Datasetpublic

d4rkninja/tanpo-hiring-sft

Tanpo Hiring SFT (10k) Ownership Owner: DarkNinja Solutions Creator: d4rkninja Community: DarkLab Domain Hiring and talent: role design, sourcing, interviewing, scorecards, offers, onboarding, and org design. Dataset summary Field Value Rows 10000 Schema Chat SFT (messages with system / user / assistant) Source file tanpo-hiring-sft-10k-format-fixed.jsonl Audit verdict PASS Empty assistant rate 0.0% Exact… See the full description on the dataset page: https://huggingface.co/datasets/d4rkninja/tanpo-hiring-sft.

sourceHugging Faceapache-2.0updated 9d agoView on Hugging Face

0likes65downloads

Dataset Card

Tanpo Hiring SFT (10k)

Ownership

—Owner: DarkNinja Solutions
—Creator: d4rkninja
—Community: DarkLab

Domain

Hiring and talent: role design, sourcing, interviewing, scorecards, offers, onboarding, and org design.

Dataset summary

Field	Value
Rows	10000
Schema	Chat SFT (`messages` with `system` / `user` / `assistant`)
Source file	`tanpo-hiring-sft-10k-format-fixed.jsonl`
Audit verdict	PASS
Empty assistant rate	0.0%
Exact user+assistant dupes	0
Assistant length (min/median/max chars)	753 / 2588 / 4024

Intended use

Supervised fine-tuning (SFT) data for LFM2.5-1.2B Tanpo specialists — specifically the hiring domain model.

Related model repo (if published): `d4rkninja/tanpo-hiring`

Format

Each line is a JSON object:

json

{
  "messages": [
    {"role": "system", "content": "..."},
    {"role": "user", "content": "..."},
    {"role": "assistant", "content": "..."}
  ]
}

File layout in this repo: data/train.jsonl

Attribution

System prompts identify the specialist as owned by DarkNinja Solutions, created by d4rkninja, under the DarkLab community.

Audit

See DarkLab internal audit: ALL_DATASETS_PUSH_AUDIT.md (2026-09-17).

License

Apache 2.0 (dataset card default; respect any third-party content constraints if present in prompts).