{ "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": "2024-08-02T05:05:38.205441Z", "iopub.status.busy": "2024-08-02T05:05:38.205008Z", "iopub.status.idle": "2024-08-02T05:05:40.188955Z", "shell.execute_reply": "2024-08-02T05:05:40.188337Z", "shell.execute_reply.started": "2024-08-02T05:05:38.205412Z" } }, "outputs": [], "source": [ "%load_ext autoreload\n", "%autoreload 2\n", "\n", "import ast\n", "import os\n", "import shutil\n", "import sys\n", "from collections import defaultdict\n", "\n", "import numpy as np\n", "import pandas as pd\n", "from preference_data_preparation_4min_13b_extend import *\n", "from preference_helper import *\n", "from sklearn.model_selection import train_test_split\n", "from suno_utils.utils.s3 import download_s3_files\n", "from suno_utils.utils.text import read_json, read_jsonl, write_json, write_jsonl\n", "from tqdm import tqdm\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": "2024-08-02T05:05:40.268439Z", "iopub.status.busy": "2024-08-02T05:05:40.268210Z", "iopub.status.idle": "2024-08-02T05:05:40.329591Z", "shell.execute_reply": "2024-08-02T05:05:40.329028Z", "shell.execute_reply.started": "2024-08-02T05:05:40.268421Z" } }, "outputs": [], "source": [ "OUT_DATA_DIR = \"/app/suno/data/dpo/13b_feedback_v7/\"\n", "os.makedirs(OUT_DATA_DIR, exist_ok=True)\n", "shutil.copyfile(\n", " \"/app/suno/data/chirp_v4/multi/tokenizer_60k.json\",\n", " os.path.join(OUT_DATA_DIR, \"tokenizer_60k.json\"),\n", ")\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": "2024-08-02T05:05:45.500474Z", "iopub.status.busy": "2024-08-02T05:05:45.500113Z", "iopub.status.idle": "2024-08-02T05:05:46.558123Z", "shell.execute_reply": "2024-08-02T05:05:46.557389Z", "shell.execute_reply.started": "2024-08-02T05:05:45.500454Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Preference data shape (30994, 47)\n" ] } ], "source": [ "df = pd.read_csv(\n", " \"/home/tony/Data/Preference/13b_v0/interesting_clips_20240802_feedback.csv\"\n", ") # , engine='python')\n", "print(\"Preference data shape\", df.shape)" ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:56.199480Z", "start_time": "2024-05-16T13:58:53.963687Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:05:46.559485Z", "iopub.status.busy": "2024-08-02T05:05:46.559302Z", "iopub.status.idle": "2024-08-02T05:07:51.747145Z", "shell.execute_reply": "2024-08-02T05:07:51.746542Z", "shell.execute_reply.started": "2024-08-02T05:05:46.559466Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "5379938\n", "5379938\n", "pre-downloaded df (30994, 47)\n", "downloaded df (30039, 47)\n" ] } ], "source": [ "converted_paths = 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": 5, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:56.467253Z", "start_time": "2024-05-16T13:58:56.207647Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:07:51.748301Z", "iopub.status.busy": "2024-08-02T05:07:51.748129Z", "iopub.status.idle": "2024-08-02T05:07:52.356820Z", "shell.execute_reply": "2024-08-02T05:07:52.356340Z", "shell.execute_reply.started": "2024-08-02T05:07:51.748283Z" } }, "outputs": [ { "data": { "text/plain": [ "is_13b\n", "True 30026\n", "False 13\n", "Name: count, dtype: int64" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df[\"is_13b\"] = df[\"model_name\"].str.contains(\"v3p5\")\n", "df[\"is_13b\"].value_counts()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# LET's do the data prep" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:07:52.357773Z", "iopub.status.busy": "2024-08-02T05:07:52.357620Z", "iopub.status.idle": "2024-08-02T05:07:52.395407Z", "shell.execute_reply": "2024-08-02T05:07:52.394937Z", "shell.execute_reply.started": "2024-08-02T05:07:52.357757Z" } }, "outputs": [], "source": [ "# this is a big hack but for feedback preference we need to make it compatible with before\n", "df[\"preference\"] = df[\"feedback_preference\"]" ] }, { "cell_type": "code", "execution_count": 7, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:56.592883Z", "start_time": "2024-05-16T13:58:56.470781Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:07:52.396596Z", "iopub.status.busy": "2024-08-02T05:07:52.396323Z", "iopub.status.idle": "2024-08-02T05:07:52.435630Z", "shell.execute_reply": "2024-08-02T05:07:52.435163Z", "shell.execute_reply.started": "2024-08-02T05:07:52.396579Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "preference model_name \n", "False chirp-v3p5-engine-s-8 11849\n", " chirp-v3p5-engine-upload-4 1487\n", " chirp-v3p5-engine-t-1 568\n", " chirp-v3p5-engine-ft 443\n", " chirp-v3p5-engine-ft-1 252\n", " chirp-v3p5-engine-upload 182\n", " chirp-v3p5-engine-s-14 62\n", " chirp-v3p5-engine-ft-5 59\n", " chirp-v3p5-engine-s-8-no-top-p 18\n", " chirp-v3p5-engine-t-1-7 15\n", " chirp-v3p5-engine-s-20 13\n", " chirp-v3p5-engine-ft-4 9\n", " chirp-v3p5-engine-s-8-paged 8\n", " chirp-v3p5-engine-ft-6 7\n", " chirp-v3p5-engine-t-1-5 7\n", " chirp-v3p5-engine-t-1-fast-sem 7\n", " chirp-v3p5-engine-ft-3 6\n", " chirp-v3p5-engine-ft-2 5\n", " chirp-v3p5-engine-t-1-fast 5\n", " chirp-v3p5-engine-s-19 4\n", " chirp-v3p5-engine-t 4\n", " chirp-v3-engine-i 3\n", " chirp-v3p5-engine-s-18 2\n", " chirp-v3p5-engine-b 1\n", "True chirp-v3p5-engine-s-8 11904\n", " chirp-v3p5-engine-upload-4 1488\n", " chirp-v3p5-engine-t-1 566\n", " chirp-v3p5-engine-ft 439\n", " chirp-v3p5-engine-ft-1 226\n", " chirp-v3p5-engine-upload 181\n", " chirp-v3p5-engine-ft-5 68\n", " chirp-v3p5-engine-s-14 38\n", " chirp-v3p5-engine-s-20 21\n", " chirp-v3p5-engine-ft-4 16\n", " chirp-v3p5-engine-s-8-no-top-p 14\n", " chirp-v3p5-engine-ft-3 9\n", " chirp-v3p5-engine-t-1-7 9\n", " chirp-v3-engine-i 8\n", " chirp-v3p5-engine-s-8-paged 8\n", " chirp-v3p5-engine-t-1-5 5\n", " chirp-v3p5-engine-t-1-fast-sem 5\n", " chirp-v3p5-engine-ft-6 4\n", " chirp-v3p5-engine-s-19 4\n", " chirp-v3p5-engine-t-1-fast 4\n", " chirp-v2-xxl-alpha 2\n", " chirp-v3p5-engine-b 1\n", " chirp-v3p5-engine-ft-2 1\n", " chirp-v3p5-engine-s-18 1\n", " chirp-v3p5-engine-t 1\n", "Name: count, dtype: int64\n", "(30039, 48)\n" ] } ], "source": [ "## for 13b this is easy for now\n", "print(df.groupby([\"preference\"])[\"model_name\"].value_counts())\n", "print(df.shape)" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:21.936897Z", "iopub.status.busy": "2024-08-02T05:08:21.936544Z", "iopub.status.idle": "2024-08-02T05:08:21.966956Z", "shell.execute_reply": "2024-08-02T05:08:21.966446Z", "shell.execute_reply.started": "2024-08-02T05:08:21.936878Z" } }, "outputs": [], "source": [ "df = df[df[\"model_name\"] == \"chirp-v3p5-engine-s-8\"].copy()\n", "# df = df[~df[\"model_name\"].str.contains(\"engine-t\")].copy()" ] }, { "cell_type": "code", "execution_count": 9, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:58:56.909539Z", "start_time": "2024-05-16T13:58:56.595736Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:23.475846Z", "iopub.status.busy": "2024-08-02T05:08:23.475558Z", "iopub.status.idle": "2024-08-02T05:08:23.512249Z", "shell.execute_reply": "2024-08-02T05:08:23.511741Z", "shell.execute_reply.started": "2024-08-02T05:08:23.475827Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(23753, 48)\n", "(23484, 48)\n", "preference model_name \n", "False chirp-v3p5-engine-s-8 11742\n", "True chirp-v3p5-engine-s-8 11742\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": "code", "execution_count": 10, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:36.043975Z", "start_time": "2024-05-16T13:58:56.910958Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:26.796792Z", "iopub.status.busy": "2024-08-02T05:08:26.796471Z", "iopub.status.idle": "2024-08-02T05:08:33.459877Z", "shell.execute_reply": "2024-08-02T05:08:33.459288Z", "shell.execute_reply.started": "2024-08-02T05:08:26.796773Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "unique_requests 11742\n" ] } ], "source": [ "# Let's use the old selection for now -- for quality assurance\n", "# expand the metadata columns -- this takes forever...~ 6 mins\n", "test_slice = df[\"metadata\"].apply(lambda x: ast.literal_eval(x))\n", "test_slice_series = test_slice.apply(pd.Series)\n", "df = pd.concat([df, test_slice_series], axis=1, join=\"inner\")\n", "print(\"unique_requests\", df[\"request_id\"].nunique())" ] }, { "cell_type": "code", "execution_count": 11, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:34.056300Z", "iopub.status.busy": "2024-08-02T05:08:34.056034Z", "iopub.status.idle": "2024-08-02T05:08:36.721280Z", "shell.execute_reply": "2024-08-02T05:08:36.720688Z", "shell.execute_reply.started": "2024-08-02T05:08:34.056282Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "pre-downloaded audio_prompt df (23484, 79)\n", "(3680, 79)\n", "downloaded df (23484, 79)\n" ] } ], "source": [ "print(\"pre-downloaded audio_prompt df\", df.shape)\n", "print(df[df[\"audio_prompt_id\"].isin(converted_paths)].shape)\n", "# df = df[df[\"audio_prompt_id\"].isin(converted_paths)].copy()\n", "print(\"downloaded df\", df.shape)" ] }, { "cell_type": "code", "execution_count": 12, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:36.393047Z", "start_time": "2024-05-16T13:59:36.048831Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:36.722312Z", "iopub.status.busy": "2024-08-02T05:08:36.722149Z", "iopub.status.idle": "2024-08-02T05:08:36.741631Z", "shell.execute_reply": "2024-08-02T05:08:36.741152Z", "shell.execute_reply.started": "2024-08-02T05:08:36.722295Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "unique_requests 11742\n" ] } ], "source": [ "# GPT requests are also fine for now\n", "print(\"unique_requests\", df[\"request_id\"].nunique())" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:37.316022Z", "iopub.status.busy": "2024-08-02T05:08:37.315861Z", "iopub.status.idle": "2024-08-02T05:08:37.365082Z", "shell.execute_reply": "2024-08-02T05:08:37.364591Z", "shell.execute_reply.started": "2024-08-02T05:08:37.316005Z" } }, "outputs": [], "source": [ "df = df.loc[:,~df.columns.duplicated()].copy()" ] }, { "cell_type": "code", "execution_count": 14, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:38.540440Z", "iopub.status.busy": "2024-08-02T05:08:38.540030Z", "iopub.status.idle": "2024-08-02T05:08:38.556203Z", "shell.execute_reply": "2024-08-02T05:08:38.555743Z", "shell.execute_reply.started": "2024-08-02T05:08:38.540421Z" } }, "outputs": [], "source": [ "df[\"total_start_s\"] = 0\n", "df[\"total_clip_s\"] = 0" ] }, { "cell_type": "code", "execution_count": 15, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:40.799375Z", "start_time": "2024-05-16T13:59:36.394236Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:41.620640Z", "iopub.status.busy": "2024-08-02T05:08:41.620394Z", "iopub.status.idle": "2024-08-02T05:08:41.776946Z", "shell.execute_reply": "2024-08-02T05:08:41.776450Z", "shell.execute_reply.started": "2024-08-02T05:08:41.620622Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1110\n", "good_continue_at\n", "True 23335\n", "False 149\n", "Name: count, dtype: int64\n", "\n", " Check some basics... \n", " preference\n", "False 11742\n", "True 11742\n", "Name: count, dtype: int64 is_13b\n", "True 23484\n", "Name: count, dtype: int64 model_name\n", "chirp-v3p5-engine-s-8 23484\n", "Name: count, dtype: int64 preference model_name \n", "False chirp-v3p5-engine-s-8 11742\n", "True chirp-v3p5-engine-s-8 11742\n", "Name: count, dtype: int64\n", "(23484, 81)\n" ] } ], "source": [ "# get the original duration of the clips, if they are concacted\n", "df[\"original_duration_s\"] = df[\"total_start_s\"] + df[\"duration\"]\n", "# classify the continue at behavoirs by the duration choice\n", "audio_prompt_id_to_continue_at = {}\n", "for _, row in df[~df[\"audio_prompt_id\"].isna()].iterrows():\n", " audio_prompt_id = row[\"audio_prompt_id\"]\n", " if audio_prompt_id not in audio_prompt_id_to_continue_at:\n", " audio_prompt_id_to_continue_at[audio_prompt_id] = row[\"continue_at\"]\n", " else:\n", " # pick the max\n", " audio_prompt_id = max(\n", " audio_prompt_id_to_continue_at[audio_prompt_id], row[\"continue_at\"]\n", " )\n", "print(len(audio_prompt_id_to_continue_at))\n", "df[\"has_continue_and_start_continue_at\"] = df[\"id\"].apply(\n", " lambda x: audio_prompt_id_to_continue_at.get(x)\n", ")\n", "# we want continue at to be at most of the clip...\n", "df[\"good_continue_at\"] = (\n", " (df[\"has_continue_and_start_continue_at\"] / df[\"duration\"]) > 0.9\n", ") | df[\"has_continue_and_start_continue_at\"].isna()\n", "print(df[\"good_continue_at\"].value_counts())\n", "\n", "\n", "print(\n", " \"\\n Check some basics... \\n\",\n", " df[\"preference\"].value_counts(),\n", " df[\"is_13b\"].value_counts(),\n", " df[\"model_name\"].value_counts(),\n", " df.groupby([\"preference\"])[\"model_name\"].value_counts(),\n", ")\n", "\n", "df = df.sort_values(by=[\"request_id\", \"preference\"])\n", "df[\"duration_rel_diff\"] = df[\"duration\"].diff()\n", "print(df.shape)\n", "# df[\"play_rel_diff\"] = df['reaction_play_count'].diff()" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.035167Z", "start_time": "2024-05-16T13:59:40.801098Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:44.085773Z", "iopub.status.busy": "2024-08-02T05:08:44.085286Z", "iopub.status.idle": "2024-08-02T05:08:44.129751Z", "shell.execute_reply": "2024-08-02T05:08:44.129233Z", "shell.execute_reply.started": "2024-08-02T05:08:44.085752Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "negative 11631 positive 10372\n", "total pair requests 11742 selected pair requests 10273 frac 0.875\n" ] } ], "source": [ "normal_pos_play_count = 3\n", "# this is lower, cause a concat is probably already ensuring that it is good\n", "concat_pos_play_count = 1\n", "# this is a filter on the concated clip\n", "concat_total_play_count = 3\n", "\n", "neg_filter_selection_mask = (\n", " (df[\"preference\"] == False) # get basics aligned\n", " # & (df[\"reaction_play_count\"] >= 1) # has to be played once\n", " # & (df[\"play_count\"] <= 3) # if it is actually bad, shouldn't be listened often\n", " & (df[\"duration\"] >= 10) # can't be too short, otherwise it is obvious\n", " & (df[\"duration\"] <= 240) # can't be badly long\n", " # & (df[\"has_continue_and_start_continue_at\"].isna()) # won't have any continues\n", " # & (df[\"dislike_count\"] >= 1) # this is kinda strict\n", " # & (\n", " # (df_slice[\"is_in_playlist\"] == False)\n", " # & (df_slice[\"concat_in_playlist\"] == False)\n", " # ) # can't be part of a playlist -- otherwise there are some like signal in it?\n", ")\n", "pos_filter_selectin_mask = (\n", " (df[\"preference\"] == True) # get basics aligned\n", " # & (\n", " # df[\"good_continue_at\"] == True\n", " # ) # if continue, needs to continue off a certain percentage\n", " # & (df[\"reaction_play_count\"] >= 1)\n", " # & (df[\"play_rel_diff\"] >= 0) # this is more like quality assurance\n", " & (df[\"duration\"] >= 10) # can't be too short, otherwise it is obvious\n", " & (df[\"duration\"] <= 240) # can't be badly long\n", " & (df[\"dislike_count\"] == 0) # can't have dislikes\n", " & (df[\"flag_count\"] == 0) # can't have issues\n", " # & (\n", " # (\n", " # (df[\"part_of_concat\"] == True)\n", " # & (df[\"play_count\"] >= concat_pos_play_count)\n", " # & (df[\"concat_play_counts\"] >= concat_total_play_count)\n", " # )\n", " # | (\n", " # (df[\"part_of_concat\"] == False)\n", " # & (df[\"play_count\"] >= normal_pos_play_count)\n", " # )\n", " # )\n", " # & (df[\"user_n_clips\"] >= 20) # user needs to have genereated at least 20\n", " # & (df[\"duration_rel_diff\"] < 10) # positive isn't just longer\n", " # & (df[\"upvote_count\"] >= 1)\n", ")\n", "print(\n", " \"negative\",\n", " sum(neg_filter_selection_mask),\n", " \"positive\",\n", " sum(pos_filter_selectin_mask),\n", ")\n", "\n", "neg_filter_requests = df[neg_filter_selection_mask][\"request_id\"].unique()\n", "pos_filter_requests = df[pos_filter_selectin_mask][\"request_id\"].unique()\n", "# looking for very strong signal here:\n", "# listen to the positive/negative more than once\n", "# disliked one of the clips\n", "unique_requests = set(pos_filter_requests).intersection(neg_filter_requests)\n", "print(\n", " \"total pair requests\",\n", " df[\"request_id\"].nunique(),\n", " \"selected pair requests\",\n", " len(unique_requests),\n", " f\"frac {len(unique_requests) / df['request_id'].nunique():.3f}\",\n", ")" ] }, { "cell_type": "code", "execution_count": 17, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:49.261780Z", "iopub.status.busy": "2024-08-02T05:08:49.261464Z", "iopub.status.idle": "2024-08-02T05:08:49.277850Z", "shell.execute_reply": "2024-08-02T05:08:49.277375Z", "shell.execute_reply.started": "2024-08-02T05:08:49.261761Z" } }, "outputs": [], "source": [ "# df[(df[\"neg_feedback\"] == True) & (df[\"upvote_count\"] != 0)][\"user_id\"].value_counts()" ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:49.760214Z", "iopub.status.busy": "2024-08-02T05:08:49.759908Z", "iopub.status.idle": "2024-08-02T05:08:49.774326Z", "shell.execute_reply": "2024-08-02T05:08:49.773892Z", "shell.execute_reply.started": "2024-08-02T05:08:49.760195Z" } }, "outputs": [], "source": [ "# df[(df[\"pos_feedback\"] == True) & (df[\"dislike_count\"] != 0)][\"user_id\"].value_counts()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Overwrite with only quality" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:51.368475Z", "iopub.status.busy": "2024-08-02T05:08:51.368016Z", "iopub.status.idle": "2024-08-02T05:08:51.382476Z", "shell.execute_reply": "2024-08-02T05:08:51.382026Z", "shell.execute_reply.started": "2024-08-02T05:08:51.368456Z" } }, "outputs": [], "source": [ "# import json\n", "\n", "# with open(\n", "# \"/home/tony/Data/Preference/13b_v0/interesting_clips_20240627_feedback_bad_audio_quality_ids.json\",\n", "# \"r\",\n", "# ) as fp:\n", "# bad_audio_quality_ids = json.load(fp)" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:52.012047Z", "iopub.status.busy": "2024-08-02T05:08:52.011767Z", "iopub.status.idle": "2024-08-02T05:08:52.293149Z", "shell.execute_reply": "2024-08-02T05:08:52.292710Z", "shell.execute_reply.started": "2024-08-02T05:08:52.012028Z" } }, "outputs": [], "source": [ "# unique_requests = df[df[\"id\"].isin(bad_audio_quality_ids)][\"request_id\"].unique()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## continue" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.250737Z", "start_time": "2024-05-16T13:59:41.036434Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:54.088368Z", "iopub.status.busy": "2024-08-02T05:08:54.088136Z", "iopub.status.idle": "2024-08-02T05:08:54.127436Z", "shell.execute_reply": "2024-08-02T05:08:54.126957Z", "shell.execute_reply.started": "2024-08-02T05:08:54.088351Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "requests 10273 clips 20546 total khrs 0.957; N gpus for 1000 iters 1.284; n unique users 684\n" ] } ], "source": [ "df_slice = df[df[\"request_id\"].isin(set(unique_requests))].copy()\n", "print(\n", " \"requests\",\n", " df_slice[\"request_id\"].nunique(),\n", " \"clips\",\n", " df_slice.shape[0],\n", " f\"total khrs {sum(df_slice['duration'] / 3600 / 1000):.3f};\",\n", " f\"N gpus for 1000 iters {df_slice.shape[0] / 8 / 2 / 1000:.3f};\",\n", " f\"n unique users {df_slice['user_id'].nunique()}\",\n", ")\n", "# 76171 152342 total khrs 2.880 n gpus for 1250 iters 3.809" ] }, { "cell_type": "code", "execution_count": 22, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:08:57.552455Z", "iopub.status.busy": "2024-08-02T05:08:57.552146Z", "iopub.status.idle": "2024-08-02T05:08:57.568577Z", "shell.execute_reply": "2024-08-02T05:08:57.568118Z", "shell.execute_reply.started": "2024-08-02T05:08:57.552436Z" } }, "outputs": [], "source": [ "# df_slice[df_slice[\"id\"].isin(bad_audio_quality_ids)][\"preference\"].value_counts()" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.277006Z", "start_time": "2024-05-16T13:59:41.252105Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:58.076360Z", "iopub.status.busy": "2024-08-02T05:08:58.075996Z", "iopub.status.idle": "2024-08-02T05:08:58.090399Z", "shell.execute_reply": "2024-08-02T05:08:58.089972Z", "shell.execute_reply.started": "2024-08-02T05:08:58.076342Z" } }, "outputs": [], "source": [ "# test_mask = (df_slice[\"preference\"] == True) & (\n", "# (df_slice[\"is_in_playlist\"] == True) | (df_slice[\"concat_in_playlist\"] == True)\n", "# )\n", "# print(\"positive in playlist\", df_slice[test_mask].shape)" ] }, { "cell_type": "code", "execution_count": 24, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.323409Z", "start_time": "2024-05-16T13:59:41.278278Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:58.200551Z", "iopub.status.busy": "2024-08-02T05:08:58.200266Z", "iopub.status.idle": "2024-08-02T05:08:58.213423Z", "shell.execute_reply": "2024-08-02T05:08:58.212977Z", "shell.execute_reply.started": "2024-08-02T05:08:58.200532Z" } }, "outputs": [], "source": [ "# interesting_clips_must_be_positive_mask = (\n", "# (df_slice[\"upvoted\"] == True)\n", "# | (df_slice[\"has_action\"] == True)\n", "# | (df_slice[\"part_of_concat\"] == True)\n", "# )\n", "# interesting_clips_must_be_not_negative_mask = (df_slice[\"downvoted\"] == False) # & (df_slice[\"dislike_count\"] < 1)\n", "# interesting_clips_mask = interesting_clips_must_be_positive_mask & interesting_clips_must_be_not_negative_mask\n", "# assert interesting_clips_mask.eq(df_slice[\"preference\"]).all()" ] }, { "cell_type": "code", "execution_count": 25, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.392244Z", "start_time": "2024-05-16T13:59:41.324472Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:59.133031Z", "iopub.status.busy": "2024-08-02T05:08:59.132678Z", "iopub.status.idle": "2024-08-02T05:08:59.144668Z", "shell.execute_reply": "2024-08-02T05:08:59.144248Z", "shell.execute_reply.started": "2024-08-02T05:08:59.133013Z" } }, "outputs": [], "source": [ "# save positive ids\n", "# positive_preference_ids = df_slice[df_slice[\"preference\"] == False][\"s3_id\"].to_json(orient='values')\n", "# with open('/home/tony/Data/Preference/7b_v2/7v_v20_full_recut_id_negative.json', 'w') as file:\n", "# file.write(positive_preference_ids)" ] }, { "cell_type": "code", "execution_count": 26, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T14:00:20.866354Z", "start_time": "2024-05-16T14:00:12.443344Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:08:59.413767Z", "iopub.status.busy": "2024-08-02T05:08:59.413621Z", "iopub.status.idle": "2024-08-02T05:08:59.425313Z", "shell.execute_reply": "2024-08-02T05:08:59.424896Z", "shell.execute_reply.started": "2024-08-02T05:08:59.413752Z" } }, "outputs": [], "source": [ "# df_slice.to_csv(\"/home/tony/Data/Preference/7b_v0/interesting_clips_v3_processed.csv\")" ] }, { "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": 27, "metadata": { "execution": { "iopub.execute_input": "2024-08-02T05:09:00.340204Z", "iopub.status.busy": "2024-08-02T05:09:00.339953Z", "iopub.status.idle": "2024-08-02T05:09:00.356119Z", "shell.execute_reply": "2024-08-02T05:09:00.355706Z", "shell.execute_reply.started": "2024-08-02T05:09:00.340187Z" } }, "outputs": [ { "data": { "text/plain": [ "0 NaN\n", "1 NaN\n", "2 NaN\n", "3 NaN\n", "4 NaN\n", " ..\n", "30983 NaN\n", "30990 NaN\n", "30991 NaN\n", "30992 NaN\n", "30993 NaN\n", "Name: continue_at, Length: 20546, dtype: float64" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df_slice[\"continue_at\"]" ] }, { "cell_type": "code", "execution_count": 28, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.932296Z", "start_time": "2024-05-16T13:59:41.932287Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:00.808116Z", "iopub.status.busy": "2024-08-02T05:09:00.807791Z", "iopub.status.idle": "2024-08-02T05:09:00.833784Z", "shell.execute_reply": "2024-08-02T05:09:00.833368Z", "shell.execute_reply.started": "2024-08-02T05:09:00.808099Z" } }, "outputs": [ { "data": { "text/plain": [ "8754" ] }, "execution_count": 28, "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": 29, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.932966Z", "start_time": "2024-05-16T13:59:41.932957Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:01.564141Z", "iopub.status.busy": "2024-08-02T05:09:01.563804Z", "iopub.status.idle": "2024-08-02T05:09:01.578182Z", "shell.execute_reply": "2024-08-02T05:09:01.577720Z", "shell.execute_reply.started": "2024-08-02T05:09:01.564122Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "10273\n" ] } ], "source": [ "final_filtered_requests = df_slice[\"request_id\"].unique()\n", "print(len(final_filtered_requests))" ] }, { "cell_type": "code", "execution_count": 30, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.933558Z", "start_time": "2024-05-16T13:59:41.933550Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:02.333549Z", "iopub.status.busy": "2024-08-02T05:09:02.333197Z", "iopub.status.idle": "2024-08-02T05:09:02.345351Z", "shell.execute_reply": "2024-08-02T05:09:02.344925Z", "shell.execute_reply.started": "2024-08-02T05:09:02.333532Z" } }, "outputs": [], "source": [ "# df_slice.to_csv(\"/home/tony/Data/Preference/7b_v2/7b_before_recode_20240412\", index=False)" ] }, { "cell_type": "code", "execution_count": 31, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.934277Z", "start_time": "2024-05-16T13:59:41.934268Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:02.812678Z", "iopub.status.busy": "2024-08-02T05:09:02.812276Z", "iopub.status.idle": "2024-08-02T05:09:02.864825Z", "shell.execute_reply": "2024-08-02T05:09:02.864357Z", "shell.execute_reply.started": "2024-08-02T05:09:02.812660Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "10170 103\n", "(20340, 82) (206, 82)\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", "\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": 32, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.934954Z", "start_time": "2024-05-16T13:59:41.934946Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:09.084716Z", "iopub.status.busy": "2024-08-02T05:09:09.084404Z", "iopub.status.idle": "2024-08-02T05:09:09.100754Z", "shell.execute_reply": "2024-08-02T05:09:09.100292Z", "shell.execute_reply.started": "2024-08-02T05:09:09.084698Z" } }, "outputs": [], "source": [ "# val_df[[\"request_id\", \"metadata\", \"updated_at\", \"user_id\", \"preference\"]].head()" ] }, { "cell_type": "code", "execution_count": 33, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.935620Z", "start_time": "2024-05-16T13:59:41.935613Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:09.637313Z", "iopub.status.busy": "2024-08-02T05:09:09.637016Z", "iopub.status.idle": "2024-08-02T05:09:10.249256Z", "shell.execute_reply": "2024-08-02T05:09:10.248684Z", "shell.execute_reply.started": "2024-08-02T05:09:09.637295Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████| 20340/20340 [00:00<00:00, 34208.13it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "947 hours of 20340 clips, 1.8160714285714286 nodes\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "total_duration = 0\n", "for i, row in 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(\n", " f\"{round(total_duration / 60 / 60):,} hours of {train_df.shape[0]} clips, {train_df.shape[0] / 8 / 2 / 700} nodes\"\n", ")" ] }, { "cell_type": "code", "execution_count": 34, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.936268Z", "start_time": "2024-05-16T13:59:41.936260Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:18.340514Z", "iopub.status.busy": "2024-08-02T05:09:18.340199Z", "iopub.status.idle": "2024-08-02T05:09:21.768680Z", "shell.execute_reply": "2024-08-02T05:09:21.767979Z", "shell.execute_reply.started": "2024-08-02T05:09:18.340495Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 206/206 [00:03<00:00, 60.56it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Total 206 clips, 0 different prompts\n", "5 hours of False\n", "5 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": 35, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.936964Z", "start_time": "2024-05-16T13:59:41.936957Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:09:22.380356Z", "iopub.status.busy": "2024-08-02T05:09:22.380158Z", "iopub.status.idle": "2024-08-02T05:15:01.536875Z", "shell.execute_reply": "2024-08-02T05:15:01.536211Z", "shell.execute_reply.started": "2024-08-02T05:09:22.380338Z" } }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████| 20340/20340 [05:39<00:00, 59.98it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Total 20340 clips, 40 different prompts\n", "473 hours of False\n", "474 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": 36, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.937879Z", "start_time": "2024-05-16T13:59:41.937870Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:01.037164Z", "iopub.status.busy": "2024-08-02T05:29:01.036780Z", "iopub.status.idle": "2024-08-02T05:29:01.345824Z", "shell.execute_reply": "2024-08-02T05:29:01.345225Z", "shell.execute_reply.started": "2024-08-02T05:29:01.037144Z" } }, "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": 37, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.938629Z", "start_time": "2024-05-16T13:59:41.938621Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:01.520691Z", "iopub.status.busy": "2024-08-02T05:29:01.520347Z", "iopub.status.idle": "2024-08-02T05:29:01.535118Z", "shell.execute_reply": "2024-08-02T05:29:01.534583Z", "shell.execute_reply.started": "2024-08-02T05:29:01.520672Z" } }, "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": 38, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.939205Z", "start_time": "2024-05-16T13:59:41.939198Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:01.848755Z", "iopub.status.busy": "2024-08-02T05:29:01.848340Z", "iopub.status.idle": "2024-08-02T05:29:01.862210Z", "shell.execute_reply": "2024-08-02T05:29:01.861695Z", "shell.execute_reply.started": "2024-08-02T05:29:01.848736Z" } }, "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": 39, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.939977Z", "start_time": "2024-05-16T13:59:41.939969Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:01.972314Z", "iopub.status.busy": "2024-08-02T05:29:01.971909Z", "iopub.status.idle": "2024-08-02T05:29:01.984089Z", "shell.execute_reply": "2024-08-02T05:29:01.983570Z", "shell.execute_reply.started": "2024-08-02T05:29:01.972295Z" } }, "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": 40, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.940610Z", "start_time": "2024-05-16T13:59:41.940603Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:02.256278Z", "iopub.status.busy": "2024-08-02T05:29:02.255988Z", "iopub.status.idle": "2024-08-02T05:29:02.268035Z", "shell.execute_reply": "2024-08-02T05:29:02.267525Z", "shell.execute_reply.started": "2024-08-02T05:29:02.256259Z" } }, "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": 41, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.941167Z", "start_time": "2024-05-16T13:59:41.941159Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:02.896414Z", "iopub.status.busy": "2024-08-02T05:29:02.896134Z", "iopub.status.idle": "2024-08-02T05:29:02.908282Z", "shell.execute_reply": "2024-08-02T05:29:02.907765Z", "shell.execute_reply.started": "2024-08-02T05:29:02.896396Z" } }, "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": 42, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.941801Z", "start_time": "2024-05-16T13:59:41.941793Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:03.132657Z", "iopub.status.busy": "2024-08-02T05:29:03.132257Z", "iopub.status.idle": "2024-08-02T05:29:03.145924Z", "shell.execute_reply": "2024-08-02T05:29:03.145360Z", "shell.execute_reply.started": "2024-08-02T05:29:03.132638Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "103 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", "\n", "validation_on_metas(test_metas)" ] }, { "cell_type": "code", "execution_count": 43, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.942520Z", "start_time": "2024-05-16T13:59:41.942511Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:03.696372Z", "iopub.status.busy": "2024-08-02T05:29:03.695950Z", "iopub.status.idle": "2024-08-02T05:29:03.710262Z", "shell.execute_reply": "2024-08-02T05:29:03.709733Z", "shell.execute_reply.started": "2024-08-02T05:29:03.696352Z" } }, "outputs": [], "source": [ "train_info = read_json(os.path.join(OUT_DATA_DIR, f\"info_tr.json\"))" ] }, { "cell_type": "code", "execution_count": 44, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.943072Z", "start_time": "2024-05-16T13:59:41.943065Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:04.468438Z", "iopub.status.busy": "2024-08-02T05:29:04.468133Z", "iopub.status.idle": "2024-08-02T05:29:04.484906Z", "shell.execute_reply": "2024-08-02T05:29:04.484423Z", "shell.execute_reply.started": "2024-08-02T05:29:04.468419Z" } }, "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": 45, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.944246Z", "start_time": "2024-05-16T13:59:41.944237Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:04.808797Z", "iopub.status.busy": "2024-08-02T05:29:04.808255Z", "iopub.status.idle": "2024-08-02T05:29:04.824204Z", "shell.execute_reply": "2024-08-02T05:29:04.823717Z", "shell.execute_reply.started": "2024-08-02T05:29:04.808773Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "total samples 20340 (20340, 82)\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": 46, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.945249Z", "start_time": "2024-05-16T13:59:41.945241Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:05.536076Z", "iopub.status.busy": "2024-08-02T05:29:05.535912Z", "iopub.status.idle": "2024-08-02T05:29:05.549317Z", "shell.execute_reply": "2024-08-02T05:29:05.548862Z", "shell.execute_reply.started": "2024-08-02T05:29:05.536060Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1 epoch per batch 4, total 635.625\n" ] } ], "source": [ "print(\"1 epoch per batch 4, total\", total_iters / 8 / 2 / 2)" ] }, { "cell_type": "code", "execution_count": 47, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.945972Z", "start_time": "2024-05-16T13:59:41.945964Z" }, "execution": { "iopub.execute_input": "2024-08-02T05:29:07.260723Z", "iopub.status.busy": "2024-08-02T05:29:07.260185Z", "iopub.status.idle": "2024-08-02T05:29:07.474356Z", "shell.execute_reply": "2024-08-02T05:29:07.473798Z", "shell.execute_reply.started": "2024-08-02T05:29:07.260702Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Submitted batch job 1303\n" ] } ], "source": [ "!cd /home/tony/Work/tony/slurm/feedback/ && sbatch sbatch_ipo_13b_feedback" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# some gymathtics loading prev data" ] }, { "cell_type": "code", "execution_count": 48, "metadata": { "ExecuteTime": { "end_time": "2024-05-16T13:59:41.946562Z", "start_time": "2024-05-16T13:59:41.946555Z" }, "execution": { "iopub.execute_input": "2024-07-19T05:01:11.985249Z", "iopub.status.busy": "2024-07-19T05:01:11.985061Z", "iopub.status.idle": "2024-07-19T05:01:12.006061Z", "shell.execute_reply": "2024-07-19T05:01:12.005532Z", "shell.execute_reply.started": "2024-07-19T05:01:11.985228Z" } }, "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 }