Upload sakthai-1.5b-t4.ipynb with huggingface_hub

sakthai-1.5b-t4.ipynb

@@ -0,0 +1,69 @@
+{
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "# SakThai Plus 1.5B v11 - Free T4 Training\n",
+        "\n",
+        "Runs the same QLoRA+rsLoRA recipe as the HF Job version, but on the **free Kaggle T4 GPU** (30h/week).\n",
+        "Dataset: `sakthai-combined-v11` (2,965 rows, bench-aligned tool schemas).\n",
+        "\n",
+        "**Setup before running:**\n",
+        "1. Upload `train-sakthai-1.5b-kaggle.py` to the notebook (Add Input > Upload as file, or paste into a cell with `%%writefile`)\n",
+        "2. Add `HF_TOKEN` as a Kaggle Secret (Settings > Add Secret, name `HF_TOKEN`)\n",
+        "3. Select Accelerator: `GPU T4 x2` (or T4 x1)\n",
+        "4. Run all cells\n",
+        "\n",
+        "Output: adapter `Nanthasit/sakthai-plus-1.5b-lora` + merged `Nanthasit/sakthai-plus-1.5b`"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from kaggle_secrets import UserSecretsClient\n",
+        "import os\n",
+        "os.environ['HF_TOKEN'] = UserSecretsClient().get_secret('HF_TOKEN')\n",
+        "print('HF_TOKEN set:', bool(os.environ['HF_TOKEN']))"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "!pip install -q transformers trl peft datasets accelerate bitsandbytes huggingface_hub\n",
+        "!pip show trl | head -1"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "!python train-sakthai-1.5b-kaggle.py"
+      ]
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {"provenance": []},
+    "kaggle": {
+      "accelerator": "GPU T4 x2",
+      "dataSources": [],
+      "kernelType": "Notebook",
+      "language": "python"
+    },
+    "language_info": {
+      "name": "python",
+      "version": "3.11"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}