models/BERT/training.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "mount_file_id": "1le-owYXMUHA7Yjgpj4QSBehRoSdOey7_",
      "authorship_tag": "ABX9TyOfK0XWch0KiCoF+QbnqOc6",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/aryamanpathak2022/Sentimental_analysis_wn/blob/main/models/BERT/training.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#BERT training"
      ],
      "metadata": {
        "id": "oKRk1ZUHzagM"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install transformers\n",
        "!pip install torch"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true,
        "id": "PfmUnzAgzkEf",
        "outputId": "de4a68bf-b9bf-4717-f6ba-552d09e2b513"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: transformers in /usr/local/lib/python3.10/dist-packages (4.40.1)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from transformers) (3.14.0)\n",
            "Requirement already satisfied: huggingface-hub<1.0,>=0.19.3 in /usr/local/lib/python3.10/dist-packages (from transformers) (0.20.3)\n",
            "Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.10/dist-packages (from transformers) (1.25.2)\n",
            "Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.10/dist-packages (from transformers) (24.0)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.10/dist-packages (from transformers) (6.0.1)\n",
            "Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.10/dist-packages (from transformers) (2023.12.25)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.10/dist-packages (from transformers) (2.31.0)\n",
            "Requirement already satisfied: tokenizers<0.20,>=0.19 in /usr/local/lib/python3.10/dist-packages (from transformers) (0.19.1)\n",
            "Requirement already satisfied: safetensors>=0.4.1 in /usr/local/lib/python3.10/dist-packages (from transformers) (0.4.3)\n",
            "Requirement already satisfied: tqdm>=4.27 in /usr/local/lib/python3.10/dist-packages (from transformers) (4.66.2)\n",
            "Requirement already satisfied: fsspec>=2023.5.0 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub<1.0,>=0.19.3->transformers) (2023.6.0)\n",
            "Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub<1.0,>=0.19.3->transformers) (4.11.0)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests->transformers) (3.3.2)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests->transformers) (3.7)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests->transformers) (2.0.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests->transformers) (2024.2.2)\n",
            "Requirement already satisfied: torch in /usr/local/lib/python3.10/dist-packages (2.2.1+cu121)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from torch) (3.14.0)\n",
            "Requirement already satisfied: typing-extensions>=4.8.0 in /usr/local/lib/python3.10/dist-packages (from torch) (4.11.0)\n",
            "Requirement already satisfied: sympy in /usr/local/lib/python3.10/dist-packages (from torch) (1.12)\n",
            "Requirement already satisfied: networkx in /usr/local/lib/python3.10/dist-packages (from torch) (3.3)\n",
            "Requirement already satisfied: jinja2 in /usr/local/lib/python3.10/dist-packages (from torch) (3.1.3)\n",
            "Requirement already satisfied: fsspec in /usr/local/lib/python3.10/dist-packages (from torch) (2023.6.0)\n",
            "Collecting nvidia-cuda-nvrtc-cu12==12.1.105 (from torch)\n",
            "  Using cached nvidia_cuda_nvrtc_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (23.7 MB)\n",
            "Collecting nvidia-cuda-runtime-cu12==12.1.105 (from torch)\n",
            "  Using cached nvidia_cuda_runtime_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (823 kB)\n",
            "Collecting nvidia-cuda-cupti-cu12==12.1.105 (from torch)\n",
            "  Using cached nvidia_cuda_cupti_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (14.1 MB)\n",
            "Collecting nvidia-cudnn-cu12==8.9.2.26 (from torch)\n",
            "  Using cached nvidia_cudnn_cu12-8.9.2.26-py3-none-manylinux1_x86_64.whl (731.7 MB)\n",
            "Collecting nvidia-cublas-cu12==12.1.3.1 (from torch)\n",
            "  Using cached nvidia_cublas_cu12-12.1.3.1-py3-none-manylinux1_x86_64.whl (410.6 MB)\n",
            "Collecting nvidia-cufft-cu12==11.0.2.54 (from torch)\n",
            "  Using cached nvidia_cufft_cu12-11.0.2.54-py3-none-manylinux1_x86_64.whl (121.6 MB)\n",
            "Collecting nvidia-curand-cu12==10.3.2.106 (from torch)\n",
            "  Using cached nvidia_curand_cu12-10.3.2.106-py3-none-manylinux1_x86_64.whl (56.5 MB)\n",
            "Collecting nvidia-cusolver-cu12==11.4.5.107 (from torch)\n",
            "  Using cached nvidia_cusolver_cu12-11.4.5.107-py3-none-manylinux1_x86_64.whl (124.2 MB)\n",
            "Collecting nvidia-cusparse-cu12==12.1.0.106 (from torch)\n",
            "  Using cached nvidia_cusparse_cu12-12.1.0.106-py3-none-manylinux1_x86_64.whl (196.0 MB)\n",
            "Collecting nvidia-nccl-cu12==2.19.3 (from torch)\n",
            "  Using cached nvidia_nccl_cu12-2.19.3-py3-none-manylinux1_x86_64.whl (166.0 MB)\n",
            "Collecting nvidia-nvtx-cu12==12.1.105 (from torch)\n",
            "  Using cached nvidia_nvtx_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (99 kB)\n",
            "Requirement already satisfied: triton==2.2.0 in /usr/local/lib/python3.10/dist-packages (from torch) (2.2.0)\n",
            "Collecting nvidia-nvjitlink-cu12 (from nvidia-cusolver-cu12==11.4.5.107->torch)\n",
            "  Downloading nvidia_nvjitlink_cu12-12.5.40-py3-none-manylinux2014_x86_64.whl (21.3 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m21.3/21.3 MB\u001b[0m \u001b[31m59.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.10/dist-packages (from jinja2->torch) (2.1.5)\n",
            "Requirement already satisfied: mpmath>=0.19 in /usr/local/lib/python3.10/dist-packages (from sympy->torch) (1.3.0)\n",
            "Installing collected packages: nvidia-nvtx-cu12, nvidia-nvjitlink-cu12, nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cufft-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, nvidia-cusparse-cu12, nvidia-cudnn-cu12, nvidia-cusolver-cu12\n",
            "Successfully installed nvidia-cublas-cu12-12.1.3.1 nvidia-cuda-cupti-cu12-12.1.105 nvidia-cuda-nvrtc-cu12-12.1.105 nvidia-cuda-runtime-cu12-12.1.105 nvidia-cudnn-cu12-8.9.2.26 nvidia-cufft-cu12-11.0.2.54 nvidia-curand-cu12-10.3.2.106 nvidia-cusolver-cu12-11.4.5.107 nvidia-cusparse-cu12-12.1.0.106 nvidia-nccl-cu12-2.19.3 nvidia-nvjitlink-cu12-12.5.40 nvidia-nvtx-cu12-12.1.105\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
      ],
      "metadata": {
        "id": "S4jCRVB230c7"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from transformers import BertTokenizer, BertForSequenceClassification\n",
        "import torch\n",
        "\n",
        "# Load pre-trained BERT tokenizer and model\n",
        "tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n",
        "model = BertForSequenceClassification.from_pretrained('bert-base-uncased',num_labels=3)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true,
        "id": "Ox8rI019zZ2i",
        "outputId": "3f1b9dfa-691b-4540-b1d5-6cb3a54b765b"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at bert-base-uncased and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "\n",
        "# Read the CSV file into a DataFrame\n",
        "df = pd.read_csv('drive/MyDrive/Sentiment_analysis/manual_processed_Articles.csv')\n",
        "\n",
        "# Extract the 'Articles' and 'sentiment' columns and convert them to lists\n",
        "articles_list = df['Processed_Article'].tolist()\n",
        "sentiment_list = df['sentiment']\n",
        "label_mapping = {'positive': 1, 'negative': 2, 'neutral': 0}\n",
        "\n",
        "df['sentiment'] = df['sentiment'].map(label_mapping)\n",
        "\n",
        "# Print the lists\n",
        "print(articles_list)\n",
        "print(sentiment_list)\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true,
        "id": "m6UWRpCJ1rGE",
        "outputId": "c7c90f76-1274-4e7c-9ce9-aee22a0de8aa"
      },
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['least people dead others injured two train collided southeast india sunday police official said train traveling visakhapatnam state andhra pradesh rayagada odisha stopped due break overhead cable hit oncoming train vizianagaram district andhra pradesh reuters said citing senior railway official collision derailed two coach carrying people stationary train official told reuters preliminary investigation suggests human error caused overshooting signal stationary train led collision statement country railway ministry read people died person identified deepika superintendent police vizianagaram district told cnn monday said railway police would investigate incident sunday collision come four month separate train accident odisha three train collided leaving people dead injured rescue operation underway accident site injured moved hospital railway minister ashwini vaishnaw said x chief minister office andhra pradesh said x ordered official take quick relief measure ensure injured get prompt medical attention advised nearby district send first responder scene office indian prime minister narendra modi said sunday spoken vaishnaw accident authority providing possible assistance affected post x read prime minister extends condolence bereaved family prays injured recover soon india minister railway ashwini vaishnaw monday announced compensation lost family member suffered grievous injury minor injury', 'one india revered entrepreneur belief young people need put exceptionally long hour work want see country become global economic powerhouse nr narayana murthy cofounder software behemoth infosys said india need highly determined extremely disciplined extremely hardworking youngster put hour week work know exactly german japanese second world war murthy told mohandas pai former cfo infosys chat published youtube thursday murthy whose wealth estimated billion forbes cofounded infosys went become one world biggest outsourcing firm also fatherinlaw british prime minister rishi sunak somehow youth habit taking desirable habit west helping country murthy said india work productivity one lowest world added indian learn policy helped emerging market particularly china tech tycoon said corporate leader india motivate young people work hard telling first time india received certain respect globally time u consolidate accelerate progress india one fastest growing major economy world projected expand year according international monetary fund murthys comment putting longer hour professionally come time drastic shift way people think work west china year many young people china embraced new philosophy theyve called lying flat fight country excessive work culture practice working pm six day week united state quiet quitting became workplace trend last year people decided stop work beyond hired murthys comment slammed social medium past year several survey shown indian feel among overworked underpaid people world cnn reached infosys comment', 'india economy like elephant world populous nation hard get moving force reckon action analyst often say global economy lurched one crisis another recent time india shed lumbering gait broken steady trot began year strutting stuff world economic forum switzerland emissary india dominant presence main street davos one investor described thoroughfare little india month later prime minister narendra modi inaugurated group g leader summit new delhi first india country stock market zoomed dizzying height growing economic confidence isnt limited earth august india joined tiny club country safely placed spacecraft moon underscoring scientific technological ambition euphoria around india come time china engine global growth decade seeing major economic slowdown southern neighbor fast emerging potential successor surging young population humming factory country lot going favor indian economy undeniably poised greatness number reform undertaken past year finally paving way solid growth said eswar prasad professor trade policy cornell university adding country receiving considerable interest foreign investor least good reason new delhi also warmer relation west view china growing suspicion mean investor consider world largest democracy bright spot increasingly fractured world past decade period global bullishness india excitement kept fizzling china raced ahead gap two asian economy massive india economy currently worth nearly trillion making world fifth largest china economy world second largest bigger almost trillion two combined expected contribute half global growth year international monetary fund said coming china overtake china biggest contributor global growth next five year india must achieve sustained growth rate barclays analyst wrote october report imf project india expand year china hand set official growth target around grapple mounting challenge ranging weak consumer spending deepening property crisis india economy comfortably placed grow annual rate least coming year barclays said achieve historicallyaspirational pace growth private sector india need step level investment added modi government aim make india trillion economy certainly laying groundwork make easier business attract company invest much like china three decade ago india beginning massive infrastructure transformation spending billion building road port airport railway year budget alone billion carved capital spending boost economic expansion result seen ground furious construction underway across nation india added km mile national highway network increase overall length modi government said daily pace national highway construction doubled since first came power nine year ago india world biggest software company also built range digital platform known digital public infrastructure transformed commerce digitization game changer country citizen business said prasad formalization economy reduced many friction business given india citizen sense vested country economic success example aadhaar program launched changed life million indian providing proof identity first time ever world largest biometric database cover india billion people helped government save million reducing corruption welfare initiative another platform unified payment interface upi allows user make payment instantly scanning qr code embraced indian walk life coffee shop owner beggar allowed million dollar flow formal economy september modi citing world bank report said thanks digital public infrastructure india achieved financial inclusion target year would otherwise taken least long year indian company getting act country biggest conglomerate including mukesh ambanis reliance industry gautam adanis eponymous group spending billion g clean energy even though built empire back traditional industry fossil fuel india aggressively trying capitalize massive rethink underway among company supply chain international firm want diversify operation away china faced obstacle pandemic threatened rising tension beijing washington asia third largest economy launched productionlinked incentive program worth billion attract company set manufacturing sector ranging electronics automobile pharmaceutical medical device result world biggest firm including apple supplier foxconn expanding operation significantly india even india heft increasing far recreating economic miracle china unleashed decade ago india like china late early government clearing obstacle fdi foreign direct investment quickly said willy shih professor harvard business school think perception different compared china early bureaucratic unpredictability around nontariff barrier thing get way unpredictability full display modi suddenly banned india cash resulting longterm pain citizen business country taking many step woo foreign company authority cracked firm china october report hsbc economist frederic neumann justin feng wrote india moment run cylinder take slack china sputtering growth engine going highlight difference consumption investment two economy china account around world investment india le even assuming zero growth china tripling investment spending growth india recent average would take another year india investment spending catch china wrote would take another year india consumption catch china today term total spending according report none say india going impact doubt though wont enough shield world economy china economy stumble badly added', 'piece debris stood munna qureshi dozen laborer team tasked rescuing deep inside himalayan tunnel previous attempt free failed could hear laborer gasping side excitement yearold said heart racing removed last rock u qureshi among specialized worker called indian authority help last month rescue construction worker trapped collapsed tunnel northern uttarakhand state nearly three week construction worker cut world meter inside mountain receiving food air thin tube frequent update rescuer outside engineer worked round clock drill safe passage broken rock using state art machine official flew expert help rescue effort ultimately day qureshi colleague succeeded bringing men safety drill broke beyond repair meter trapped worker known locally rat hole miner belong niche group highly skilled poorly paid excavator typically crawl narrow tunnel extract coal deep within ground profession dangerous banned part country thrust spotlight recent week men celebrated hero many across country rat hole mining may illegal lt general syed ata hasnain retired official india national disaster management authority told reporter shortly rescue rat miner talent experience worker employed dangerous profession among vulnerable marginalized india hence unflattering local moniker mostly migrant india poorest state paid day work according local report slimly built nimble expected enter tiny crevice mine often deprived oxygen risk buried loose soil coal mining india take place northeastern meghalaya state home country largest coal deposit amounting million metric ton rat hole mining banned state national green tribunal ngt due health environmental risk still carried illegally secluded pocket region according hasina kharbhih founder impulse nonprofit advocate safety worker estimated rat hole miner died practice banned four year ban implemented another died becoming trapped illegal coal mine two week number however tip iceberg kharbhih said sure region happened thoroughly researched number would go men called rescue laborer said knew risk joined profession always thought job would take life someday one worker nasir khan said never thought would earn respect however justice bp katoki retired judge appointed tribunal implement ban rat mining said country shouldnt normalize dangerous profession last effort rescue life people argument made necessary told cnn mean one celebrates practice dangerous implication uttarakhand chief minister pushkar singh dhami announced rupee handed among worker token appreciation rat miner said still waiting detail compensation despite medium attention said thanked rewarded authority two hour men rescued list men took part effort circulated uttarakhand government official medium whatsapp group conspicuously missing name rat miner put life line complete final breakthrough surprise said mohammad irshad ansari one rat miner worker laborer seen laborer whatever may done change poor cnn contacted uttarakhand government comment compensation one men monu kumar said received hero welcome arrived home northern uttar pradesh state music playing garlanded family neighbour distributed kilogram worth sweet everyone said people village saying much put life line get anything return get trapped worker rat miner crawl centimeter foot diameter pipe inserted debris crouch hour small space dig final meter foot rubble bare hand khan said unlike anything ever seen debris stone mountain steel pipe water rope used many tool slowly get added two rat miner went time rotating fourhour shift one cutting stone pulling debris pipe difficult risky said kumar doubt ansari want keep going rescue men situation arise khan family urged quit job saying risk worth low pay despite working three decade industry afford send three child school work also seasonal summer month access oxygen problem monsoon water make unsafe said working four month year afford send child school go school kumar feel medium attention wont last long soon call stop coming said one going remember u', 'half century ago india prime minister indira gandhi talked staggering challenge developing nation industrialize without harming environment one hand rich look askance continuing poverty warn u method told united nation event stockholm first global conference make environment major issue wish impoverish environment yet moment forget grim poverty large number people added word never relevant tension economic growth environmental protection heart global discussion tackle ever accelerating climate crisis addressing opening session cop climate talk dubai friday india current prime minister narendra modi said developing country must given fair share global carbon budget amount planetwarming carbon pollution world emit still avoid climate catastrophe even though earth heating dangerous level many government around world persist viewing coal oil gas source economic development energy security geopolitical power un said year result world fossil fuel production set twice amount needed limit global temperature rise degree celsius preindustrial level goal paris climate agreement recent un environment programme report found one major contributor disastrous overshoot india burning ever greater amount coal oil try meet need billion people plan double domestic coal production even world populous nation cling coal one hand sign attempting chart sustainable course india made significant investment set ambitious target renewable energy unep report said noting world fastest growing major economy earmarked billion toward energy transition year national budget global agency also noted india growing ambition pursuit green energy parisbased international energy agency said report october country moving dynamic new phase energy development marked longterm net zero emission ambition modi pledged india would achieve net zero emission still couple decade later developed economy transforming india energy sector like thing country going chaotic muddled farreaching implication global energy market race limit global warming india net zero target met full time along respective national target take world degree celsius global average surface temperature rise said siddharth singh energy investment analyst iea getting need come cost growth already early sign gradually loosening link economic development carbon emission said singh country able meet pledge carbon emission fall even gdp quadruple period iea said report india world thirdlargest energy consuming country although energy use emission per person le half world average iea data show could change rapidly thanks rising income energy demand doubled since demand still met coal oil solid biomass next three decade south asian nation see largest energy demand growth country world iea said superlative isnt surprising country expected achieve impressive economic milestone world fifth largest economy comfortably placed grow annual rate least coming year analyst say may become third country annual gdp trillion develops modernizes urban population going shoot leading massive rise construction home office shop building india add equivalent london urban population every year next year said singh modi government also trying boost domestic manufacturing unleashed building boom everything road bridge port railway springing around country infrastructure bonanza lead surge demand coal steel huge source carbon emission electricity demand also expected skyrocket coming year factor ranging improved standard living climate change latter fueling deadly heatwaves across india result air conditioner ownership set see sharp spike coming year india total electricity demand residential air conditioner set exceed total energy consumption whole africa today iea said year coal account almost country electricity generation likely change near future future role fossil fuel one controversial issue nation grappling cop pushing phaseout others calling weaker language phasedown india said former option feasible moment phase fossil fuel unless nuclear energy storage becomes viable india power minister rk singh said last week day said country wont rushed reducing use coal going going compromise availability power growth even requires add coalbased capacity minister said november still fact india developing point history give unique opportunity repeat climate sin wealthier nation country reluctant signatory paris climate accord country pledged limit precipitous rise global temperature well degree celsius preindustrial level preferred goal degree pledge havent yet delivered world careening toward nearly degree global warming even current climate policy met united nation warned modi government set ambitious goal promised renewable energy fulfil india energy requirement end decade government also set target gw nonfossil fuel electricity generating capacity period gw last year india launched incentive program boost domestic manufacturing critical sector including solar module advanced chemistry cell battery program work could see india establish reliable exporter solar module iea said report country already world fourth largest combined capacity wind solar pv electricity generation third largest next year singh said context development journey india clean energy target indeed impressive asia third largest economy able meet pledge would also offer new roadmap growth one allows developing country get rich go green time major conglomerate keen seize green opportunity india richest men including mukesh ambani gautam adani investing billion clean energy even though made empire back fossil fuel probably never better time india grow sustainably said tim buckley director sydneybased think tank climate energy finance thats two main reason world seeing unprecedented level investment clean technology following russia invasion ukraine solar energy getting significantly cheaper explained still big funding gap iea said energy investment need nearly triple end decade india trajectory meet zero emission target modi government want rich nation come climate financing developed world agreed decade ago transfer least billion year developing country help green transition effort adapt climate crisis pledge reaffirmed paris agreement target never met speech cop friday modi warned thinking one self interest lead world darkness hopefully message heard india vast bureaucracy well need departmental action urgency filtering respect excellent renewables ambition modi said buckley india need make sure there alignment implementation really ambitious target going met', 'india top court monday upheld government controversial decision revoke special status jammu kashmir also ruling disputed muslimmajority territory regain state designation local election held next year supreme court judgment chief justice dy chandrachud said restoration statehood shall take place earliest soon possible directed country election commission hold poll jammu kashmir end september claimed entirety india pakistan mountainous kashmir region epicenter oftenviolent territorial struggle nucleararmed neighbor year region one world dangerous flashpoint de facto border called line control divide area overseen new delhi islamabad four year ago prime minister narendra modis ruling bharatiya janata party bjp scrapped article constitutional provision granted special status former state including power constitution flag autonomy matter save certain policy area foreign affair defense day later india parliament voted split jammu kashmir two union territory highly contentious move gave new delhi greater control disputed region meanwhile remote mountainous region ladakh previously part jammu kashmir turned standalone territory part disputed region claimed india china reclassification ladakh also upheld supreme court said monday revoking kashmir special status one modis key promise general election campaign supreme court verdict come month expected run rare third term modi monday hailed court verdict resounding declaration hope progress unity today supreme court verdict abrogation article historic constitutionally upholds decision taken parliament india th august modi wrote x formerly twitter verdict beacon hope promise brighter future testament collective resolve build stronger united india added scrapping article brought jammu kashmir provision indian constitution move allowed nonresident buy property valley apply job scholarship previously reserved state existing population move also prompted uproar opposition party time critic saying change kashmir status could inflame tension risk violence already unstable region monday kashmiri politician expressed disappointment court verdict people jk going lose hope give fight honour dignity continue regardless wrote mehbooba mufti former chief minister current leader jammu kashmir people democratic party isnt end road u omar abdullah vice president jammu kashmir national conference said disappointed disheartened struggle continue took bjp decade reach also prepared long haul wrote x', 'fbi director christopher wray india week trip aimed strengthening security cooperation deepening partnership come wake major law enforcement issue two nation one far sinister potential cause crack within alliance two week ago united state accused indian government official involved conspiracy kill american citizen home soil citizen question gurpatwant singh pannun sikh separatist allegedly target murderforhire plot new york city pannun wanted man india considered government terrorist national security threat sikh overseas pannun outspoken activist man rallying cause come unite large swath community international diaspora late last month u federal prosecutor charged indian national suspicion trying kill according explosive indictment alleged hitman acting upon order unnamed indian government official stunning revelation came little month canada prime minister justin trudeau publicly alleged india may involved murder another sikh separatist hardeep singh nijjar soil prompting furious response new delhi provoking diplomatic fallout two country india vehemently denied involvement nijjars murder u plot foiled detail indictment yet heard court episode cast dark shadow new delhi credibility wrote suhasini haider diplomatic affair editor hindu newspaper one recent editorial many asking badly hit new delhi tie washington india government denied involvement alleged plot kill pannun contrast vocal condemnation made canada accusation set highlevel committee investigate accusation u indictment regard case individual filed u court allegedly linking indian official matter concern india foreign ministry said indictment revealed cause pannun campaigning creation separate sikh homeland one would known khalistan include state punjab india khalistan long outlawed world largest democracy painful memory deadly insurgency sikh separatist continue haunt many indian citizen garner level sympathy among sikh diaspora overseas protected free speech law people like pannun able openly advocate secession india topic highly charged four decade ago india former prime minister indira gandhi assassinated sikh bodyguard sent army troop charging inside religion holiest shrine flush separatist harboring inside operation known blue star destroyed much building left hundred dead riot broke day gandhi murder killing people predominantly sikh according official figure one worst outbreak communal violence since india partition year since pursuing khalistan cause formed minority punjab given government banning movement linked extremism doesnt pose significant threat country analyst say seeking reconciliation view human right abuse committed community several overseas sikh organization continue advocate punjab secession say khalistan movement falsely equated terrorism indian government despite government response campaigner overseas remains robust particularly current prime minister narendra modi india went difficult phase khalistan movement defined national security said harsh pant vicepresident observer research foundation new delhi think tank therefore indian state tends take serious view thing pannun sikh separatist wanted indian government close associate nijjar also wanted india terrorism related offense fiery orator pannun repeatedly made comment perceived secessionist threatening toward india government wanted dozen terrorism charge india accused trying radlicalize community support creation khalistan independent sikh homeland would include part india india outlawed usbased secessionist group sikh justice headed pannun hold referendum among diaspora group support khalistan widely circulated video last month warned sikh fly air india november citing danger life headline reverberated across india day followed claiming pannun threatening blow aircraft pannun told cnn surprised indictment outlining foiled plot kill government india officially declared reward head running global khalistan referendum campaign said referring vote sikh diaspora around world held gathering support separate homeland asked cnn threat india new delhi claim replied advocacy surely challenge india territorial integrity challenge ballot bullet hence challenge pose considered terrorism un international law remark air india said comment indian medium often misrepresented video message boycott air india bomb said disinformation tactic india modi government equate nonviolent call boycott terrorism analyst say pannuns inflammatory rhetoric medium hype surround india could provoke backlash citizen prompt government response someone making statement regularly indian medium broadcasting take conversation beyond confines diplomacy pant said many indian might ask u government tolerating behavior public perception danger may much higher danger indian government alleged pursuit american citizen home soil might appear risky put dent relation two country recently rapidly strengthened economic technological defense tie however analyst say chance major fallout low committed countering rise neighboring superpower view joint concern china common wisdom u india need vital strategic purpose mount effective front competition china said daniel markey senior adviser south asia united state institute peace therefore manage issue without sidetracked mission date two side avoided acrimonious public statement pledging take matter seriously recent past precedent kind pragmatic realpolitik murder saudi arabia istanbul consulate jamal khashoggi u resident washington post columnist critical saudi government caused global outrage u condemnation time ultimately vital relationship washington riyadh stabilized statement monday announcing wrays visit india central bureau investigation equivalent fbi said meeting signified step towards deepening cooperation shared commitment combat crime manifestation spirit international police cooperation statement mention u indictment measured usindia response might suggest india topmost official werent privy alleged plot find difficult accept logic go way top said pant primarily think cost high assume rogue element think question lost control u indictment alleged nikhil gupta worked unnamed indian official set meeting undercover officer believed hitman target pannun indian official agreed pay undercover officer arranged murder according prosecutor hitman contacted gupta fact undercover officer drug enforcement agency according dea gupta arrested june year czech republic pursuant bilateral extradition treaty india least incident done little damage government reputation hardliner may actually take pride episode demonstration indian muscle believe necessary deterrent overseas critic indian state said markey usip pant said might fringe support alleged attempt pannuns life government understands complexity case thats youve seen swiftly response america committee constituted investigate added think implication quite serious however alleged tie indian government proven markey said country professional intelligence officer pay reputational cost india diplomat especially washington ottawa left clean mess said', 'act vandalism th century monument built powerful mughal empire sparked anger india historian urging authority provide better protection country vast cultural heritage destruction part three emperor tomb zafar mahal palace new delhi discovered historian sunday heritage walk site overseen government agency responsible conserving historic monument believed last great mughal monument said historian sam dalrymple alerted security guard damage zafar mahal slowly falling apart year symbolic wider neglect mughal history delhi vandalism includes damage part marble lattice screen jaali surround empty grave bahadur shah zafar final mughal emperor exiled burma myanmar india former british ruler destruction really heartbreaking said artist srishti rana menon one hauntingly beautiful place sic jaalis nothing short exquisite wrote instagram whats lost really priceless absolutely shocked hurt angry whatever took place wrote another user considering lucky witness vandalism palace free enter public among heritage site delhi protected governmentrun archaeological survey india asi praveen singh superintending archaeologist asi delhi circle said nothing huge vandalism zafar mahal doubt damaged whole monument vanished told cnn adding complaint lodged police delhi southern mehrauli district arrest made authority investigate case mehrauli police official said delhi metropolis million people seen rapid urbanization past decade vast sparkling modern metro system burgeoning middle class come symbolize india economic growth meandering road stand centuriesold monument medieval ruin put indian capital par world great ancient city historian say one rare city scale heritage rome cairo said dalrymple grew delhi sheer extent history remarkable also questioned vandal able cause damage monument capital one greatest tragedy happening delhi whats happening outside said even vandalism zafar mahal fallen disrepair weed growing around tomb dirt wall structure according amateur historian shah umair asi promising year renovate palace step taken far said umair discovered damage alongside dalrymple singh asi said agency expected start repair zafar mahal april work delayed cost estimate set regular preservation conservation monument required singh said mughal empire ruled india reached pinnacle emperor shah jahan commissioned iconic taj mahal agra mausoleum wife built wealth artistic marvel across country destroyed jaali zafar mahal considered archetype indoislamic art movement flourished mughal rule recent year however rise hindu nationalism india put mughalera monument center controversy example anger erupted unescolisted taj mahal excluded official government tourism brochure former tourism minister alphons kannanthanam dismissed controversy time stating taj mahal india pride many however controversy showed creeping intent ruling hindunationalist bharatiya janata party exclude india former muslim ruler cultural history historian rana safvi said mughal architecture found south delhi mehrauli district unique india called authority protect typical indoislamic art dont see anywhere region said safvi referring wellknown motif like marble jaalis floral symbol safvi also urged authority protect lesser known equally significant monument take pride history', 'india southern state tamil nadu braced downpour heavy monsoon rain brought deadly flash flood submerged entire neighborhood earlier week indian meteorological department met forecast heavy rain millimeter inch fell part state sunday monday thats nearly half tamil nadu typically receives year according cnn weather analysis met data least people died flooding tamil nadu chief secretary shiv da meena told reporter tuesday downpour also occurred state million people reel aftermath cyclone michaung battered tamil nadu earlier month killing least people video photo broadcast local television week showed people wading waistdeep muddy water indian air force helicopter dropping food essential item people stranded rooftop according video published country defense ministry deluge weakened since december condition improving rescue crew india northeast monsoon usually run october december bringing heavy rain particularly south part southern india recent year experienced spell extremely heavy rainfall destroying house flooding road claiming dozen life least people killed heavy rain battered region month earlier people killed another spell downpour tamil nadu flood occur regularly india monsoon season expert say climate change increasing frequency severity world populous nation one country worst affected climate crisis according intergovernmental panel climate change potentially affecting billion people nationwide', 'uninitiated idea playing test cricket match five day result might end draw sometimes hard fathom thursday india accelerated format cricket defeating south africa seven wicket shortest completed test match ever requiring le two day secure victory match become shortest decided game term over bowled history test cricket over bowled across four inning match held newlands cricket ground cape town south africa remarkable first day action wednesday wicket fell south africa bowled run toppling india india lost final six wicket run remarkable collapse south africa returned crease start second inning ended opening day day two thursday saw despite protea aiden markrams valiant effort stem flow wicket falling markram scored remarkable ball team next highest score run south africa eventually fell leaving india modest run target win match touring side chased target over finishing level series test match displaced game australia south africa shortest test time per espncricinfo stats', 'starbucks plan operate store india opening equivalent one new location every three day next four year demand coffee burgeons world fastestgrowing major economy achieve target company focus opening coffee shop smaller city also double workforce country said statement tuesday starbucks currently store india operated year joint venture tata consumer product unit one india bestknown conglomerate first location mumbai opened twolevel store furnished local furniture decorated image metropolis apart usual range latte frappuccinos starbucks also offer beverage call indianinspired including south indian filter coffee masala chai blend sweetened milky tea spice many indian still begin day steaming cup chai appetite coffee growing local international chain proliferating across country visit india week starbucks ceo laxman narasimhan said india become one starbucks fastestgrowing market world south asian nation growing middle class becoming key market western company especially time china grappling number economic problem including muted consumer demand however despite slowdown china latter still starbucks largest fastestgrowing overseas market beverage giant aim store nearly currently last year starbucks poured million new campus eastern part china time opening september company said largest investment ever made coffee manufacturing distribution center outside united state', 'next week could shape future walt disney world populous country star india network among crown jewel disney dis acquired bought st century fox rupert murdoch billion five year ago blockbuster deal magical kingdom took fox business india gaining new audience million people south asian country one world hottest medium market disney hasnt happily ever hoping ceo bob iger admitted earnings call late last year part business india challenged u house mouse hit particularly hard lost digital right stream hugely popular indian premier league ipl cricket match billionaire mukesh ambanis conglomerate u company trying salvage india dream disney ambanis reliance industry reportedly discussing combining indian medium business form entertainment behemoth indian tycoon would upper hand company appointed law firm started antitrust diligence merger reuters reported last week citing unnamed source ambanis energytotelecom conglomerate would disney would hold remaining economic time reported december citing unnamed source merger likely completed next month indian newspaper added disney respond cnns request comment reliance declined comment disney search partner world fastest growing major economy come time burbankheadquartered company facing range problem home turf like competitor yearold hollywood stalwart face uncertain environment united state viewer increasingly tune linear tv favor tiktok youtube disney hit particularly hard big miss box office corporate upheaval iger said november company looking expansively india considering option also added would like stay market easy see relatively free market vast english speaking population india attractive country global entertainment company prime minister narendra modis government expects nation soon become world third largest medium entertainment market fifth currently fox acquisition disney served market platter star india built vast audience spending billion right broadcast india biggest sport including country national obsession cricket beat facebook meta sony sony billion deal five year ipl one world valuable sport property network big advantage local content country nearly two dozen language spoken star india offer tv channel language still disney struggled seize opportunity tv business well india iger said november company struggling area streaming app hotstar shed million subscriber since lost ipl right reliance almost two year ago march hotstar suffered another blow stopped streaming hbo content week later warner bros discovery wbd parent company hbo cnn moved content ambanis jiocinema taking loyal indian viewer hit show game throne succession along apart loss ambani analyst questioned disney strategy india particularly aggressive spending sport company entertainment asset would attractive acquirer partner disney india sport business faced challenge noted mihir shah vice president research firm medium partner asia disney lost digital right ipl match retain tv right paying billion also kept right show international cricket council tournament another staggering billion shah said financial difficulty business continue coming year largely attributed disney aggressive bidding renewing right added medium giant also failed fully capitalize streaming service technical prowess loss ipl also limited investment local entertainment content shah said american company misstep come time competition intensifying india potential reliancedisney deal isnt merger work sony india zee entertainment talk two year merge operation create billion giant fate deal still unclear analyst say corporate marriage key achieving scale competing global streaming giant netflix nflx amazon amzn established big presence india potential deal sign india entertainment industry entering phase consolidation handful player deep pocket able operate said aliasgar shakir analyst motilal oswal financial service november earnings call iger said disney want continue india also aim see whether strengthen hand improve bottom line ambani asia second richest man billion deepening medium ambition help disney merged entity would massive tv channel two streaming platform early interpret disney scaling back india shah said contour deal still unknown looking like partnership reliance industry disney could also start power couple go beyond medium industry insider speculating combined push theme park remember company business interest beyond medium entertainment partnership could start something bigger shah said', 'boeing getting respite month terrible news india newest airline purchasing boeing max aircraft first major order troubled planemaker announced since dramatic alaska airline incident part fuselage max fell midflight akasa air ordered max max aircraft delivered airline said event southern indian city hyderabad order include max plane grounded u regulator since inflight blowout boeing said order finalized december efficiency versatility support akasa air expansion meet soaring demand air travel region many year come stephanie pope boeings chief operating officer said statement thursday earlier month alaska airline flight carrying people made emergency landing shortly takeoff portland oregon part wall new aircraft detached left gaping hole side plane huge new blow reputation max grounded month crash indonesia early ethopia killed people board flight boeings ba share plunged almost month analyst well fargo week warned ongoing investigation alaska airline incident open whole new worm aircraft manufacturer facing quality control engineering problem year wednesday u federal aviation administration said expanding probe include spirit aerosystems spr build fuselage max another highprofile embarrassment boeing week u secretary state antony blinken forced change plane return washington world economic forum davos switzerland modified boeing plane suffered traveling press told critical failure related oxygen leak airshow india vinay dube founder ceo akasa air said airline pursues highest standard safety global aviation large historic aircraft order put akasa path becoming one top leading airline world turn decade added akasa air launched indian carrier buying lot boeing plane last year air india said would buy boeing plane order one largest boeings history announced president joe biden call indian prime minister narendra modi india aviation market growing rapid pace modis government belief become world largest', 'saffron flag flying majority hindu town ayodhya excited local prepare host indian prime minister narendra modi inauguration new multimilliondollar temple like many town muslim yearold maulana badshah khan say hell staying home fear repeat religious violence erupted year ago hindu nationalist destroyed babri masjid th century mosque triggering riot across country monday modi officially open ram janmabhoomi mandir lavish temple built site analyst say monument hindu nationalist ambition khan say belief celebration clear sign muslim becoming marginalized leadership modis bharatiya janata party bjp wound babri mosque demolition always even feel despondent voicing said khan temple hold symbolic value showing muslim place new india people invited ayodhya attend ceremony person including highprofile politician flying across vast country take part ten thousand pious hindu thronging small town place flower gift inside temple amid increasing crowd apprehension among ayodhyas muslim azam qadri yearold head local religious body said lived violence fear arrival outsider every time people come outside trouble one afford lose precious belonging saving identity paper anymore easy restart life said haji mahboob lost two relative violence said local muslim worried emboldened crowd chant provocative slogan call muslim expelled ayodhya demand hindu rashtra nation said mahboob said sense hopelessness among community many muslim believed supreme court granted hindu permission build temple contested site controversy might come end said instead emboldened hindu began targeting even mosque across country campaigning tear hindu stand u see u said modi rose power pledge reform country economy usher new era development also heavily pushed hindutva agenda ideology belief india become land hindu many state passed legislation critic say rooted hindutva discriminatory toward muslim including law make increasingly difficult interfaith couple marry banning slaughter transport cow animal considered sacred hindu one modis key promise voter build ram temple desecrated mosque site he hoping construction firm chance rare third election win year nilanjan mukhopadhyay author demolition verdict book mosque demolition said modis decision preside monday festivity sign hindu hegemony india modis involvement ceremony indicative line state religion becoming increasingly blurred said mukhopadhyay added sentiment among muslim one celebration heard muslim advising travel train drive car alone wear clothing identifies muslim going immense sadness also going tremendous fear among india muslim said speaking cnn bjp spokesperson nalin kohli said muslim marginalized india inauguration ram mandir cause celebration term prime minister modis government work initiative developmental agenda one scheme program anything distinguishes indian citizen basis religion caste region said year following demolition babri masjid hindu nationalist rallied build ram mandir site destroyed mosque setting stage emotional politically charged showdown worried indian liberal feared outburst sectarian violence judgement paved way building ram mandir also allocated land town muslim build another mosque kilometer mile ram mandir village called dhannipur mahboob one petitioner fought babri mosque supreme court said muslim ayodhya construction hold emotional sway built mosque close babri stood could tried convince injustice happening u however land far away construction even begun say late last year arafat shaikh member bjp appointed lead construction mosque told cnn never stepped foot ayodhya took role said delay due disagreement design building belief new mosque unique first india five minaret amongst ambitious plan shaikh say want additional land educational institution vegetarian kitchen footlong quran painted saffron color closely associated hinduism one become increasingly politicized appropriated hinduright shaikh said color chosen gharib nawaz famous sufi saint also revered saffron bring two community closer said rightwing hindu nationalist organization say temple opening symbol new hindu nation mughals tried change u english tried change u inauguration ram temple show world hindu tradition practice belief still intact new india see resurrection hindu civilization said vinod bansal spokesperson rightwing vishwa hindu parishad group mahant jairam da local head rashtriya swayamsevak sangh parent organization bjp said believe mosque built ayodhya instead claiming call build structure call war go saudi arabia pakistan muslimdominated country build mosque like india said hassan ali nine spent two night local police station trying escape violence sectarian fault line clearly visible many story local hindu muslim helping however lot poison fed people said one tell anymore one tell people heart', 'top pakistan official thursday accused india killing two pakistani national soil citing called pattern alleged indian assassination plot overseas raising tension two nucleararmed rival news conference pakistani foreign secretary muhammad syrus sajjad qazi claimed islamabad credible evidence linking indian agent killing muhammad riaz pakistanadministered kashmir september last year shahid latif city sialkot near kashmir month later killingsforhire case involving sophisticated international setup spread multiple jurisdiction qazi said accused two indian agent recruiting assassin kill men slain vicinity mosque provide information victim suggest new delhi might targeted india long accused pakistan harboring terrorist particularly disputed heavily militarized region kashmir country claim entirety de facto border called line control divide new delhi islamabad dispute led three war two nation qazi thursday compared alleged killing pakistan purported recent indian assassination plot north america fit pattern similar case come light country including canada united state qazi said clearly indian network extrajudicial extraterritorial killing become global phenomenon india ministry external affair mea rejected qazis allegation calling pakistan latest attempt peddling false malicious antiindia propaganda world know pakistan long epicenter terrorism organized crime illegal transnational activity india many country publicly warned pakistan cautioning would consumed culture terror violence mea spokesperson randhir jaiswal said pakistan reap sow last november u accused indian government official involved conspiracy kill american citizen home soil revelation came little month canadian prime minister justin trudeau publicly alleged india may involved murder canadian national territory prompting furious response new delhi diplomatic fallout two country indian government denied involvement american canadian case set highlevel committee investigate accusation u', 'consecration controversial hindu temple symbolizes seismic shift india secular founding value analyst say prime minister narendra modi disregard norm separating religion state push win rare third term year modi presided lavish opening ceremony last month ram janmabhoomi mandir sacred town ayodhya fulfilling longstanding promise voter helped propel hindunationalist bharatiya janata party bjp power today beginning new time cycle modi said new temple honoring hindu deity lord ram century waiting ram arrived modis vision divine india far cry idea modern country founding father nearly decade power prime minister enveloped language religion pursuit hindu nationalist agenda isolating million among india sizable religious minority moment culmination political project year making new departure india longer secular republic said political scientist gilles vernier senior fellow centre policy research new delhi india becomes de facto hindu nation task building national hindu religious symbol fall state leader officiates simultaneously prime minister head priest country ram janmabhoomi mandir stand site babri masjid th century mosque destroyed hardline hindu setting wave deadly sectarian violence seen india since bloody partition temple inauguration attended thousand handpicked guest including cricket legend sachin tendulkar billionaire industrialist mukesh ambani streamed live million across country ayodhya billboard celebrating temple opening featured image hindu deity ram alongside modis face leader bjp even dubbing prime minister king god modi fasted day purification ritual event visited temple across country performing custom sacramental india majority faith publicly called instrument lord ram anointed divine represent people india consecration modi presided pran pratishtha unveiling muchanticipated ram idol taking role typically reserved priest move well received quarter supporter praising leader action hindu modis action represent betrayal religion political capital obviously electoral stunt happening name faith said indian american activist sunita viswanath member usbased hindu human right group statement day temple opening modi priest leading ceremony political gain technically morally wrong weaponization religion trample whats left india secular democratic value yet blurring line state religion modi achieved predecessor unable analyst say much solidifying image india someone devout someone delivers promise said foreign policy editorinchief former cnn new delhi bureau chief ravi agrawal popular move criticized remains popular country hindu modi rose power pledge reform india economy usher new era development also heavily pushed hindutva agenda ideology belief india become land hindu stood reelection modis hindutva policy became brazen according analyst month winning announced stripping statehood india muslimmajority territory jammu kashmir turning two union territory bringing federal control later year government passed controversial citizenship law considered many discriminatory muslim reiterated party desire build ram temple contested holy site many hindu believe babri masjid built ruin hindu temple allegedly destroyed babar first mughal emperor south asia year rallied tear mosque make way temple dispute reached climax spurred bjp rightwing group hindu hardliner attacked mosque triggering widespread communal violence killed people nationwide victory modi supporter india supreme court granted hindu permission build temple ending decadeslong dispute dealing blow million muslim fear religious division becoming pronounced modis bjp government indian government denies discriminating minority analyst say last week festivity emboldened rightwing hindu act impunity minority communal tension rose western maharashtra state three reported altercation hindu muslim according local police separate incident central madhya pradesh state group rightwing hindu seen placing saffron flag top christian church color closely associated hinduism india become majoritarian india become nationalist india become prohindu agrawal said partly due government ability point india history wrong perceive india faced since assuming power nearly decade ago modi positioned disrupter india colonial legacy often speech marked emotive language emphasized need liberate india slavery mindset making step steer country away government called vestige british rule similarly modi also made comment india erstwhile islamic ruler mughals ruled much country many hardline hindu believe era period oppression muslim rule view also echoed member bjp hindu group decade claimed mughals destroyed hindu temple building mosque monument place many case debated court across country move indian liberal fear could spark violence disharmony week court city varanasi ruled hindu pray inside disputed gyanvapi mosque built former mughal ruler aurangzeb purportedly site destroyed hindu temple another major religious dispute aurangzeb severed many head could shake faith modi said speech referring ruler died three century ago country head toward national election expected held april may government see addressing injustice said foreign policy agrawal gilles political scientist said last week display hindu nationalism ram temple show strength alliance bjp india business cultural elite temple inauguration dark day india religious minority said officially become secondclass citizen', 'india prime minister narendra modi wednesday inaugurated abu dhabis first hindu temple largest united arab emirate boosting credential global statesman month head poll nationwide election seeking rare third term power bap hindu mandir built sprawling acre site abu dhabi desert city first traditional hindu stone mandir pink sandstone column topped seven spire representing number sheikh rule emirate modi greeted hug country president sheikh mohamed bin zayed al nahyan presented guard honor showing close two nation come strategic economic relation islam official religion uae modis trip come muslim india say feel marginalized threatened modis bharatiya janata party hindu nationalist policy gain momentum world largest democracy yet analyst expect present issue modi visit given india rising prominence growing economy strategic position global stage back home analyst say modis leading role temple inauguration could give party boost buildup election month time decade power modi created image protector hinduism said new delhi based political analyst asim ali taking message beyond india border he hoisting flag hinduism across world like religion pride india going overseas ali said he sending message world india protector hindu religion opening abu dhabi temple come week modi inaugurated controversial ram mandir temple built foundation centuriesold mosque torn hardline hindu crowd early northern india ceremony seen seismic shift away modern india secular founding principle hailed hindu nationalist crowning moment decadeslong campaign reshape nation temple opening abu dhabi must seen political operation modi visvis constituency abroad home said nicolas blarel associate professor international relation institute political science leiden university netherlands temple storied history uaebased devotee long advocated construction said blarel focus foreign policy south asia middle east country million home million indian national alone making largest population indian citizen world uae government allocated land development temple major win diaspora became joint venture indian uae government entrusted bap swaminarayan sanstha responsibility build administer mandir blarel said nine year later important cultural political project materialized framed bjp modi evidence global reach india religious cultural heritage also cooperation diaspora indian prime minister seventh trip islamic country third last eight month indicating importance modi lead election indiauae relation grown economic term perhaps significantly strategic security term said kadira pethiyagoda former diplomat brookings fellow product india seeking expand strategic reach middle east secure energy source added pethiyagoda author indian foreign policy cultural value bid boost industrial sector lift export indian government sought sign free trade deal move thats warmly received around world last september u president joe biden modi leader uae saudi arabia announced launch new trade route connecting india middle east europe railway port india billionaire including mukesh ambani gautam adani also invested big money country last august uae among six country invited join brics bloc india part uae like gulf state trying recent year diversify strategic security partnership great power said pethiyagoda mean moving away sole reliance u toward partnership india china russia given reason analyst say little chance uae bring criticism modis government treatment muslim bring private pethiyagoda said gulf government always taken practical approach noninterference domestic affair major power also unclear much extra support uae temple give prime minister already popular devout hindu follower bap temple win much adulation amongst bjp supporting hindu ram mandir ayodhya already garnered pethiyagoda said may however win vote south much india gulf workforce hail sends remittance bjp weaker added could also promoted properly help improve reputation amongst muslim voter could portrayed hindumuslim friendship', 'india main opposition party indian national congress said friday bank account frozen federal tax authority week expected nationwide election authority later temporarily restored access account party said move drew widespread criticism congress ally accused government conducting attack democracy congress treasurer ajay maken told news conference account frozen investigation party income tax return financial year added income tax department issued payment demand billion rupee million connection ongoing probe congress party bank account havent frozen democracy frozen maken said one party system country later friday income tex tribunal granted party temporary access account ahead hearing wednesday lawyer congress lawmaker vivek tankha confirmed cnn cnn reached india income tax department finance ministry multiple leader within congress party comment friday congress supporter gathered delhi iconic jantar mantar area protest action party chief mallikarjun kharge called judiciary save multiparty system country protect india democracy ally congress party repeatedly claimed democracy stifled india friday case latest series investigation targeting prominent opponent prime minister narendra modis ruling bharatiya janata party bjp former congress leader rahul gandhi scion political dynasty produced three prime minister last year disqualified lawmaker handed twoyear jail sentence defamation ruling supporter say politically motivated gandhi later reinstated lawmaker following intervention india supreme court gandhi friday said party fight tooth nail protect india democracy never bowed dictatorship ever bow wrote x formerly twitter congress party formidable force indian politics languishing poll since modi rose power decade ago promise economic growth prosperous society congress party attempted form alliance opposition group bid unseat modis bjp upcoming election alliance already starting see crack emerge support bjp surge week estimated million indian expected head poll often called world largest experiment democracy indian election come crucial time country gain prominence world stage modi presenting statesman cementing south asian nation modern superpower tension running high home minority feeling sidelined bjps majoritarian hindu nationalist policy recent gauge voter sentiment congress party lost three four regional vote key state election december giving boost modi bjp developing story follow', 'disney joining force asia richest man create new medium giant india say reach domestic audience million people billionaire mukesh ambanis reliance industry disney combined digital streaming platform tv channel country joint venture worth billion company said statement wednesday talk potential deal swirling time disney struggled take advantage opportunity presented nation billion people english widely spoken company also facing multitude problem back home disney dis made major push country acquired st century fox including vast star india network reliance stake merged entity directly viacom controlling stake disney hold rest landmark agreement herald new era indian entertainment industry said mukesh ambani whose sprawling business empire worth billion span retail technology renewable energy ambanis wife nita ambani chairperson joint venture combine disney acclaimed film show viacoms renowned production sport offering company said adding venture would also cater indian diaspora across world india world populous market excited opportunity joint venture provide create longterm value company said disney ceo bob iger disney faced multiple challenge india boast vibrant medium entertainment sector house mouse hit particularly hard lost digital right stream hugely popular indian premier league cricket match ambanis conglomerate disney streaming app india hotstar shed million subscriber since suffered another blow last march stopped streaming hbo content week later warner bros discovery wbd parent company hbo cnn moved content ambanis jiocinema taking indian viewer hit show game throne succession along latest financial year ending september disney earned revenue cent average per hotstar subscriber cent compared nonus subscriber disney main streaming service according annual report hotstar subscriber number plunged million latest financial year earnings call november iger said disney tv business well india part business country struggling opportunity strengthen hand said time wed like stay indian market deal one india richest conglomerate allows disney achieve goal according analyst merger creates powerful entity poised enhance scale profitability competitiveness converged tv streaming video landscape said mihir shah vice president research firm medium partner asia story updated additional information', 'month ago nita ambani stepped away frontline business focus philanthropy wife asia richest man back biggest job ferociously competitive world india medium chairperson billion behemoth formed merger disney dis medium interest india reliance industry business empire controlled billionaire mukesh ambani yearolds new position come le two year husband set succession plan motion billion conglomerate span retail technology renewable energy plan hand empire three child nita ambani steering medium giant alone uday shankar former president disney asia pacific unit vice chairperson joint venture provide strategic guidance company said statement wednesday disneyreliance deal expected completed late early provided regulator shareholder grant approval nita ambani first stepped business limelight year ago becoming first woman director board reliance industry appointment made month regulator urged listed indian company improve gender diversity boardroom criticized analyst tokenism resigned board last august devote energy time reliance foundation focus philanthropic initiative group company said statement time described educationist philanthropist businesswoman group ambanis wife already extensive experience promoting sport art country owner mumbai indian cricket team hugely popular indian premier league ipl sport key success new medium venture disney india facing multiple challenge merger announced hit particularly hard lost digital right stream ipl cricket match reliance resulting loss million subscriber trained indian classical dance form bharatnatyam founded nita mukesh ambani cultural centre mumbai last year brought india first major broadway show sound music may married mukesh ambani since also member international olympic committee honorary trustee board metropolitan museum art new york three child daughter isha two son anant akash also playing bigger role family business even though mukesh ambani insisted laying succession plan intention retiring time soon never differentiated isha akash anant whatever boy could daughter could nita ambani said last year rare tv interview asked promoting woman corporate india young girl breaking shattering glass ceiling give opportunity shine said adding equal pay necessity corporate world representation woman indian boardroom improved significantly last decade nearly top company listed national stock exchange one female board member according report consultancy ey however le company female chairperson still room improvement ey added report published new job title nita ambani become powerful woman indian medium entertainment sector worth billion growing rapidly reliance stake merged entity directly viacom controlling stake disney holding rest joint venture expected reach domestic audience million viewer company said statement', 'charter jet tycoon celebrity performance rihanna direction jungle fever outfit visiting animal rescue center thats billionaire mukesh ambani kickstarting big fat indian wedding celebration son head indian conglomerate reliance asia richest man net worth billion according forbes ambani known never party half occasion moment india transfixed threeday prewedding jamboree youngest son anant ambani set marry longtime girlfriend radhika merchant begin friday entertain guest take place township jamnagar near reliance main oil refinery western state gujarat wedding take place july festivity de facto began wednesday anant bridetobe hosted communal dinner local villager served addition pop rb superstar rihanna u magician david blaine perform also traditional ceremony temple complex anants prewedding bash likely attended bill gate metas mark zuckerberg fellow indian billionaire gautam adani kumar mangalam birla well many cricketer bollywood film star shah rukh khan salman khan according two source direct knowledge plan declined identified also likely attend bob iger ceo disney announced billion merger india medium asset reliance wednesday anant director reliance new energy business one ambanis three child heir empire merchant director india encore healthcare ambani also made headline daughter ishas wedding described minidavos summit former u secretary state hillary clinton john kerry among hit dance floor bollywood star back guest anant ambanis celebration set savor dish created around chef pampered hair styling makeup artist indian wear drapists available first come first serve basis according planning document provided invitee seen reuters visit rescue center reliance say home animal one world largest elephant hospital dress code jungle fever guest advised wear animal print clothing hawaiian shirt return charter jet flight new delhi mumbai offer guest asked limit two luggage item three suitcase per couple document caution bring there guarantee arrive flight guest also asked reasonable expectation laundry service clothes given steam press returned within hour expecting requesting anything quicker may feasible said', 'police eastern india arrested three men alleged gang rape foreign tourist assault husband hunt four suspect case highlight country decadeslong struggle curb sexual violence woman couple traveling motorcycle state west bengal neighboring nepal found late friday police officer patrol said pitambar singh kherwar superintendent dumka district police jharkhand state taken hospital woman told doctor raped said police know identity wanted suspect formed special investigative team kherwar said unclear whether three arrested suspect legal representation cnn naming female victim accordance indian law prevents naming victim sexual violence police disclosed name nationality couple arrest come travel vlogger couple saturday posted instagram account knife held throat attack india woman raped brought hospital dna testing said couple post spanish woman say instagram page brazilian instagram story woman showed bruise face saying face look like isnt hurt thought going die followup post sunday couple thanked follower support saying well police everything possible catch remaining suspect cnn reached couple comment india national commission woman ncw condemned alleged attack ncw chairperson rekha sharma spoken victim extended required assistance organization posted social platform x saturday jharkhand minister mithilesh kumar thakur called alleged assault condemnable incident crime committed culprit spared said saturday india struggled year tackle high rate violence woman number highprofile rape case involving foreign visitor drawing international attention issue british woman allegedly raped walking hotel western state goa popular tourist destination two year earlier american woman allegedly drugged raped group men fivestar hotel room new delhi six men sentenced life prison gang rape swiss tourist according india national crime record bureau total rape case recorded average case per day expert warn number case recorded small fraction may real number deeply patriarchal country shame stigma surround rape victim family india current law still crime man force sex sexual act wife long perhaps india infamous case recent year gangrape medical student beaten tortured left die following brutal attack public bus new delhi case ensuing nationwide protest drew international medium scrutiny prompted authority enact legal reform rape law amended broaden definition crime set strict punishment rape also sexual assault voyeurism stalking despite change rape case remain prevalent country victim advocate saying government still enough protect woman punish attacker', 'lineup india sporting hero gained celebrity status cricket neeraj chopra anomaly fact champion javelin thrower barely bowled ball swung bat since childhood even though cant help notice similarity discipline country favorite pasttime feel india natural fast arm chopra tell cnn sport basic thing javelin throwing fast arm throw far south asian country home million cricket fan according report ormax medium chopras ability hurl javelin close foot earned sizeable following right stardom grown passing success winning olympic gold tokyo three year ago first asian athlete javelin chopra crowned world champion budapest last year becoming first indian win world title track field event kind perhaps come surprise given historic career date understated yearold proven hit back home many people country support watch competition say chopra even im competing diamond league stay awake one two night watch competition really amazing people support athletics popularity august celebrated national javelin day india anniversary chopra winning olympic gold medal tokyo prior chopra india practically history track field winning olympic medal field hockey wrestling genuine excitement time superstar athlete competes jockey runway quivering javelin hand nation hold breath ask anybody india favorite sportsperson moment unanimously answer neeraj chopra despite fact athletics popular sport veteran journalist norris pritam last year published book chopras life career tell cnn sport hand cricket popular sport common man sport people talk breathe eat cricket kind religion still neeraj chopra best sportsperson he hero he absolutely hero india growing rural village khandra northern state haryana chopra played cricket volleyball another sport fast arm useful weapon preteen year overweight child carrying recently described excess flab frequently told sport wasnt changed soon caught sight javelin shimmering sky athletics stadium nearby city panipat random thing background family whole village nobody athlete nobody know javelin throwing say chopra go stadium really like javelin fly going like fish swimming sea really like thing even throw always enjoy javelin fly enough thenyearold chopra inspired try sport father uncle part family farmer pooled together enough money buy steel javelin plowing field near family home chopra attempted first throw seed soon sown fledgling athlete would spend hour hour launching throw throw started without training without family background without sort formal school could javelin thrower say pritam fact school village culture sport ground sport proper coaching chopra relied youtube video hone technique talent soon became evident moving sport academy city panchkula arm speed got faster throw international competition quickly followed today chopras resum long impressive boasting two asian game title commonwealth game gold medal top world olympic triumph favorite defend title paris olympics get underway july though he allowing thought success enter head stage main focus stay healthy say chopra im taking pressure winning gold medal ill training heart ill give best day decide win medal performing pressure particularly huge fanbase back home india common theme chopras career popularity doesnt faze instead feeding selfbelief right final throw pressure add feel really good people talking say okay neeraj win medal neeraj throw far always take thing positively first world title previously obvious gap trophy cabinet last year chopra want throw beyond meter foot describes magical mark javelin throwing men history ever thrown meter chopras personal best six centimeter inch shy distance matter rather tick goal india however nothing prove status national hero already secured everybody know neeraj say pritam poster billboard he book magazine everything put tv find neeraj talking he brand ambassador hundred thing sanitation education selling product television moving billboard one say win second gold medal olympics treated like god add pritam chopra rarely overwhelmed adulation taking public recognition stride nearly always train competes outside india back home often struck farreaching influence people meet always say son daughter also start javelin athletics also win medal india say chopra good thing', 'india said monday joined world top nuclear power mastering ability put multiple warhead atop single intercontinental ballistic missile successful test multiple independently targetable reentry vehicle mirv technology indigenously developed agniv icbm put india club includes united state russia china france united kingdom neighboring pakistan also claimed mirv technology expert say claim unverified indian prime minister narendra modi praised country scientist development one series announced government month national election prime minister seeking rare third consecutive term power proud drdo defence research development organisation scientist mission divyastra first flight test indigenously developed agni missile multiple independently targetable reentry vehicle mirv technology modi said x monday indian scientist conducted test facility abdul kalam island bay bengal india northeast coast defense ministry said statement various telemetry radar station tracked monitored multiple reentry vehicle mission accomplished designed parameter statement saidindia give exact number reentry vehicle released agniv test mirved missile carry dozen possibly mirv warhead indian defense minister rajnath singh congratulated scientist team behind test india proud wrote x home minister amit shah called momentous day nation adding technology accelerate modis vision selfreliant bharat india warhead released space rocket missile carried aloft programmed hit separate target kilometer mile apart according center arm control nonproliferation overall agniv missile range kilometer mile according center strategic international study missile defense project put india rival like china pakistan well within range weapon mirv technology new united state first deployed minuteman iii icbm according national museum u air force minuteman iii designed carry three warhead u missile carry one comply arm control treaty russia mirved missile present problem ballistic missile defense interceptor missile contend number warhead traveling target hundred mile apart also considered destabilizing weapon according expert present tempting firststrike target creates use lose scenarioan incentive strike first time crisis otherwise first strike attack destroyed country mirved missile would disproportionately damage country ability retaliate according website posting union concerned scientist india announcement mirv capability come rival china expanding nuclear force similar fashion u defense department report congress china military said beijing developing new icbm significantly improve nuclearcapable missile force require increased nuclear warhead production partially due introduction multiple independently targetable reentry vehicle mirv capability china india share disputed border himalaya deadly clash occurred recently', 'british singer ed sheeran wowed stage mumbai diljit dosanjh weekend pair sang rendition indian artist hit track lover punjabi duet promptly lit india social medium artist posted video performance instagram huge cheer crowd sheeran invited singing partner onto stage got bring diljitdosanjh tonight mumbai sing punjabi first time ive incredible time india come sheeran wrote post singer actor dosanjh one biggest south asian artist worldwide played major role putting punjabi music map internationally made history first punjabi artist perform coachella past year sheeran gone selfreleasing eps playing small bar london one modern pop bestselling song writer middle mammoth twoanda halfyear long global mathematics tour concertgoers packed mahalaxmi racecourse see dosanjhs surprise appearance indian star hyped audience sheeran rocked acoustic guitar singing part chorus idea diljit going felt like moon descended fan said video recap posted indian star still processing happened never going forget night another concertgoer said spotify india commented video posted sheeran saying see fire stage reference one british star hit fan indian celebrity social medium praised dosanjh getting sheeran sing indian language bollywood actor varun dhawan called global domination comment dosanjhs instagram video standing audience shrieking like fan girl multiverse event nobody expected commented comedian sahil bulla hat diljit wrote another user man literally blownup punjabi music scene making artist sing punjabi waiting collab dosanjh collaborated englishspeaking singer last year released bilingual track ha ha sia australiaborn artist behind chandelier elastic heart fan hoping international crossover indian artist one person commenting first sia ed diljit gon na make everyone sing punjabi india official language including english home hundred mother tongue punjabi native punjab region india well neighboring pakistan around world million speaker punjabi across south asian subcontinent diaspora community concert sheeran also thanked fan braving hour traffic reach event road leading sprawling acre venue equivalent roughly american football field jammed car five hour gig began pm saturday according cnn staff attended traffic bad fan left car walked final kilometer mile stretch entrance know india big place everyone concert necessarily mumbai people traveled long way today people got train plane drove got kid know lot go spending saturday night sheeran said beginning added made promise would return india', 'popular food delivery app india sparked controversy abrupt retreat launching initiative attract vegetarian customer earlier week zomato announced plan introduce curation restaurant serve pure vegetarian food list exclude restaurant serve nonveg food item company founder ceo deepinder goyal said x tuesday nonveg short nonvegetarian refers food contains meat poultry seafood india largest percentage vegetarian world goyal wrote x adding many particular food cooked handled order cater strict dietary preference food would delivered dedicated pure veg fleet rider would wear new green uniform opposed red one worn rest delivery worker company said mean nonveg meal even veg meal served nonveg restaurant never go inside green delivery box meant pure veg fleet goyal added many widely followed religion world populous country require adherence strict dietary law jains many hindu stick vegetarian diet following principle ahimsa nonviolence respect life almost four indian adult say vegetarian according pew research center dietary law deeply entrenched even instance property developer selling apartment meat eater restriction cow slaughter become politically contentious india recent year many among country majority hindu population consider animal sacred goyal emphasized new plan meant alienate religious political preference company received huge backlash social medium user criticized move casteist many hindu dominant caste tend vegetarian many people marginalized caste india caste system officially abolished yearold social hierarchy imposed people birth still exists many aspect life caste system categorizes hindu birth defining place society job marry zomatos move also slammed risky meat eater delivery staff unsafe illogical wrote one user x strengthens artificial divide vegnonveg eater enables colony identify persecute citizen basis food preference following outcry zomato rolled back plan dress rider delivering vegetarian food green rider regular fleet fleet vegetarian wear colour red goyal said wednesday ensure red uniform delivery partner incorrectly associated nonveg food blocked housing society rider physical safety paramount importance u said realise even customer could get trouble landlord would nice thing happened u added', 'world biggest election kick gear next week first ballot cast india mammoth national poll considered consequential decade potential shape country future nearly billion people eligible decide whether grant prime minister narendra modi rare third consecutive term office extend year rule bharatiya janata party bjp modi india become world fastest growing major economy pushing country billion people nearsuperpower status india also become increasingly polarized along religious line critic say another fiveyear term give modis rightwing bjp mandate continue policy transformed country secular republic hindufirst nation here need know largest election human history india national election giant exercise democracy logistics take month complete million people eligible vote population united state european union russia combined given size electorate isnt single date everyone vote instead polling unfold seven phase around country beginning april ending june vote country state eight union territory counted result released june multiparty firstpastthepost system indian vote fill seat lower house parliament called lok sabha two seat nominated country president party majority form government appoint one winning candidate prime minister modi bjp remain hugely popular widely expected secure another five year power yearold first elected prime minister roaring majority ticket development anticorruption credited implementing welfare social reform easily secured second term time apparent platform hindu nationalism election indian cast ballot highest voter turnout country history main challenger bjp indian national congress governed country much year since independence find doldrums effort prevent another modi win congress formed alliance opposition leader including major regional party indian national developmental inclusive alliance india launched campaign late last month platform saving democracy political scion rahul gandhi face congress party key figure include popular leader aam aadmi party delhi chief minister arvind kejriwal also regional heavyweight west bengal chief minister india trinamool congress mamata banerjee south tamil nadu chief minister muthuvel karunanidhi stalin vying block bjp takeover respective state modi yet break india bloc beset infighting yet put forward candidate prime minister analyst say struggling match kind star quality appeal projected modi opposition also pressure say campaign bjp weaken modis opponent several prominent member opposition including delhi chief minister kejriwal arrested investigated state agency move decried political party congress bank account frozen federal tax authority said asked pay million tax bjp denied political interference meanwhile another victory modi would cement place one india important longestserving leader modis party ally aiming win seat lok sabha giving enough vote change india constitution figure also improve absolute majority secured bjp seat congress party india world largest democracy decade bjp rule analyst say populist modi tightened grip democratic institution minority especially country million muslim say persecuted bjps hindunationalist policy modi accused silencing critic including cracking india independent medium annual world press freedom index india dropped th year modi came power nation last year list demolishing islamic monument renaming city founded ancient muslim ruler revising history textbook modis rule seismic shift india secular founding value hindufirst nation analyst say another bjp term push india along road modis supporter point economic record india poised become stcentury powerhouse economy rapidly expands say presence world stage including hosting g cemented country modern global power along historymaking moon landing modi also delivered many key promise crowning achievement inauguration controversial ram janmabhoomi mandir hindu temple holy city ayodhya built site destroyed babri mosque india economy rising world populous country failed generate enough job million young people struggling enter workforce unemployment among yearolds staggering end farmer also emerged biggest threat ruling party recent year repeatedly taking street demand stronger protection economic welfare including higher fixed price crop agriculture primary source livelihood india population farmer previously success strongarming modi repealing unpopular agriculture policy india election believed world expensive even outstripping u presidential election political party candidate regulatory body spent billion according delhibased centre medium study year spending expected surge past high peak himalaya remote forest india central state vote cast electronically across million polling station million election worker deployed traveling via road boat camel train helicopter reach every indian voter foot meter tashigang village far northern state himachal pradesh border china world highest polling station resident voted state government election candidate contested election year candidate still announced party fight seat lok sabha including six national state party party symbol allotted election commission appears ballot helping dozen party differentiate making easier voter make choice country around quarter population illiterate everyday object like ceiling fan comb mango ruling bjps symbol lotus congress party raised openpalmed hand kejriwals aam aadmi party common man party broom representing origin anticorruption street movement', 'elon musk postponed planned trip india citing heavy obligation tesla tesla ceo due arrive india next week visit expected include meeting prime minister narendra modi confirmation tesla build factory world populous country unfortunately heavy tesla obligation require visit india delayed much look forward visiting later year musk posted x saturday musk posted last week x looking forward meeting modi also planning meet leader indian space tech startup visit including pawan chandana cofounder skyroot aerospace launched india first privately developed rocket apart tesla tsla musk also bos rocket company spacex x formerly twitter currently world fourthrichest man musk expected announce investment billion billion india mainly building new tesla factory reuters financial time reported citing anonymous source trip originally scheduled day india commenced world largest democratic election estimated million people country billion eligible vote poll started friday take month complete modi widely expected clinch rare third consecutive fiveyear term prime minister', 'india one world top spacefaring nation first asian country reach mar orbit fourth planet take spacecraft moon landing closer south pole known cratered terrain anyone else time private sector played limited role space exploration acting mostly supplier vendor national space agency indian space research organization isro thats changing quickly owing raft government reform aiming boost private participation space sector united state proliferation private company like elon musk spacex boosted washington space ambition driving cost indian homegrown startup like skyroot aerospace launched country first private rocket leading charge commercialize india space sector bolster status space superpower inside sleek rocket hanger southern city hyderabad company preparing liftoff sevenstory tall vikram rocket take india first privately launched satellite orbit thatll major milestone u pawan chandana former isro scientist cofounded skyroot aerospace told cnn company globally able make orbit yet india pushing pioneering national initiative including plan space station put indian astronaut moon meanwhile prime minister narendra modi currently running third consecutive term made major push commercialize space activity recent year including allowing private enterprise take part easing approval foreign investment space sector company around world catapulting thousand satellite low earth orbit leo extends altitude kilometer mile form megaconstellations group small satellite work together beam broadband internet terrestrial user according analytics firm brycetech spacecraft launched small satellite thats driving exponential growth global space economy worth billion according report mckinsey company world economic forum could grow trillion small satellite market skyroot want capture plan offer customized launch satellite kilogram pound organization satellite often piggyback onto rideshare mission last year example one spacex journey powered falcon rocket carry kilogram pound leo launched small satellite orbit multiple company there often waiting period hitch ride satellite go rocket heading may disadvantageous satellite want reach particular orbit observe certain part planet model provides advantage spacex say cost getting kilogram pound payload leo per kilogram really competitive spacex offering say susmita mohanty director general indian thinktank spaceport sarabhai skyroot want offer dedicated launch per kilogram much expensive spacexs le bespoke service chandana say mass production better technology may enable cost reduction chandana compare spacexs ridesharing mission train go space skyroots operation cab get space take satellite customized location add dont directly compete spacex spacex bigger vehicle satellite customized launch solution price drop interest people wanting choose dedicated launch without compromising going rideshare mission said customized launch might useful want send smaller payload aiming reach diverse amount orbit might possible get ridesharing say kari bingen director aerospace security project center strategic international study thinktank based u could useful commercial company academic institution government willing pay price add chandana optimistic make affordable noting frugality dna india space program country mar orbiter budget million modi pointed cost le hollywood scifi thriller gravity chandana add company aim build reusable rocket eventually capability helped spacex reduce launch cost india home almost space startup one according report published deloitte period private investment company topped million skyroot already raised million backer like singapore sovereign wealth fund gic make india mostfunded space startup according data platform tracxn company work reach profitability need demonstrate consistently reach orbit interest turn contract help u scale say chandana yet spot vikram fully booked mix domestic international client indian space startup mature coming year expects government become customer enabling growth many private space company source income government including space x raked billion dollar u government contract india government allocated billion develop space prorgram interim budget comparison nasa received billion unless government back reform significant funding coming india difficult u rival development u market chinese market say mohanty add indian startup play meaningful role global marketplace need go beyond servicing space agency domestic mandate start bidding contract chandana sound challenge want become globally one top three launch company say fascination rocket become obsession']\n",
            "0     negative\n",
            "1      neutral\n",
            "2     positive\n",
            "3     negative\n",
            "4     negative\n",
            "5      neutral\n",
            "6     negative\n",
            "7     negative\n",
            "8      neutral\n",
            "9     positive\n",
            "10    positive\n",
            "11     neutral\n",
            "12    positive\n",
            "13    negative\n",
            "14    negative\n",
            "15    negative\n",
            "16    positive\n",
            "17    negative\n",
            "18    positive\n",
            "19    positive\n",
            "20    positive\n",
            "21    negative\n",
            "22    positive\n",
            "23    positive\n",
            "24     neutral\n",
            "25     neutral\n",
            "26     neutral\n",
            "27     neutral\n",
            "28    positive\n",
            "Name: sentiment, dtype: object\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QUVE7lOqzUnp",
        "outputId": "5b9af4d8-4939-4c60-b5a9-f70429454312"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[2, 0, 1, 2, 2, 0, 2, 2, 0, 1, 1, 0, 1, 2, 2, 2, 1, 2, 1, 1, 1, 2, 1, 1, 0, 0, 0, 0, 1]\n"
          ]
        }
      ],
      "source": [
        "from torch.utils.data import DataLoader, Dataset\n",
        "from transformers import AdamW, get_linear_schedule_with_warmup\n",
        "import numpy as np\n",
        "\n",
        "class NewsDataset(Dataset):\n",
        "    def __init__(self, texts, labels, tokenizer, max_len):\n",
        "        self.texts = texts\n",
        "        self.labels = labels\n",
        "        self.tokenizer = tokenizer\n",
        "        self.max_len = max_len\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.texts)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        text = self.texts[idx]\n",
        "        label = self.labels[idx]\n",
        "\n",
        "        encoding = self.tokenizer.encode_plus(\n",
        "            text,\n",
        "            add_special_tokens=True,\n",
        "            max_length=self.max_len,\n",
        "            return_token_type_ids=False,\n",
        "            padding='max_length',\n",
        "            truncation=True,\n",
        "            return_attention_mask=True,\n",
        "            return_tensors='pt',\n",
        "        )\n",
        "\n",
        "        return {\n",
        "            'text': text,\n",
        "            'input_ids': encoding['input_ids'].flatten(),\n",
        "            'attention_mask': encoding['attention_mask'].flatten(),\n",
        "            'label': torch.tensor(label, dtype=torch.long)\n",
        "        }\n",
        "\n",
        "# Example usage\n",
        "# take data formm file\n",
        "texts = articles_list\n",
        "labels = sentiment_list.to_numpy()\n",
        "# get labels in numeric\n",
        "labels = [label_mapping[label] for label in labels]\n",
        "print(labels)\n",
        "\n",
        "dataset = NewsDataset(texts, labels, tokenizer, max_len=128)\n",
        "# divide the dataset in test and training\n",
        "train_size = int(0.8 * len(dataset))\n",
        "test_size = len(dataset) - train_size\n",
        "train_dataset, test_dataset = torch.utils.data.random_split(dataset, [train_size, test_size])\n",
        "\n",
        "# DataLoader\n",
        "\n",
        "train_dataloader = DataLoader(train_dataset, batch_size=8, shuffle=True)\n",
        "test_dataloader = DataLoader(test_dataset, batch_size=8, shuffle=False)\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Set up optimizer and learning rate scheduler\n",
        "num_epochs = 3\n",
        "\n",
        "optimizer = AdamW(model.parameters(), lr=2e-5, correct_bias=False)\n",
        "total_steps = len(train_dataloader) * num_epochs\n",
        "scheduler = get_linear_schedule_with_warmup(optimizer, num_warmup_steps=0, num_training_steps=total_steps)\n",
        "\n",
        "# Fine-tuning loop\n",
        "model = model.to(device)\n",
        "model.train()\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "    for batch in train_dataloader:\n",
        "        optimizer.zero_grad()\n",
        "        input_ids = batch['input_ids'].to(device)\n",
        "        attention_mask = batch['attention_mask'].to(device)\n",
        "        labels = batch['label'].to(device)\n",
        "\n",
        "        outputs = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels)\n",
        "        loss = outputs.loss\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        scheduler.step()\n",
        "\n",
        "print(\"Fine-tuning completed\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_8UOjnQM3qBa",
        "outputId": "00b7dfde-12d4-46cb-fb61-ffa949c83a0b"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/transformers/optimization.py:521: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use the PyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Fine-tuning completed\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "model.eval()\n",
        "correct_preds = 0\n",
        "total_preds = 0\n",
        "\n",
        "with torch.no_grad():\n",
        "  for batch in train_dataloader:\n",
        "    input_ids = batch['input_ids'].to(device)\n",
        "    attention_mask = batch['attention_mask'].to(device)\n",
        "    labels = batch['label'].to(device)\n",
        "\n",
        "    outputs = model(input_ids=input_ids, attention_mask=attention_mask)\n",
        "    _, predicted = torch.max(outputs.logits, dim=1)\n",
        "\n",
        "    total_preds += labels.size(0)\n",
        "    correct_preds += (predicted == labels).sum().item()\n",
        "\n",
        "accuracy = correct_preds / total_preds\n",
        "print(f\"Initial model accuracy :{accuracy}\")\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bGlrQgoP_BF8",
        "outputId": "951939e6-fb9b-40aa-a1ff-1811e7bf341c"
      },
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Initial model accuracy :0.391304347826087\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model_save_path = 'drive/MyDrive/Sentiment_analysis/fine_tuned_bert_model.pt'\n",
        "\n",
        "# Save the model state dict and other information you may need\n",
        "torch.save({\n",
        "            'model_state_dict': model.state_dict(),\n",
        "            # Add other information if needed\n",
        "            }, model_save_path)\n",
        "\n",
        "print(f\"Model saved to {model_save_path}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "idB_su6Q_2r1",
        "outputId": "afba71a9-fac8-4ded-8072-0af6441756a8"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model saved to drive/MyDrive/Sentiment_analysis/fine_tuned_bert_model.pt\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "from transformers import BertTokenizer\n",
        "\n",
        "# Load unlabelled CSV file\n",
        "csv_file = 'drive/MyDrive/Sentiment_analysis/processed_Articles.csv'\n",
        "df = pd.read_csv(csv_file)\n",
        "\n",
        "# Extract text data\n",
        "unlabeled_texts = df['Processed_Article'].tolist()\n",
        "\n"
      ],
      "metadata": {
        "id": "nXv6UQXnBeJ7"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "encoded_unlabeled_texts = tokenizer(unlabeled_texts, padding=True, truncation=True, return_tensors='pt')\n",
        "self_train_dataloader = DataLoader(encoded_unlabeled_texts['input_ids'], batch_size=8)  # Use input_ids directly for self-training"
      ],
      "metadata": {
        "id": "hjbmOJM4CIMa"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "learning_rate = 2e-5\n",
        "batch_size = 8"
      ],
      "metadata": {
        "id": "KSTZOs2KCN6q"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# from torch.utils.data import DataLoader, Dataset\n",
        "# train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
        "# self_train_dataloader = DataLoader(encoded_texts, batch_size=8)\n",
        "confidence_threshold = 0.4\n",
        "\n",
        "model.train()\n",
        "for epoch in range(num_epochs):\n",
        "  train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "    # Self-training with unlabeled data\n",
        "  for input_ids in self_train_dataloader:\n",
        "        input_ids = input_ids.to(device)\n",
        "\n",
        "        with torch.no_grad():\n",
        "            outputs = model(input_ids=input_ids)\n",
        "            print(outputs)\n",
        "            logits = outputs.logits\n",
        "            probabilities = torch.softmax(logits, dim=1)\n",
        "            max_probabilities, predicted_labels = torch.max(probabilities, dim=1)\n",
        "            print(max_probabilities)\n",
        "            print(predicted_labels)\n",
        "\n",
        "        # Add confident predictions to labeled dataset for re-training\n",
        "        new_labeled_data = {\n",
        "            'input_ids': input_ids[max_probabilities >= confidence_threshold].cpu(),\n",
        "            'labels': predicted_labels[max_probabilities >= confidence_threshold].cpu()\n",
        "        }\n",
        "\n",
        "\n",
        "        # Convert tensors to list and append to existing dataset\n",
        "        texts_to_add = tokenizer.batch_decode(new_labeled_data['input_ids'], skip_special_tokens=True)\n",
        "        labels_to_add = new_labeled_data['labels'].tolist()\n",
        "        print(f\"Texts to add: {len(texts_to_add)}\")\n",
        "        print(f\"Labels to add: {labels_to_add}\")\n",
        "\n",
        "        print(labels_to_add)\n",
        "        texts.extend(texts_to_add)\n",
        "        labels=labels.tolist()\n",
        "        labels.extend(labels_to_add)  # Concatenate labels\n",
        "        # convert labels to np arrya\n",
        "        labels = np.array(labels)\n",
        "        print(labels)\n",
        "\n",
        "\n",
        "        print(f\"Total texts: {len(texts)}\")\n",
        "        print(f\"Total labels: {len(labels)}\")\n",
        "\n",
        "        assert len(texts) == len(labels), \"Lengths of texts and labels do not match\"\n",
        "\n",
        "        # Update dataset and dataloader with new labeled data\n",
        "        dataset = NewsDataset(texts, labels, tokenizer, max_len=128)\n",
        "        train_size = int(0.8 * len(dataset))\n",
        "        test_size = len(dataset) - train_size\n",
        "        train_dataset, test_dataset = torch.utils.data.random_split(dataset, [train_size, test_size])\n",
        "        train_dataloader = DataLoader(train_dataset, batch_size=8, shuffle=True)\n",
        "\n",
        "  for batch in train_dataloader:\n",
        "        optimizer.zero_grad()\n",
        "        input_ids = batch['input_ids'].to(device)\n",
        "        attention_mask = batch['attention_mask'].to(device)\n",
        "        labels = batch['label'].to(device)\n",
        "\n",
        "        outputs = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels)\n",
        "        loss = outputs.loss\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        scheduler.step()\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "xQjKqE-XCO-g",
        "outputId": "e332ee76-39b3-42b0-ea7b-af4e981ae450"
      },
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2820, -0.1112, -0.1235],\n",
            "        [-0.0861, -0.0268, -0.0388],\n",
            "        [-0.2345,  0.1159, -0.1165],\n",
            "        [-0.2809,  0.1104, -0.2192],\n",
            "        [-0.3468,  0.1699,  0.0677],\n",
            "        [-0.2065, -0.0765,  0.2390],\n",
            "        [-0.0583, -0.0641, -0.2126],\n",
            "        [-0.0598,  0.0949, -0.0456]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3533, 0.3412, 0.4005, 0.4175, 0.4001, 0.4219, 0.3507, 0.3669])\n",
            "tensor([1, 1, 1, 1, 1, 2, 0, 1])\n",
            "Texts to add: 4\n",
            "Labels to add: [1, 1, 1, 2]\n",
            "[1, 1, 1, 2]\n",
            "Total texts: 33\n",
            "Total labels: 33\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.3049,  0.1578, -0.0976],\n",
            "        [-0.3358,  0.0088,  0.1639],\n",
            "        [-0.2608,  0.1263,  0.1847],\n",
            "        [-0.2168,  0.0773, -0.0762],\n",
            "        [-0.0359, -0.0437, -0.2233],\n",
            "        [-0.0685,  0.0110, -0.1817],\n",
            "        [-0.2010,  0.1078, -0.1741],\n",
            "        [-0.1600, -0.0933, -0.3164]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4159, 0.4060, 0.3870, 0.3842, 0.3544, 0.3639, 0.4018, 0.3656])\n",
            "tensor([1, 2, 2, 1, 0, 1, 1, 1])\n",
            "Texts to add: 3\n",
            "Labels to add: [1, 2, 1]\n",
            "[1, 2, 1]\n",
            "Total texts: 36\n",
            "Total labels: 36\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.4009,  0.1662,  0.0478],\n",
            "        [-0.2347,  0.0545, -0.2089],\n",
            "        [-0.1096,  0.0173, -0.0681],\n",
            "        [-0.3298,  0.0152, -0.0148],\n",
            "        [-0.1348,  0.0328, -0.1920],\n",
            "        [-0.1613, -0.1549, -0.1248],\n",
            "        [-0.3920, -0.0029, -0.1352],\n",
            "        [-0.2133, -0.1089,  0.0804]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4072, 0.3972, 0.3573, 0.3733, 0.3782, 0.3408, 0.3916, 0.3886])\n",
            "tensor([1, 1, 1, 1, 1, 2, 1, 2])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 37\n",
            "Total labels: 37\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.1947,  0.1704, -0.0723],\n",
            "        [ 0.0657, -0.1079, -0.1283],\n",
            "        [-0.2736, -0.0392,  0.0991],\n",
            "        [-0.3235,  0.0304, -0.0527],\n",
            "        [-0.2811,  0.0964, -0.0520],\n",
            "        [-0.1212,  0.0304,  0.0262],\n",
            "        [-0.3065,  0.1144,  0.0043],\n",
            "        [-0.0348,  0.0229, -0.1379]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4035, 0.3753, 0.3907, 0.3814, 0.3925, 0.3502, 0.3918, 0.3577])\n",
            "tensor([1, 0, 2, 1, 1, 1, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 38\n",
            "Total labels: 38\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2135, -0.0505, -0.1826],\n",
            "        [-0.1612,  0.0233,  0.0238],\n",
            "        [-0.2751,  0.0738, -0.0787],\n",
            "        [-0.2878,  0.1000,  0.1013],\n",
            "        [-0.4397,  0.1200, -0.0258],\n",
            "        [-0.1278, -0.1031, -0.1635],\n",
            "        [-0.2425,  0.0671,  0.0026],\n",
            "        [-0.2334, -0.0191, -0.0356]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3669, 0.3533, 0.3900, 0.3736, 0.4106, 0.3428, 0.3744, 0.3583])\n",
            "tensor([1, 2, 1, 2, 1, 1, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 39\n",
            "Total labels: 39\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.1706,  0.1503,  0.1224],\n",
            "        [-0.1598, -0.0444, -0.0246],\n",
            "        [-0.1265, -0.0331,  0.0664],\n",
            "        [-0.1268, -0.0977,  0.1102],\n",
            "        [-0.1554,  0.0752, -0.0784],\n",
            "        [-0.1296,  0.2047, -0.2144],\n",
            "        [-0.1985, -0.1105,  0.0855],\n",
            "        [-0.0936, -0.0511,  0.1200]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3706, 0.3504, 0.3663, 0.3844, 0.3771, 0.4213, 0.3884, 0.3773])\n",
            "tensor([1, 2, 2, 2, 1, 1, 2, 2])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 40\n",
            "Total labels: 40\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2812, -0.0217,  0.1289],\n",
            "        [-0.1745,  0.1240,  0.0443],\n",
            "        [-0.2831,  0.0440, -0.0994],\n",
            "        [-0.2697, -0.1344, -0.1411],\n",
            "        [-0.1705,  0.2163, -0.2542],\n",
            "        [-0.2331,  0.0410, -0.1689],\n",
            "        [-0.1843, -0.0672,  0.0085],\n",
            "        [-0.1944, -0.0085, -0.0983]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3962, 0.3752, 0.3865, 0.3488, 0.4341, 0.3890, 0.3634, 0.3644])\n",
            "tensor([2, 1, 1, 1, 1, 1, 2, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 41\n",
            "Total labels: 41\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2138, -0.0751,  0.1573],\n",
            "        [-0.2538,  0.0556,  0.0247],\n",
            "        [-0.0850, -0.0550, -0.0192],\n",
            "        [-0.2689,  0.1597, -0.0249],\n",
            "        [-0.0979,  0.0924, -0.1376],\n",
            "        [-0.2936, -0.0598, -0.1601],\n",
            "        [-0.1641, -0.0350, -0.0645],\n",
            "        [-0.1794,  0.2048, -0.1881]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4028, 0.3699, 0.3447, 0.4028, 0.3815, 0.3709, 0.3509, 0.4244])\n",
            "tensor([2, 1, 2, 1, 1, 1, 1, 1])\n",
            "Texts to add: 3\n",
            "Labels to add: [2, 1, 1]\n",
            "[2, 1, 1]\n",
            "Total texts: 44\n",
            "Total labels: 44\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2307,  0.0728, -0.1297],\n",
            "        [-0.1782, -0.0358, -0.1007],\n",
            "        [-0.0943,  0.0124, -0.1183],\n",
            "        [-0.0942, -0.2054, -0.0893],\n",
            "        [-0.2212, -0.0127,  0.0831],\n",
            "        [-0.3894,  0.2831, -0.0081],\n",
            "        [-0.1973,  0.0571,  0.0488],\n",
            "        [-0.2587, -0.0316,  0.0056]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3914, 0.3566, 0.3602, 0.3466, 0.3779, 0.4429, 0.3614, 0.3661])\n",
            "tensor([1, 1, 1, 2, 2, 1, 1, 2])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 45\n",
            "Total labels: 45\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2838,  0.1059, -0.1677],\n",
            "        [-0.1737,  0.0035,  0.0212],\n",
            "        [ 0.0161, -0.1222,  0.0072],\n",
            "        [-0.0845,  0.0637, -0.1434],\n",
            "        [-0.0789, -0.0422, -0.1801],\n",
            "        [-0.0270, -0.0985, -0.1701],\n",
            "        [-0.5213,  0.2985,  0.0466],\n",
            "        [-0.1273, -0.0262, -0.2363]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4102, 0.3565, 0.3494, 0.3738, 0.3527, 0.3574, 0.4509, 0.3684])\n",
            "tensor([1, 2, 0, 1, 1, 0, 1, 1])\n",
            "Texts to add: 2\n",
            "Labels to add: [1, 1]\n",
            "[1, 1]\n",
            "Total texts: 47\n",
            "Total labels: 47\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2784,  0.0805, -0.0288],\n",
            "        [-0.2652,  0.0259,  0.0663],\n",
            "        [-0.0451, -0.0399,  0.0307],\n",
            "        [-0.0878, -0.0500, -0.2772],\n",
            "        [-0.2473, -0.0218, -0.1362],\n",
            "        [-0.3987,  0.0210, -0.0453],\n",
            "        [-0.5662,  0.2042, -0.1429],\n",
            "        [-0.1641, -0.0379, -0.2104]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3854, 0.3734, 0.3498, 0.3624, 0.3718, 0.3856, 0.4609, 0.3672])\n",
            "tensor([1, 2, 2, 1, 1, 1, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 48\n",
            "Total labels: 48\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.3300,  0.1685,  0.1288],\n",
            "        [-0.2804,  0.0740, -0.1557],\n",
            "        [-0.1221, -0.1281, -0.0687],\n",
            "        [-0.2917, -0.0094, -0.1582],\n",
            "        [-0.1188, -0.0217, -0.0158],\n",
            "        [-0.0674, -0.0360, -0.0078],\n",
            "        [-0.3837,  0.0379,  0.1001],\n",
            "        [-0.1958,  0.1198, -0.0549]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3893, 0.4006, 0.3460, 0.3823, 0.3453, 0.3431, 0.3912, 0.3893])\n",
            "tensor([1, 1, 2, 1, 2, 2, 2, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 49\n",
            "Total labels: 49\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.4871, -0.0449,  0.2119],\n",
            "        [-0.1095,  0.1126,  0.1056],\n",
            "        [-0.1391, -0.0519, -0.1020],\n",
            "        [-0.1778, -0.1961,  0.0040],\n",
            "        [-0.2391, -0.0165,  0.0174],\n",
            "        [-0.1001,  0.0913,  0.0548],\n",
            "        [-0.2179, -0.1677, -0.0519],\n",
            "        [-0.2767, -0.0458, -0.1234]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4404, 0.3579, 0.3487, 0.3770, 0.3649, 0.3584, 0.3653, 0.3678])\n",
            "tensor([2, 1, 1, 2, 2, 1, 2, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [2]\n",
            "[2]\n",
            "Total texts: 50\n",
            "Total labels: 50\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.0926, -0.1196,  0.1422],\n",
            "        [-0.1594, -0.1685, -0.1163],\n",
            "        [-0.1222, -0.0207,  0.2740],\n",
            "        [-0.1349, -0.1808, -0.1064],\n",
            "        [-0.1659,  0.0046,  0.1050],\n",
            "        [-0.0459, -0.0936, -0.1288],\n",
            "        [-0.3575,  0.0257, -0.2265],\n",
            "        [-0.1449, -0.1244, -0.0562]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3906, 0.3440, 0.4136, 0.3448, 0.3749, 0.3480, 0.4067, 0.3510])\n",
            "tensor([2, 2, 2, 2, 2, 0, 1, 2])\n",
            "Texts to add: 2\n",
            "Labels to add: [2, 1]\n",
            "[2, 1]\n",
            "Total texts: 52\n",
            "Total labels: 52\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.3721,  0.1613, -0.2871],\n",
            "        [-0.4110,  0.0842, -0.1097],\n",
            "        [-0.0574, -0.0965, -0.1002],\n",
            "        [-0.2472,  0.3376, -0.0035],\n",
            "        [-0.2192,  0.0722, -0.2945],\n",
            "        [-0.0998,  0.0819,  0.0623],\n",
            "        [-0.1738, -0.0750, -0.0323],\n",
            "        [-0.1780, -0.0856,  0.0191]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4494, 0.4110, 0.3425, 0.4409, 0.4098, 0.3553, 0.3538, 0.3674])\n",
            "tensor([1, 1, 0, 1, 1, 1, 2, 2])\n",
            "Texts to add: 4\n",
            "Labels to add: [1, 1, 1, 1]\n",
            "[1, 1, 1, 1]\n",
            "Total texts: 56\n",
            "Total labels: 56\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[ 0.0067,  0.1205, -0.1710],\n",
            "        [-0.3539,  0.0713, -0.0578],\n",
            "        [-0.2337,  0.0338,  0.0100],\n",
            "        [-0.2407,  0.1181, -0.3241],\n",
            "        [-0.1028,  0.0085, -0.1491],\n",
            "        [ 0.0304,  0.0049,  0.1032],\n",
            "        [-0.2057, -0.0471, -0.2417],\n",
            "        [-0.3013,  0.1152, -0.0140]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3788, 0.3949, 0.3647, 0.4271, 0.3638, 0.3526, 0.3736, 0.3940])\n",
            "tensor([1, 1, 1, 1, 1, 2, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 57\n",
            "Total labels: 57\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2134,  0.1455, -0.1995],\n",
            "        [-0.1506, -0.1036, -0.1480],\n",
            "        [-0.1516, -0.2094, -0.1974],\n",
            "        [-0.1710,  0.1424, -0.1647],\n",
            "        [-0.3321,  0.0244, -0.1719],\n",
            "        [-0.2577, -0.0297, -0.0957],\n",
            "        [-0.2505,  0.0964, -0.2349],\n",
            "        [-0.2625,  0.2257, -0.0082]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4155, 0.3436, 0.3449, 0.4054, 0.3965, 0.3660, 0.4124, 0.4158])\n",
            "tensor([1, 1, 0, 1, 1, 1, 1, 1])\n",
            "Texts to add: 4\n",
            "Labels to add: [1, 1, 1, 1]\n",
            "[1, 1, 1, 1]\n",
            "Total texts: 61\n",
            "Total labels: 61\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2138, -0.2079, -0.0354],\n",
            "        [-0.2038, -0.1688, -0.1230],\n",
            "        [-0.1727,  0.0072, -0.3688],\n",
            "        [-0.0677, -0.0290, -0.1491],\n",
            "        [-0.1432, -0.0525,  0.0031],\n",
            "        [-0.2085, -0.0230,  0.1208],\n",
            "        [-0.2032,  0.0367, -0.2106],\n",
            "        [-0.0171,  0.2249, -0.3271]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3734, 0.3475, 0.3965, 0.3510, 0.3559, 0.3868, 0.3895, 0.4236])\n",
            "tensor([2, 2, 1, 1, 2, 2, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 62\n",
            "Total labels: 62\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2462,  0.1039, -0.2808],\n",
            "        [-0.2202, -0.0180, -0.0417],\n",
            "        [-0.3639, -0.1100, -0.1237],\n",
            "        [-0.3165,  0.1423, -0.0611],\n",
            "        [-0.0032,  0.0096,  0.1571],\n",
            "        [-0.2941,  0.0812, -0.0790],\n",
            "        [-0.2879,  0.0717,  0.0029],\n",
            "        [-0.2114, -0.1667, -0.0874]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4192, 0.3580, 0.3620, 0.4085, 0.3684, 0.3938, 0.3800, 0.3562])\n",
            "tensor([1, 1, 1, 1, 2, 1, 1, 2])\n",
            "Texts to add: 2\n",
            "Labels to add: [1, 1]\n",
            "[1, 1]\n",
            "Total texts: 64\n",
            "Total labels: 64\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.0040, -0.0934, -0.1218],\n",
            "        [-0.1058,  0.0456,  0.0791],\n",
            "        [-0.4579, -0.1456, -0.0898],\n",
            "        [-0.1624, -0.0749, -0.2918],\n",
            "        [-0.0551,  0.0092, -0.2802],\n",
            "        [-0.1984, -0.0387, -0.2858],\n",
            "        [-0.1959,  0.1996,  0.1990],\n",
            "        [-0.0310,  0.0696, -0.0325]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3567, 0.3574, 0.3791, 0.3675, 0.3722, 0.3797, 0.3742, 0.3562])\n",
            "tensor([0, 2, 2, 1, 1, 1, 1, 1])\n",
            "Texts to add: 0\n",
            "Labels to add: []\n",
            "[]\n",
            "Total texts: 64\n",
            "Total labels: 64\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2415,  0.0035, -0.1008],\n",
            "        [ 0.0499, -0.0449, -0.1345],\n",
            "        [-0.0600,  0.0681, -0.0799],\n",
            "        [-0.1960,  0.0816, -0.1249],\n",
            "        [-0.2171,  0.0282,  0.0351],\n",
            "        [-0.2076,  0.0156, -0.2717],\n",
            "        [-0.0686, -0.1298, -0.0804],\n",
            "        [-0.2256,  0.0528,  0.0385]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3726, 0.3648, 0.3647, 0.3890, 0.3610, 0.3921, 0.3414, 0.3646])\n",
            "tensor([1, 0, 1, 1, 2, 1, 0, 1])\n",
            "Texts to add: 0\n",
            "Labels to add: []\n",
            "[]\n",
            "Total texts: 64\n",
            "Total labels: 64\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.0887, -0.1211,  0.0303],\n",
            "        [-0.1212, -0.0152, -0.1595],\n",
            "        [-0.1661,  0.0244,  0.0150],\n",
            "        [-0.0209, -0.1649,  0.0615],\n",
            "        [-0.2260,  0.0530, -0.0782],\n",
            "        [-0.1733, -0.1446, -0.1048],\n",
            "        [-0.0309, -0.0966, -0.1764],\n",
            "        [-0.1601, -0.0317, -0.1201]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3640, 0.3617, 0.3550, 0.3679, 0.3797, 0.3454, 0.3570, 0.3578])\n",
            "tensor([2, 1, 1, 2, 1, 2, 0, 1])\n",
            "Texts to add: 0\n",
            "Labels to add: []\n",
            "[]\n",
            "Total texts: 64\n",
            "Total labels: 64\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.2025,  0.0586, -0.3276],\n",
            "        [-0.2708,  0.0595,  0.1342],\n",
            "        [-0.1483, -0.0390,  0.0348],\n",
            "        [-0.2492,  0.0687, -0.2485],\n",
            "        [-0.1587, -0.1174, -0.0139],\n",
            "        [-0.3446,  0.0338, -0.2101],\n",
            "        [-0.3868, -0.0134, -0.0688],\n",
            "        [-0.4597,  0.1738, -0.0500]]), hidden_states=None, attentions=None)\n",
            "tensor([0.4082, 0.3854, 0.3621, 0.4072, 0.3614, 0.4051, 0.3796, 0.4291])\n",
            "tensor([1, 2, 2, 1, 2, 1, 1, 1])\n",
            "Texts to add: 4\n",
            "Labels to add: [1, 1, 1, 1]\n",
            "[1, 1, 1, 1]\n",
            "Total texts: 68\n",
            "Total labels: 68\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.1065, -0.0067,  0.0934],\n",
            "        [-0.0812,  0.0777, -0.0289],\n",
            "        [-0.2520,  0.1131,  0.0257],\n",
            "        [-0.1417,  0.4148,  0.0427],\n",
            "        [-0.2083,  0.0824, -0.0127],\n",
            "        [-0.1882, -0.0890, -0.0380],\n",
            "        [-0.1083,  0.0086, -0.1085],\n",
            "        [-0.4372,  0.1307, -0.0201]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3672, 0.3634, 0.3831, 0.4420, 0.3764, 0.3558, 0.3598, 0.4121])\n",
            "tensor([2, 1, 1, 1, 1, 2, 1, 1])\n",
            "Texts to add: 2\n",
            "Labels to add: [1, 1]\n",
            "[1, 1]\n",
            "Total texts: 70\n",
            "Total labels: 70\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.0312,  0.0027, -0.2080],\n",
            "        [-0.3551, -0.0616, -0.1269],\n",
            "        [-0.1503, -0.0603, -0.0748],\n",
            "        [-0.3031, -0.0228, -0.1689],\n",
            "        [-0.0694, -0.0037, -0.2390],\n",
            "        [-0.3481, -0.0241,  0.1497]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3601, 0.3728, 0.3449, 0.3817, 0.3667, 0.4084])\n",
            "tensor([1, 1, 1, 1, 1, 2])\n",
            "Texts to add: 1\n",
            "Labels to add: [2]\n",
            "[2]\n",
            "Total texts: 71\n",
            "Total labels: 71\n",
            "SequenceClassifierOutput(loss=None, logits=tensor([[-0.3828, -0.0661,  0.0100],\n",
            "        [-0.1579, -0.1556, -0.2602],\n",
            "        [-0.3735,  0.1286, -0.0940],\n",
            "        [-0.1116,  0.0260, -0.0553],\n",
            "        [-0.3497, -0.0237, -0.0147],\n",
            "        [-0.4137, -0.0567,  0.0146],\n",
            "        [-0.1102, -0.0164, -0.0695],\n",
            "        [-0.0474, -0.0334, -0.1170]]), hidden_states=None, attentions=None)\n",
            "tensor([0.3843, 0.3450, 0.4157, 0.3580, 0.3695, 0.3872, 0.3498, 0.3441])\n",
            "tensor([2, 1, 1, 1, 2, 2, 1, 1])\n",
            "Texts to add: 1\n",
            "Labels to add: [1]\n",
            "[1]\n",
            "Total texts: 72\n",
            "Total labels: 9\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "AssertionError",
          "evalue": "Lengths of texts and labels do not match",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mAssertionError\u001b[0m                            Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-31-e13c7e712839>\u001b[0m in \u001b[0;36m<cell line: 7>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     44\u001b[0m         \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mf\"Total labels: {len(labels)}\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     45\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 46\u001b[0;31m         \u001b[0;32massert\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtexts\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"Lengths of texts and labels do not match\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     47\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     48\u001b[0m         \u001b[0;31m# Update dataset and dataloader with new labeled data\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mAssertionError\u001b[0m: Lengths of texts and labels do not match"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(labels.tolist())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "T6R8IcDosjLb",
        "outputId": "58fd5a8f-b823-4d21-b0be-13d59f6aa04e"
      },
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[1, 1, 0, 2, 0, 0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "model.eval()\n",
        "correct_preds = 0\n",
        "total_preds = 0\n",
        "\n",
        "with torch.no_grad():\n",
        "  for batch in test_dataloader:\n",
        "    input_ids = batch['input_ids'].to(device)\n",
        "    attention_mask = batch['attention_mask'].to(device)\n",
        "    labels = batch['label'].to(device)\n",
        "\n",
        "    outputs = model(input_ids=input_ids, attention_mask=attention_mask)\n",
        "    _, predicted = torch.max(outputs.logits, dim=1)\n",
        "\n",
        "    total_preds += labels.size(0)\n",
        "    correct_preds += (predicted == labels).sum().item()\n",
        "\n",
        "accuracy = correct_preds / total_preds\n",
        "print(f\"Initial model accuracy :{accuracy}\")\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IuiVyCNFEOJF",
        "outputId": "9a4bc49d-f98b-4085-dbb0-93afb4e325a1"
      },
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Initial model accuracy :0.5\n"
          ]
        }
      ]
    }
  ]
}