File size: 1,848 Bytes
b51903a 682bee1 b51903a 682bee1 b51903a 682bee1 b51903a 682bee1 b51903a 682bee1 b51903a 682bee1 b51903a 682bee1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 | {
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# SinhalaMMLU — team-riswan, submission #138\n",
"This notebook runs the reconstructed Ollama pipeline from this repository. Read README.md and REPRODUCIBILITY.md first. Fresh model reproduction is unverified; Pass 3 retains historical sampling. The exact submitted predictions stay untouched.\n",
"Provision the original Ollama model/runtime before running the last cell. The model setup currently differs from the organisers' Hugging-Face-only template; approval is outstanding.\n"
],
"id": "team-riswan-0"
},
{
"cell_type": "code",
"metadata": {},
"execution_count": null,
"outputs": [],
"source": [
"%pip install huggingface_hub==2.3.0\n",
"from huggingface_hub import snapshot_download\n",
"from pathlib import Path\n",
"import subprocess, sys\n",
"repo = Path(snapshot_download('SinhalaMMLU-SharedTask/team-riswan', allow_patterns=['*.py', '*.md', '*.txt', 'training/src/*', 'training/ensemble_v9.py']))\n"
],
"id": "team-riswan-1"
},
{
"cell_type": "code",
"metadata": {},
"execution_count": null,
"outputs": [],
"source": [
"# Full model run: expensive. No archived predictions are used.\n",
"run_dir = Path.cwd() / 'rerun'\n",
"run_dir.mkdir(exist_ok=True)\n",
"subprocess.run([sys.executable, str(repo / 'inference.py')], cwd=run_dir, check=True)\n"
],
"id": "team-riswan-2"
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"name": "python",
"version": "3"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|