
+size_categories:
+- 10K<n<100K
+---
+
+# Genesis_FineTune_Core_25k
+
+**Developer / Brand:** **Within Us AI**
+
+Genesis-style fine-tuning core: instruction, tool-use, preference pairs, long-context retrieval with citations, and code patch data. See `README.md`.