nondual
Datasets
All datasets matching “nondual”alpaca-gpt4-nondual
Alpaca-GPT4 (Non-dual)
Version: v0.1.0 (pin with revision="v0.1.0")
This dataset is a derivative of [vicgalle/alpaca-gpt4].It preserves the original Alpaca schema and only changes the output to a non-dual response.
Schema
instruction (str)
input (str; may be empty)
output (str) — non-dual style rewrite
Provenance & Changes
Base: vicgalle/alpaca-gpt4
Modification: Only output was rewritten into a non-dual voice; instruction and input are unchanged.
Rows: ~52,000… See the full description on the dataset page: https://huggingface.co/datasets/marciodiaz/alpaca-gpt4-nondual.Nondual-Bot-Benchmark
Nondual Bot Benchmark v1.0
50 questions that separate genuine contemplative understanding from parroting, new-age contamination, and spiritual bypassing.
This benchmark was designed for one purpose: to stress-test AI models claiming to engage with nondual or contemplative teaching. It is harder than it looks.
The Problem This Solves
Every major LLM has been trained on spiritual text. Every major LLM will confidently produce spiritual-sounding output. Most of it is… See the full description on the dataset page: https://huggingface.co/datasets/Sathman/Nondual-Bot-Benchmark.openmath-nondual
nondual_openmath_final
A non-dual reformulation of the unsloth/OpenMathReasoning-mini dataset.All assistant solutions have been rewritten into impersonal, non-dual language using OpenAI models, and finalized so that the dataset no longer contains duplicate *_nondual fields.
Dataset Summary
Source: unsloth/OpenMathReasoning-mini
Format: JSONL, each line is a dictionary with the following fields:
Field
Description
problem
Math problem statement (rewritten… See the full description on the dataset page: https://huggingface.co/datasets/marciodiaz/openmath-nondual.non_dual_alpaca_datasetfine-tome-100k-nondual
fine_tome_100k_nondual
A non-dual reformulation of the mlabonne/FineTome-100k dataset.All assistant outputs (from: gpt) have been rewritten into impersonal, non-dual language using OpenAI models.User inputs and other roles remain unchanged.
Dataset Summary
Source: FineTome-100k
Size: ~100,000 conversations (JSONL, one per line)
Format: ShareGPT-style conversations, with fields:{
"conversations": [
{"from": "user", "value": "User message..."},
{"from": "gpt"… See the full description on the dataset page: https://huggingface.co/datasets/marciodiaz/fine-tome-100k-nondual.non_dual_alpaca_dataset-2