Micro Kiki V3: 489K Technical Instruction-Following Examples for LLM Fine-Tuning

489,000 instruction-following examples across 35 technical domains, designed for fine-tuning code and electronics large language models. The dataset was created by author clemsail for the micro-kiki project and was last updated on 2026-04-17.

Use Cases

• Fine-tuning code generation models based on the 35 technical domains mentioned in the description
• Training domain-expert LoRA adapters for electronics-focused LLMs based on the dataset's stated purpose
• Evaluating instruction-following capabilities in multi-domain technical contexts based on the dataset's scope

Strengths

• 489,000 examples provide substantial volume for training
• Covers 35 distinct technical domains for specialized fine-tuning
• Designed for a specific project architecture (MoE-LoRA on Qwen3.5-35B-A3B)

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

huggingface, author clemsail

Collection Method

Includes ~50,000 examples from multi-machine Claude Code sessions across GrosMac, Studio, Tower, CILS, KXKM-AI

Time Range

null

Freshness

Last updated 2026-04-17 15:48:51

Geography

null