From 9be73325a38c3ad5aa2fd8c2b31aa7fd9171fb91 Mon Sep 17 00:00:00 2001
From: Josh Gordon <random-forests@users.noreply.github.com>
Date: Thu, 21 Feb 2019 14:37:00 -0500
Subject: [PATCH] Add files via upload

---
 examples/easy_gpt_demo.ipynb | 184 +++++++++++++++++++++++++++++++++++
 1 file changed, 184 insertions(+)
 create mode 100644 examples/easy_gpt_demo.ipynb

diff --git a/examples/easy_gpt_demo.ipynb b/examples/easy_gpt_demo.ipynb
new file mode 100644
index 0000000..3632d85
--- /dev/null
+++ b/examples/easy_gpt_demo.ipynb
@@ -0,0 +1,184 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "easy-gpt-demo.ipynb",
+      "version": "0.3.2",
+      "provenance": [],
+      "private_outputs": true,
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "accelerator": "GPU"
+  },
+  "cells": [
+    {
+      "metadata": {
+        "id": "nOV0EV6vArTZ",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "!git clone https://github.com/openai/gpt-2.git"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "tjouOxR6BHdq",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "!cd gpt-2 && sh download_model.sh 117M"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "JpsJG2qZBPWh",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "!cd gpt-2 && pip3 install -r requirements.txt"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "H7wLvPqPCfCJ",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "!mv gpt-2/models models"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "Dju6D9MABz1v",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "import sys\n",
+        "import fire\n",
+        "import json\n",
+        "import os\n",
+        "import numpy as np\n",
+        "import tensorflow as tf\n",
+        "import textwrap\n",
+        "\n",
+        "sys.path.insert(0, './gpt-2/src')\n",
+        "import model, sample, encoder, generate_unconditional_samples"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "H8UpxUHVGPPF",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "def interact_model(\n",
+        "    model_name='117M',\n",
+        "    seed=None,\n",
+        "    nsamples=1,\n",
+        "    batch_size=None,\n",
+        "    length=None,\n",
+        "    temperature=1,\n",
+        "    top_k=0,\n",
+        "    raw_text='test',\n",
+        "):\n",
+        "    if batch_size is None:\n",
+        "        batch_size = 1\n",
+        "    assert nsamples % batch_size == 0\n",
+        "    np.random.seed(seed)\n",
+        "    tf.set_random_seed(seed)\n",
+        "\n",
+        "    enc = encoder.get_encoder(model_name)\n",
+        "    hparams = model.default_hparams()\n",
+        "    with open(os.path.join('models', model_name, 'hparams.json')) as f:\n",
+        "        hparams.override_from_dict(json.load(f))\n",
+        "\n",
+        "    if length is None:\n",
+        "        length = hparams.n_ctx // 2\n",
+        "    elif length > hparams.n_ctx:\n",
+        "        raise ValueError(\"Can't get samples longer than window size: %s\" % hparams.n_ctx)\n",
+        "\n",
+        "    with tf.Session(graph=tf.Graph()) as sess:\n",
+        "        context = tf.placeholder(tf.int32, [batch_size, None])\n",
+        "        output = sample.sample_sequence(\n",
+        "            hparams=hparams, length=length,\n",
+        "            context=context,\n",
+        "            batch_size=batch_size,\n",
+        "            temperature=temperature, top_k=top_k\n",
+        "        )\n",
+        "\n",
+        "        saver = tf.train.Saver()\n",
+        "        ckpt = tf.train.latest_checkpoint(os.path.join('models', model_name))\n",
+        "        saver.restore(sess, ckpt)\n",
+        "\n",
+        "        context_tokens = enc.encode(raw_text)\n",
+        "        generated = 0\n",
+        "        for _ in range(nsamples // batch_size):\n",
+        "            out = sess.run(output, feed_dict={\n",
+        "                context: [context_tokens for _ in range(batch_size)]\n",
+        "            })[:, len(context_tokens):]\n",
+        "            for i in range(batch_size):\n",
+        "                generated += 1\n",
+        "                text = enc.decode(out[i])\n",
+        "                print(\"=\" * 40 + \" SAMPLE \" + str(generated) + \" \" + \"=\" * 40)\n",
+        "                print(textwrap.fill(text, 150))\n",
+        "        print(\"=\" * 80)"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    },
+    {
+      "metadata": {
+        "id": "jkrryjyWLZvS",
+        "colab_type": "text"
+      },
+      "cell_type": "markdown",
+      "source": [
+        "### Generate text based on a conditioning text"
+      ]
+    },
+    {
+      "metadata": {
+        "id": "iBQIGgUgG0wh",
+        "colab_type": "code",
+        "colab": {}
+      },
+      "cell_type": "code",
+      "source": [
+        "# Text below from 'The Onion'\n",
+        "interact_model(\n",
+        "    length=90,\n",
+        "    nsamples=3,\n",
+        "    raw_text=\"NEW YORK—At 4:32 p.m. Tuesday, every single resident of New York City decided to evacuate the famed metropolis, having realized it was nothing more than a massive, trash-ridden hellhole that slowly sucks the life out of every one of its inhabitants.\")"
+      ],
+      "execution_count": 0,
+      "outputs": []
+    }
+  ]
+}
\ No newline at end of file