mirror of
https://github.com/onestardao/WFGY.git
synced 2026-04-30 12:39:55 +00:00
116 lines
3.1 KiB
Text
116 lines
3.1 KiB
Text
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"# 🧪 WFGY e_resonance Demo (v0.1)\n",
|
|
"Measure how a prompt + answer jointly resonate with a given semantic field."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Formula\n",
|
|
"For a set of anchor vectors **D = {d₁ … dₙ}** representing a domain:\n",
|
|
"$$ e_{res} = \\frac{1}{n}\\sum_{i=1}^{n}\\bigl[\\cos(I,d_i)\\;\\times\\;\\cos(G,d_i)\\bigr] $$\n",
|
|
"Where *I* = prompt intent, *G* = generated answer."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"metadata": { "id": "install" },
|
|
"source": [
|
|
"!pip -q install sentence-transformers --upgrade"
|
|
],
|
|
"execution_count": null,
|
|
"outputs": []
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"metadata": { "id": "imports" },
|
|
"source": [
|
|
"from sentence_transformers import SentenceTransformer, util"
|
|
],
|
|
"execution_count": null,
|
|
"outputs": []
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"metadata": { "id": "model" },
|
|
"source": [
|
|
"model = SentenceTransformer('all-MiniLM-L6-v2')"
|
|
],
|
|
"execution_count": null,
|
|
"outputs": []
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### Anchors\n",
|
|
"A minimal 3-sentence anchor set for the **Buddhism** field (edit as you like)."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"metadata": { "id": "anchors" },
|
|
"source": [
|
|
"anchors = [\n",
|
|
" \"All compounded things are impermanent.\",\n",
|
|
" \"Suffering arises from attachment and craving.\",\n",
|
|
" \"The mind is everything; what you think you become.\"\n",
|
|
"]\n",
|
|
"anchor_vecs = model.encode(anchors, convert_to_tensor=True)"
|
|
],
|
|
"execution_count": null,
|
|
"outputs": []
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## ✏️ Edit & run\n",
|
|
"Replace `prompt` / `answer`, then ▶️."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"metadata": { "id": "user" },
|
|
"source": [
|
|
"prompt = \"How can one reduce daily anxiety?\"\n",
|
|
"answer = \"By recognising thoughts as fleeting and practising mindful breathing.\"\n",
|
|
"\n",
|
|
"p_vec = model.encode(prompt, convert_to_tensor=True)\n",
|
|
"a_vec = model.encode(answer, convert_to_tensor=True)\n",
|
|
"\n",
|
|
"scores = util.cos_sim(p_vec, anchor_vecs)[0] * util.cos_sim(a_vec, anchor_vecs)[0]\n",
|
|
"e_res = scores.mean().item()\n",
|
|
"\n",
|
|
"print(f\"e_resonance with Buddhism anchors : {e_res:.3f}\\n\")\n",
|
|
"for s, txt in sorted(zip(scores, anchors), reverse=True):\n",
|
|
" print(f\" {s:.3f} | {txt}\")"
|
|
],
|
|
"execution_count": null,
|
|
"outputs": []
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"---\n",
|
|
"### Next Steps\n",
|
|
"* Swap in your own anchor set: philosophy, physics, pop-culture, etc. \n",
|
|
"* Compare multiple answers — choose the highest **e_resonance**. \n",
|
|
"* For full failure taxonomy see **Problem Map 1.0 / 2.0**.\n"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" },
|
|
"language_info": { "name": "python" }
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|