{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:21.040680Z", "start_time": "2024-05-16T13:58:19.777010Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:17:04.169987Z", "iopub.status.busy": "2024-07-01T02:17:04.169843Z", "iopub.status.idle": "2024-07-01T02:17:05.271262Z", "shell.execute_reply": "2024-07-01T02:17:05.270757Z", "shell.execute_reply.started": "2024-07-01T02:17:04.169970Z" } }, "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", "\n", "import pandas as pd\n", "import numpy as np\n", "import os\n", "from tqdm import tqdm\n", "from sklearn.model_selection import train_test_split\n", "from suno_utils.utils.s3 import download_s3_files\n", "import sys\n", "from collections import defaultdict\n", "from suno_utils.utils.text import (\n", " write_jsonl,\n", " read_jsonl,\n", " write_json,\n", " read_json,\n", ")\n", "import shutil\n", "import ast\n", "from preference_helper import *\n", "from preference_data_preparation_4min_13b_extend import *\n", "\n", "import numpy as np\n", "\n", "pd.set_option('display.max_rows', 500)\n", "pd.set_option('display.max_columns', 500)\n", "pd.set_option('display.width', 1000)" ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:21.082172Z", "start_time": "2024-05-16T13:58:21.041926Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:17:05.272150Z", "iopub.status.busy": "2024-07-01T02:17:05.271954Z", "iopub.status.idle": "2024-07-01T02:17:05.309973Z", "shell.execute_reply": "2024-07-01T02:17:05.309519Z", "shell.execute_reply.started": "2024-07-01T02:17:05.272132Z" } }, "outputs": [], "source": [ "OUT_DATA_DIR = \"/app/suno/data/dpo/13b_extend_mix_cycle/\"\n", "os.makedirs(OUT_DATA_DIR, exist_ok=True)\n", "shutil.copyfile(\"/app/suno/data/dpo/7v_v20_full/tokenizer_60k.json\", os.path.join(OUT_DATA_DIR, \"tokenizer_60k.json\"))\n", "NPZ_DIR = \"/app/suno/data/dpo/13b_npz\"" ] }, { "cell_type": "code", "execution_count": 3, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:53.962528Z", "start_time": "2024-05-16T13:58:21.105919Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:17:05.310648Z", "iopub.status.busy": "2024-07-01T02:17:05.310511Z", "iopub.status.idle": "2024-07-01T02:17:05.974544Z", "shell.execute_reply": "2024-07-01T02:17:05.974005Z", "shell.execute_reply.started": "2024-07-01T02:17:05.310633Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Preference data shape (14129, 82)\n" ] } ], "source": [ "df = pd.read_csv(\"/home/tony/Data/Preference/13b_v0/interesting_clips_20240624_mix_selected_pos.csv\") # , engine='python')\n", "print(\"Preference data shape\", df.shape)" ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:17:05.975315Z", "iopub.status.busy": "2024-07-01T02:17:05.975170Z", "iopub.status.idle": "2024-07-01T02:17:05.995323Z", "shell.execute_reply": "2024-07-01T02:17:05.994893Z", "shell.execute_reply.started": "2024-07-01T02:17:05.975300Z" } }, "outputs": [ { "data": { "text/plain": [ "is_13b\n", "True 14129\n", "Name: count, dtype: int64" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df[\"is_13b\"] = df[\"model_name\"].str.contains(\"v3p5\")\n", "df[\"is_13b\"].value_counts()" ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:17:05.996168Z", "iopub.status.busy": "2024-07-01T02:17:05.996031Z", "iopub.status.idle": "2024-07-01T02:17:06.042712Z", "shell.execute_reply": "2024-07-01T02:17:06.042251Z", "shell.execute_reply.started": "2024-07-01T02:17:05.996153Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(28258, 82)\n" ] } ], "source": [ "df_negative = df.copy()\n", "df_negative[\"preference\"] = False\n", "df_negative[\"s3_id\"] += \"_gen_cycle\"\n", "df = pd.concat([df, df_negative])\n", "print(df.shape)" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:56.199480Z", "start_time": "2024-05-16T13:58:53.963687Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:17:06.044400Z", "iopub.status.busy": "2024-07-01T02:17:06.044250Z", "iopub.status.idle": "2024-07-01T02:18:52.035053Z", "shell.execute_reply": "2024-07-01T02:18:52.034480Z", "shell.execute_reply.started": "2024-07-01T02:17:06.044385Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "4829780\n", "4829780\n", "pre-downloaded df (28258, 82)\n", "downloaded df (28216, 82)\n" ] } ], "source": [ "converted_paths = set(os.listdir(NPZ_DIR))\n", "print(len(converted_paths))\n", "\n", "converted_paths = set([f.replace(\".npz\", \"\") for f in converted_paths])\n", "print(len(converted_paths))\n", "\n", "print(\"pre-downloaded df\", df.shape)\n", "df[df[\"s3_id\"].isin(converted_paths)].shape\n", "df = df[df[\"s3_id\"].isin(converted_paths)].copy()\n", "print(\"downloaded df\", df.shape)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:18:52.035915Z", "iopub.status.busy": "2024-07-01T02:18:52.035753Z", "iopub.status.idle": "2024-07-01T02:18:52.341742Z", "shell.execute_reply": "2024-07-01T02:18:52.341245Z", "shell.execute_reply.started": "2024-07-01T02:18:52.035898Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(28216, 82)\n", "(28174, 82)\n", "preference model_name \n", "False chirp-v3p5-engine-s-2 10638\n", " chirp-v3p5-engine-upload 3449\n", "True chirp-v3p5-engine-s-2 10638\n", " chirp-v3p5-engine-upload 3449\n", "Name: count, dtype: int64\n" ] } ], "source": [ "print(df.shape)\n", "df = df[\n", " df[\"request_id\"].isin(\n", " df[\"request_id\"].value_counts().index[df[\"request_id\"].value_counts() == 2]\n", " )\n", "]\n", "print(df.shape)\n", "print(df.groupby([\"preference\"])[\"model_name\"].value_counts())\n", "assert df.shape[0] == df[\"request_id\"].nunique() * 2" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# LET's do the data prep" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Need to kick out the ones has gpt prompt -- these are pairs with different text inputs" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:18:52.342645Z", "iopub.status.busy": "2024-07-01T02:18:52.342502Z", "iopub.status.idle": "2024-07-01T02:18:52.364415Z", "shell.execute_reply": "2024-07-01T02:18:52.363980Z", "shell.execute_reply.started": "2024-07-01T02:18:52.342629Z" } }, "outputs": [], "source": [ "df_slice = df.copy()" ] }, { "cell_type": "code", "execution_count": 9, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:18:52.365116Z", "iopub.status.busy": "2024-07-01T02:18:52.364978Z", "iopub.status.idle": "2024-07-01T02:18:52.397689Z", "shell.execute_reply": "2024-07-01T02:18:52.397272Z", "shell.execute_reply.started": "2024-07-01T02:18:52.365101Z" } }, "outputs": [ { "data": { "text/plain": [ "0 NaN\n", "1 NaN\n", "2 59.24\n", "3 NaN\n", "4 16.32\n", " ... \n", "14124 NaN\n", "14125 NaN\n", "14126 NaN\n", "14127 8.00\n", "14128 NaN\n", "Name: continue_at, Length: 28174, dtype: float64" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_slice[\"continue_at\"]" ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.932296Z", "start_time": "2024-05-16T13:59:41.932287Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:52.398405Z", "iopub.status.busy": "2024-07-01T02:18:52.398269Z", "iopub.status.idle": "2024-07-01T02:18:52.441065Z", "shell.execute_reply": "2024-07-01T02:18:52.440662Z", "shell.execute_reply.started": "2024-07-01T02:18:52.398390Z" } }, "outputs": [ { "data": { "text/plain": [ "9971" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# don't have continue at\n", "df_slice[df_slice[\"continue_at\"].isna()][\"request_id\"].nunique()" ] }, { "cell_type": "code", "execution_count": 11, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.933558Z", "start_time": "2024-05-16T13:59:41.933550Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:52.441772Z", "iopub.status.busy": "2024-07-01T02:18:52.441634Z", "iopub.status.idle": "2024-07-01T02:18:52.465754Z", "shell.execute_reply": "2024-07-01T02:18:52.465337Z", "shell.execute_reply.started": "2024-07-01T02:18:52.441757Z" } }, "outputs": [], "source": [ "# df_slice.to_csv(\"/home/tony/Data/Preference/13b_v0/interesting_clips_20240624_mix_selected_pos_cycled.csv\", index=False)" ] }, { "cell_type": "code", "execution_count": 12, "metadata": { "execution": { "iopub.execute_input": "2024-07-01T02:18:52.466436Z", "iopub.status.busy": "2024-07-01T02:18:52.466296Z", "iopub.status.idle": "2024-07-01T02:18:52.503505Z", "shell.execute_reply": "2024-07-01T02:18:52.503068Z", "shell.execute_reply.started": "2024-07-01T02:18:52.466422Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "14087\n" ] } ], "source": [ "final_filtered_requests = df_slice[\"request_id\"].unique()\n", "print(len(final_filtered_requests))" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.934277Z", "start_time": "2024-05-16T13:59:41.934268Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:52.504189Z", "iopub.status.busy": "2024-07-01T02:18:52.504056Z", "iopub.status.idle": "2024-07-01T02:18:52.594855Z", "shell.execute_reply": "2024-07-01T02:18:52.594403Z", "shell.execute_reply.started": "2024-07-01T02:18:52.504175Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "13946 141\n", "(27892, 83) (282, 83)\n" ] } ], "source": [ "train_requests, val_requests = train_test_split(\n", " sorted(list(final_filtered_requests)), test_size=0.01, random_state=42\n", ")\n", "print(len(train_requests), len(val_requests))\n", "df_slice = df_slice.sort_values(by=[\"request_id\", \"preference\"])\n", "train_df = df_slice[df_slice[\"request_id\"].isin(set(train_requests))].copy()\n", "val_df = df_slice[df_slice[\"request_id\"].isin(set(val_requests))].copy()\n", "train_df = train_df.sort_values(by=[\"request_id\", \"preference\"])\n", "train_df = train_df.reset_index()\n", "val_df = val_df.sort_values(by=[\"request_id\", \"preference\"])\n", "val_df = val_df.reset_index()\n", "\n", "print(train_df.shape, val_df.shape)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Actually make" ] }, { "cell_type": "code", "execution_count": 14, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.934954Z", "start_time": "2024-05-16T13:59:41.934946Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:52.595595Z", "iopub.status.busy": "2024-07-01T02:18:52.595455Z", "iopub.status.idle": "2024-07-01T02:18:52.607439Z", "shell.execute_reply": "2024-07-01T02:18:52.607012Z", "shell.execute_reply.started": "2024-07-01T02:18:52.595580Z" } }, "outputs": [], "source": [ "# val_df[[\"request_id\", \"metadata\", \"updated_at\", \"user_id\", \"preference\"]].head()" ] }, { "cell_type": "code", "execution_count": 15, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.935620Z", "start_time": "2024-05-16T13:59:41.935613Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:52.608116Z", "iopub.status.busy": "2024-07-01T02:18:52.607979Z", "iopub.status.idle": "2024-07-01T02:18:53.487803Z", "shell.execute_reply": "2024-07-01T02:18:53.487280Z", "shell.execute_reply.started": "2024-07-01T02:18:52.608102Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████| 27892/27892 [00:00<00:00, 33129.81it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "1,364 hours of 27892 clips, 2.490357142857143 nodes\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "total_duration = 0\n", "for i, row in tqdm.tqdm(train_df.iterrows(), total=len(train_df)):\n", " # we need to alternate between preference: neg, pos\n", " # print(i, row)\n", " try:\n", " assert row[\"preference\"] == (i % 2 == 1)\n", " except:\n", " print(i, row)\n", " total_duration += row[\"duration\"]\n", "print(f\"{round(total_duration / 60 / 60):,} hours of {train_df.shape[0]} clips, {train_df.shape[0] / 8 / 2 / 700} nodes\")" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.936268Z", "start_time": "2024-05-16T13:59:41.936260Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:53.488581Z", "iopub.status.busy": "2024-07-01T02:18:53.488434Z", "iopub.status.idle": "2024-07-01T02:18:56.658350Z", "shell.execute_reply": "2024-07-01T02:18:56.657815Z", "shell.execute_reply.started": "2024-07-01T02:18:53.488564Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████| 282/282 [00:03<00:00, 91.44it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Total 282 clips\n", "7 hours of False\n", "7 hours of True\n", "Done\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "make_dataset(val_df, OUT_DATA_DIR, is_val=True, npz_dir=NPZ_DIR)" ] }, { "cell_type": "code", "execution_count": 17, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.936964Z", "start_time": "2024-05-16T13:59:41.936957Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:18:56.659168Z", "iopub.status.busy": "2024-07-01T02:18:56.659018Z", "iopub.status.idle": "2024-07-01T02:25:39.622864Z", "shell.execute_reply": "2024-07-01T02:25:39.622314Z", "shell.execute_reply.started": "2024-07-01T02:18:56.659152Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████| 27892/27892 [06:42<00:00, 69.23it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Total 27892 clips\n", "711 hours of False\n", "711 hours of True\n", "Done\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "make_dataset(train_df, OUT_DATA_DIR, is_val=False, npz_dir=NPZ_DIR)" ] }, { "cell_type": "markdown", "metadata": { "ExecuteTime": { "end_time": "2024-01-29T19:46:47.549860Z", "start_time": "2024-01-29T19:46:47.548015Z" } }, "source": [ "# Validation" ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.937879Z", "start_time": "2024-05-16T13:59:41.937870Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:39.623734Z", "iopub.status.busy": "2024-07-01T02:25:39.623576Z", "iopub.status.idle": "2024-07-01T02:25:39.993789Z", "shell.execute_reply": "2024-07-01T02:25:39.993313Z", "shell.execute_reply.started": "2024-07-01T02:25:39.623718Z" } }, "outputs": [], "source": [ "# verify\n", "mm = np.memmap(os.path.join(OUT_DATA_DIR, f\"data_val.bin\"), dtype=np.uint16, mode=\"r\")\n", "test_metas = read_jsonl(os.path.join(OUT_DATA_DIR, f\"meta_val.jsonl\"))\n", "test_info = read_json(os.path.join(OUT_DATA_DIR, f\"info_val.json\"))\n", "mm = mm.reshape(-1, 6016, 13)\n", "assert len(mm) == len(test_metas)\n", "assert mm[:100, :, 0].min() >= 0\n", "assert mm[:100, :, 0].max() <= 4000\n", "assert mm[:100, :, 1:].min() >= 0\n", "assert mm[:100, :, 1:].max() <= 2048" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.938629Z", "start_time": "2024-05-16T13:59:41.938621Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:39.994543Z", "iopub.status.busy": "2024-07-01T02:25:39.994399Z", "iopub.status.idle": "2024-07-01T02:25:40.008485Z", "shell.execute_reply": "2024-07-01T02:25:40.008072Z", "shell.execute_reply.started": "2024-07-01T02:25:39.994527Z" } }, "outputs": [], "source": [ "# # randomly listen to some stuff\n", "# from suno_utils.tasks.dac_2c_12cb import preload_models as preload_codec_models\n", "# from suno_utils.tasks.dac_2c_12cb import (\n", "# encode as codec_encode,\n", "# decode_stream_to_full_audio as codec_decode,\n", "# EMBEDDING_RATE as CODEC_EMBEDDING_RATE,\n", "# decode as decode\n", "# )\n", "# os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"0\"\n", "# _ = preload_codec_models(\"/app/suno/data/dpo/models/dac_2c_25x12.pt\", device=\"cuda\")\n", "# assert len(test_metas) == len(mm)\n", "# idx_list = list(range(len(test_metas)))\n", "# # random.shuffle(idx_list)\n", "# # idx_list = [idx for idx in idx_list if \"text\" in test_metas[idx]]\n", "# print(len(mm))" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.939205Z", "start_time": "2024-05-16T13:59:41.939198Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.009181Z", "iopub.status.busy": "2024-07-01T02:25:40.009045Z", "iopub.status.idle": "2024-07-01T02:25:40.047200Z", "shell.execute_reply": "2024-07-01T02:25:40.046786Z", "shell.execute_reply.started": "2024-07-01T02:25:40.009167Z" } }, "outputs": [], "source": [ "# import random\n", "# idx = random.choice(test_info[\"perference_0\"][\"idx_list\"])\n", "# assert \"original_duration_s\" in test_metas[idx]\n", "# # positive index should be shifted by 1\n", "# pos_idx = idx + 1\n", "# print(\n", "# \"tags:\",\n", "# test_metas[idx].get(\"tags\") == test_metas[pos_idx].get(\"tags\"),\n", "# test_metas[idx].get(\"tags\"),\n", "# )\n", "# arr = mm[idx, 1:].copy().astype(np.int16)[:, 1:]\n", "# pos_arr = mm[pos_idx, 1:].copy().astype(np.int16)[:, 1:]\n", "# pad_idx_arr = np.where(arr == COARSE_PAD_TOKEN)[0]\n", "# if len(pad_idx_arr) > 0:\n", "# arr = arr[: pad_idx_arr[0], :]\n", "# pos_pad_idx_arr = np.where(pos_arr == COARSE_PAD_TOKEN)[0]\n", "# if len(pos_pad_idx_arr) > 0:\n", "# pos_arr = pos_arr[: pos_pad_idx_arr[0], :]\n", "# a = decode(arr)\n", "# print(\"\\n negative example \\n\", test_metas[idx])\n", "# a.play(compress=False)\n", "# pos_a = decode(pos_arr)\n", "# print(\"\\n positive example \\n\", test_metas[pos_idx])\n", "# pos_a.play(compress=False)\n", "# print(\n", "# \"text:\",\n", "# test_metas[idx].get(\"text\") == test_metas[pos_idx].get(\"text\"),\n", "# test_metas[idx].get(\"text\"),\n", "# )" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.939977Z", "start_time": "2024-05-16T13:59:41.939969Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.049283Z", "iopub.status.busy": "2024-07-01T02:25:40.049134Z", "iopub.status.idle": "2024-07-01T02:25:40.083787Z", "shell.execute_reply": "2024-07-01T02:25:40.083369Z", "shell.execute_reply.started": "2024-07-01T02:25:40.049267Z" } }, "outputs": [], "source": [ "# val_df[val_df[\"tags\"] == 'a vibrant blend of experimental jazz fusion, drum-and-bass and swagger fuzzed-out guitars']" ] }, { "cell_type": "code", "execution_count": 22, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.940610Z", "start_time": "2024-05-16T13:59:41.940603Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.084468Z", "iopub.status.busy": "2024-07-01T02:25:40.084335Z", "iopub.status.idle": "2024-07-01T02:25:40.121440Z", "shell.execute_reply": "2024-07-01T02:25:40.121017Z", "shell.execute_reply.started": "2024-07-01T02:25:40.084453Z" } }, "outputs": [], "source": [ "# from collections import Counter\n", "# c = Counter()\n", "# for _, row in df_slice.iterrows():\n", "# # print(row[\"metadata\"])\n", "# for k in ast.literal_eval(row[\"metadata\"]).keys():\n", "# c[k] += 1" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.941167Z", "start_time": "2024-05-16T13:59:41.941159Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.122115Z", "iopub.status.busy": "2024-07-01T02:25:40.121980Z", "iopub.status.idle": "2024-07-01T02:25:40.160666Z", "shell.execute_reply": "2024-07-01T02:25:40.160247Z", "shell.execute_reply.started": "2024-07-01T02:25:40.122099Z" } }, "outputs": [], "source": [ "# original_npz_path = f\"/app/suno/data/dpo/7b_npz/{test_metas[idx]['id']}.npz\"\n", "# original_npz_path = \"/app/suno/data/dpo/7b_npz/729c3011-f672-4ccd-8d82-1cbf2b52ff69.npz\"\n", "# original_arr = np.load(original_npz_path)[\"v2_raw\"]" ] }, { "cell_type": "code", "execution_count": 24, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.941801Z", "start_time": "2024-05-16T13:59:41.941793Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.161442Z", "iopub.status.busy": "2024-07-01T02:25:40.161304Z", "iopub.status.idle": "2024-07-01T02:25:40.194876Z", "shell.execute_reply": "2024-07-01T02:25:40.194428Z", "shell.execute_reply.started": "2024-07-01T02:25:40.161427Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "141 0\n" ] } ], "source": [ "def validation_on_metas(input_metas):\n", "\n", " total_bad = 0\n", " total_good = 0\n", " for idx in range(len(input_metas)):\n", " if idx % 2 == 0:\n", " pos_idx = idx + 1\n", " if input_metas[idx].get(\"tags\") != input_metas[pos_idx].get(\"tags\"):\n", " # print(test_metas[idx].get(\"text\") == test_metas[pos_idx].get(\"text\"), test_metas[idx].get(\"tags\"), test_metas[pos_idx].get(\"tags\"))\n", " total_bad += 1\n", " else:\n", " total_good += 1\n", " print(total_good, total_bad)\n", " return\n", "\n", "validation_on_metas(test_metas)" ] }, { "cell_type": "code", "execution_count": 25, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.942520Z", "start_time": "2024-05-16T13:59:41.942511Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.195670Z", "iopub.status.busy": "2024-07-01T02:25:40.195535Z", "iopub.status.idle": "2024-07-01T02:25:40.228981Z", "shell.execute_reply": "2024-07-01T02:25:40.228547Z", "shell.execute_reply.started": "2024-07-01T02:25:40.195655Z" } }, "outputs": [], "source": [ "train_info = read_json(os.path.join(OUT_DATA_DIR, f\"info_tr.json\"))" ] }, { "cell_type": "code", "execution_count": 26, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.943072Z", "start_time": "2024-05-16T13:59:41.943065Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.229684Z", "iopub.status.busy": "2024-07-01T02:25:40.229541Z", "iopub.status.idle": "2024-07-01T02:25:40.260840Z", "shell.execute_reply": "2024-07-01T02:25:40.260418Z", "shell.execute_reply.started": "2024-07-01T02:25:40.229669Z" } }, "outputs": [], "source": [ "n_neg_tr = train_info[\"perference_0\"][\"idx_list\"]\n", "n_pos_tr = train_info[\"perference_1\"][\"idx_list\"]\n", "assert len(n_pos_tr) == len(n_neg_tr)" ] }, { "cell_type": "code", "execution_count": 27, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.944246Z", "start_time": "2024-05-16T13:59:41.944237Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.261540Z", "iopub.status.busy": "2024-07-01T02:25:40.261402Z", "iopub.status.idle": "2024-07-01T02:25:40.297326Z", "shell.execute_reply": "2024-07-01T02:25:40.296889Z", "shell.execute_reply.started": "2024-07-01T02:25:40.261525Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "total samples 27892 (27892, 83)\n" ] } ], "source": [ "total_iters = (len(n_neg_tr) + len(n_pos_tr))\n", "print(\"total samples\", total_iters, train_df.shape)" ] }, { "cell_type": "code", "execution_count": 34, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.945249Z", "start_time": "2024-05-16T13:59:41.945241Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:29:27.098138Z", "iopub.status.busy": "2024-07-01T02:29:27.097849Z", "iopub.status.idle": "2024-07-01T02:29:27.111866Z", "shell.execute_reply": "2024-07-01T02:29:27.111416Z", "shell.execute_reply.started": "2024-07-01T02:29:27.098121Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1 epoch per batch 4, total 217.90625\n" ] } ], "source": [ "print(\"1 epoch per batch 4, total\", total_iters / 8 / 4 / 4)" ] }, { "cell_type": "code", "execution_count": 29, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.945972Z", "start_time": "2024-05-16T13:59:41.945964Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.334016Z", "iopub.status.busy": "2024-07-01T02:25:40.333879Z", "iopub.status.idle": "2024-07-01T02:25:40.365682Z", "shell.execute_reply": "2024-07-01T02:25:40.365264Z", "shell.execute_reply.started": "2024-07-01T02:25:40.334001Z" } }, "outputs": [], "source": [ "# !cd /home/tony/Work/tony/slurm/extend/ && sbatch sbatch_ipo_13b_extend" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# some gymathtics loading prev data" ] }, { "cell_type": "code", "execution_count": 30, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.946562Z", "start_time": "2024-05-16T13:59:41.946555Z" }, "execution": { "iopub.execute_input": "2024-07-01T02:25:40.366384Z", "iopub.status.busy": "2024-07-01T02:25:40.366247Z", "iopub.status.idle": "2024-07-01T02:25:40.403935Z", "shell.execute_reply": "2024-07-01T02:25:40.403523Z", "shell.execute_reply.started": "2024-07-01T02:25:40.366370Z" } }, "outputs": [], "source": [ "# prev_v3_data = \"/app/suno/data/dpo/7v_v20_full/\"\n", "\n", "# test_val_metas = read_jsonl(os.path.join(prev_v3_data, f\"meta_val.jsonl\"))\n", "# test_tr_metas = read_jsonl(os.path.join(prev_v3_data, f\"meta_tr.jsonl\"))\n", "\n", "# all_ids = set()\n", "# for meta in test_val_metas:\n", "# all_ids.add(meta[\"id\"])\n", "# for meta in test_tr_metas:\n", "# all_ids.add(meta[\"id\"])\n", "# print(len(all_ids), len(test_val_metas) + len(test_tr_metas))\n", "\n", "# all_ids = list(all_ids)\n", "# with open(\"/home/tony/Data/Preference/7b_v2/7v_v20_full_recut_id.json\", \"w\") as fp:\n", "# json.dump(all_ids, fp)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.14" }, "toc": { "base_numbering": 1, "nav_menu": {}, "number_sections": true, "sideBar": true, "skip_h1_title": false, "title_cell": "Table of Contents", "title_sidebar": "Contents", "toc_cell": false, "toc_position": {}, "toc_section_display": true, "toc_window_display": false } }, "nbformat": 4, "nbformat_minor": 4 }