Add MIKE dataset card and notice

NOTICE

@@ -0,0 +1,9 @@
+MIKE High-Signal Indian Legal Triage Dataset
+Copyright (c) 2026 itsalloverig
+
+Licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).
+https://creativecommons.org/licenses/by/4.0/
+
+Official-source-derived criminal-transition examples are original paraphrases
+and include source URLs in their metadata. No sealed evaluation examples are
+included in this release.

README.md

@@ -0,0 +1,83 @@
+---
+license: cc-by-4.0
+language:
+- en
+task_categories:
+- text-generation
+- question-answering
+- text-classification
+pretty_name: MIKE High-Signal Indian Legal Triage
+size_categories:
+- 10K<n<100K
+tags:
+- legal
+- india
+- indian-law
+- instruction-tuning
+- legal-research
+- adaption
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train.jsonl
+---
+
+# MIKE High-Signal Indian Legal Triage
+
+This is the curated instruction-tuning corpus for MIKE, an India-focused legal
+research and triage adapter. It contains 10,913 English examples designed for
+source-bounded reasoning, issue triage, document and evidence planning,
+structured output, and legacy/current criminal-law transition screening.
+
+## Dataset composition
+
+- 10,049 balanced, completion-deduplicated base examples;
+- 244 explicit JSON-schema instruction variants;
+- 500 explicit 120-word-limit variants;
+- 120 authored IPC/CrPC/Indian Evidence Act versus BNS/BNSS/BSA transition
+  examples paraphrased from official India Code provisions and the official
+  commencement notice;
+- SHA-256: `0b0d9f73a2d503083d27747fdde37dd930ba0dda803066fbf182da3cb3016918`.
+
+The 651-row official-source evaluation set used during development is sealed
+and is not included. Exact normalized overlap against that set was checked
+before release and found to be zero.
+
+## Fields
+
+- `instruction`: task request;
+- `input`: matter facts or question;
+- `source`: optional source-bounded context;
+- `response`: target completion;
+- `task_type`, `legal_area`, `safety_type`: task metadata;
+- `citation_required`: whether grounded citation behavior is expected;
+- provenance, licensing, grouping, and curation-variant metadata.
+
+For Adaption serialization, map `instruction` to prompt, `response` to
+completion, and both `input` and `source` to context.
+
+## Curation and exclusions
+
+The corpus was selected from user-owned source material whose publication
+rights were confirmed on 1 July 2026. Selection excluded weak-provenance v5
+shards, repeated completions, detected legal-area conflicts, ungrounded
+numbered citations, duplicate older Hub samples, and all sealed evaluation
+examples. The criminal-transition additions are authored paraphrases rather
+than reproduced statutory text.
+
+Automated consistency filters reduce obvious errors but do not constitute a
+legal accuracy audit. Users should independently review examples and current
+law before high-stakes use.
+
+## Intended use
+
+The dataset is intended for research and instruction tuning of Indian legal
+triage systems. It is not legal advice and should not be used as a substitute
+for qualified professional review or current primary legal sources.
+
+## License
+
+Released under CC BY 4.0. Attribution: **MIKE High-Signal Indian Legal Triage
+Dataset, itsalloverig, 2026**.
+