JimmyTarbender/GPT2HistoryEvents

1import streamlit as st2import numpy as np3import pandas as pd4import os5import torch6import torch.nn as nn7from transformers.activations import get_activation8from transformers import AutoTokenizer, AutoModelForCausalLM9 10 11st.title('GPT2: To see all prompt outlines: https://huggingface.co/BigSalmon/BigSalmon/InformalToFormalLincoln91Paraphrase')12 13device = torch.device("cuda" if torch.cuda.is_available() else "cpu")14 15@st.cache(allow_output_mutation=True)16def get_model():17    tokenizer = AutoTokenizer.from_pretrained("BigSalmon/HistoryCurrentEvents")18    model = AutoModelForCausalLM.from_pretrained("BigSalmon/HistoryCurrentEvents")19    20    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincolnMediumParaphraseConcise")21    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincolnMediumParaphraseConcise")22 23    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln91Paraphrase")24    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln91Paraphrase")25 26    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln90Paraphrase")27    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln90Paraphrase")28 29    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln88Paraphrase")30    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln88Paraphrase")31 32    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln86Paraphrase")33    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln86Paraphrase")34    35    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln82Paraphrase")36    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln82Paraphrase")37 38    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln79Paraphrase")39    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln79Paraphrase")40    41    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln74Paraphrase")42    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln74Paraphrase")43 44    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln72Paraphrase")45    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln72Paraphrase")46 47    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln64Paraphrase")48    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln64Paraphrase")49 50    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln60Paraphrase")51    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln60Paraphrase")52 53    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/GPTNeo1.3BInformalToFormal")54    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/GPTNeo1.3BInformalToFormal")55 56    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln55")57    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln55")58 59    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln51")60    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln51")61 62    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln45")63    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln49")64    65    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln43")66    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln43")67 68    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln41")69    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln41")70 71    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln38")72    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln38")73 74    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln37")75    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln37")76 77    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln36")78    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln36")79 80    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/MediumInformalToFormalLincoln")81    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/MediumInformalToFormalLincoln")82 83    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln35")84    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln35")85 86    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln31")87    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln31")88    89    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln21")90    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln21")91    92    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/PointsOneSent")93    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/PointsOneSent")94    95    #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/PointsToSentence")96    #model = AutoModelForCausalLM.from_pretrained("BigSalmon/PointsToSentence")97    98    return model, tokenizer99    100model, tokenizer = get_model()101 102g = """informal english: garage band has made people who know nothing about music good at creating music.103Translated into the Style of Abraham Lincoln: garage band ( offers the uninitiated in music the ability to produce professional-quality compositions / catapults those for whom music is an uncharted art the ability the realize masterpieces / stimulates music novice's competency to yield sublime arrangements / begets individuals of rudimentary musical talent the proficiency to fashion elaborate suites ).104informal english: chrome extensions can make doing regular tasks much easier to get done.105Translated into the Style of Abraham Lincoln: chrome extensions ( yield the boon of time-saving convenience / ( expedite the ability to / unlock the means to more readily ) accomplish everyday tasks / turbocharges the velocity with which one can conduct their obligations ).106informal english: broadband is finally expanding to rural areas, a great development that will thrust them into modern life.107Translated into the Style of Abraham Lincoln: broadband is ( ( finally / at last / after years of delay ) arriving in remote locations / springing to life in far-flung outposts / inching into even the most backwater corners of the nation ) that will leap-frog them into the twenty-first century.108informal english: google translate has made talking to people who do not share your language easier.109Translated into the Style of Abraham Lincoln: google translate ( imparts communicability to individuals whose native tongue differs / mitigates the trials of communication across linguistic barriers / hastens the bridging of semantic boundaries / mollifies the complexity of multilingual communication / avails itself to the internationalization of discussion / flexes its muscles to abet intercultural conversation / calms the tides of linguistic divergence ).110informal english: corn fields are all across illinois, visible once you leave chicago.111Translated into the Style of Abraham Lincoln: corn fields ( permeate illinois / span the state of illinois / ( occupy / persist in ) all corners of illinois / line the horizon of illinois / envelop the landscape of illinois ), manifesting themselves visibly as one ventures beyond chicago.112informal english: """113 114number_of_outputs = st.sidebar.slider("Number of Outputs", 5, 100)115log_nums = st.sidebar.slider("How Many Log Outputs?", 50, 600)116 117def BestProbs(prompt):118  prompt = prompt.strip()119  text = tokenizer.encode(prompt)120  myinput, past_key_values = torch.tensor([text]), None121  myinput = myinput122  logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)123  logits = logits[0,-1]124  probabilities = torch.nn.functional.softmax(logits)125  best_logits, best_indices = logits.topk(10)126  best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]127  for i in best_words[0:10]:128    print("_______")129    st.write(f"${i} $\n")130    f = (f"${i} $\n")131    m = (prompt + f"{i}")132    BestProbs2(m)133  return f134 135def BestProbs2(prompt):136  prompt = prompt.strip()137  text = tokenizer.encode(prompt)138  myinput, past_key_values = torch.tensor([text]), None139  myinput = myinput140  logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)141  logits = logits[0,-1]142  probabilities = torch.nn.functional.softmax(logits)143  best_logits, best_indices = logits.topk(20)144  best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]145  for i in best_words[0:20]:146    print(i)147    st.write(i)148    149def LogProbs(prompt):150  col1 = []151  col2 = []152  prompt = prompt.strip()153  text = tokenizer.encode(prompt)154  myinput, past_key_values = torch.tensor([text]), None155  myinput = myinput156  logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)157  logits = logits[0,-1]158  probabilities = torch.nn.functional.softmax(logits)159  best_logits, best_indices = logits.topk(10)160  best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]161  for i in best_words[0:10]:162    print("_______")163    f = i164    col1.append(f)165    m = (prompt + f"{i}")166    #print("^^" + f + " ^^")167    prompt = m.strip()168    text = tokenizer.encode(prompt)169    myinput, past_key_values = torch.tensor([text]), None170    myinput = myinput171    logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)172    logits = logits[0,-1]173    probabilities = torch.nn.functional.softmax(logits)174    best_logits, best_indices = logits.topk(20)175    best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]176    for i in best_words[0:20]:177      #print(i)178      col2.append(i)179  #print(col1)180  #print(col2)181  d = {col1[0]: [col2[0], col2[1], col2[2], col2[3], col2[4], col2[5], col2[6], col2[7], col2[8], col2[9], col2[10], col2[11], col2[12], col2[13], col2[14], col2[15], col2[16], col2[17], col2[18], col2[19]],182    col1[1]: [col2[20], col2[21], col2[22], col2[23], col2[24], col2[25], col2[26], col2[27], col2[28], col2[29], col2[30], col2[31], col2[32], col2[33], col2[34], col2[35], col2[36], col2[37], col2[38], col2[39]],183    col1[2]: [col2[40], col2[41], col2[42], col2[43], col2[44], col2[45], col2[46], col2[47], col2[48], col2[49], col2[50], col2[51], col2[52], col2[53], col2[54], col2[55], col2[56], col2[57], col2[58], col2[59]],184    col1[3]: [col2[60], col2[61], col2[62], col2[63], col2[64], col2[65], col2[66], col2[67], col2[68], col2[69], col2[70], col2[71], col2[72], col2[73], col2[74], col2[75], col2[76], col2[77], col2[78], col2[79]],185    col1[4]: [col2[80], col2[81], col2[82], col2[83], col2[84], col2[85], col2[86], col2[87], col2[88], col2[89], col2[90], col2[91], col2[92], col2[93], col2[94], col2[95], col2[96], col2[97], col2[98], col2[99]],186    col1[5]: [col2[100], col2[101], col2[102], col2[103], col2[104], col2[105], col2[106], col2[107], col2[108], col2[109], col2[110], col2[111], col2[112], col2[113], col2[114], col2[115], col2[116], col2[117], col2[118], col2[119]],187    col1[6]: [col2[120], col2[121], col2[122], col2[123], col2[124], col2[125], col2[126], col2[127], col2[128], col2[129], col2[130], col2[131], col2[132], col2[133], col2[134], col2[135], col2[136], col2[137], col2[138], col2[139]],188    col1[7]: [col2[140], col2[141], col2[142], col2[143], col2[144], col2[145], col2[146], col2[147], col2[148], col2[149], col2[150], col2[151], col2[152], col2[153], col2[154], col2[155], col2[156], col2[157], col2[158], col2[159]],189    col1[8]: [col2[160], col2[161], col2[162], col2[163], col2[164], col2[165], col2[166], col2[167], col2[168], col2[169], col2[170], col2[171], col2[172], col2[173], col2[174], col2[175], col2[176], col2[177], col2[178], col2[179]],190    col1[9]: [col2[180], col2[181], col2[182], col2[183], col2[184], col2[185], col2[186], col2[187], col2[188], col2[189], col2[190], col2[191], col2[192], col2[193], col2[194], col2[195], col2[196], col2[197], col2[198], col2[199]]}191  df = pd.DataFrame(data=d)192  print(df)193  st.write(df)194  return df195  196def BestProbs5(prompt):197  prompt = prompt.strip()198  text = tokenizer.encode(prompt)199  myinput, past_key_values = torch.tensor([text]), None200  myinput = myinput201  logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)202  logits = logits[0,-1]203  probabilities = torch.nn.functional.softmax(logits)204  best_logits, best_indices = logits.topk(number_of_outputs)205  best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]206  for i in best_words[0:number_of_outputs]:207    #print(i)208    print("\n")209    g = (prompt + i)210    st.write(g)211    l = run_generate(g, "hey")212    st.write(l)213    214def run_generate(text, bad_words):215  yo = []216  input_ids = tokenizer.encode(text, return_tensors='pt')217  res = len(tokenizer.encode(text))218  bad_words = bad_words.split()219  bad_word_ids = [[7829], [40940]]220  for bad_word in bad_words: 221    bad_word = " " + bad_word222    ids = tokenizer(bad_word).input_ids223    bad_word_ids.append(ids)224  sample_outputs = model.generate(225    input_ids,226    do_sample=True, 227    max_length= res + 5, 228    min_length = res + 5, 229    top_k=50,230    temperature=1.0,231    num_return_sequences=3,232    bad_words_ids=bad_word_ids233  )234  for i in range(3):235    e = tokenizer.decode(sample_outputs[i])236    e = e.replace(text, "")237    yo.append(e)238  print(yo)239  return yo240 241with st.form(key='my_form'):242    prompt = st.text_area(label='Enter sentence', value=g, height=500)243    submit_button = st.form_submit_button(label='Submit')244    submit_button2 = st.form_submit_button(label='Fast Forward')245    submit_button3 = st.form_submit_button(label='Fast Forward 2.0')246    submit_button4 = st.form_submit_button(label='Get Top')247 248    if submit_button:249      with torch.no_grad():250        text = tokenizer.encode(prompt)251        myinput, past_key_values = torch.tensor([text]), None252        myinput = myinput253        myinput= myinput.to(device)254        logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)255        logits = logits[0,-1]256        probabilities = torch.nn.functional.softmax(logits)257        best_logits, best_indices = logits.topk(log_nums)258        best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]259        text.append(best_indices[0].item())260        best_probabilities = probabilities[best_indices].tolist()261        words = []              262        st.write(best_words)263    if submit_button2:264        print("----")265        st.write("___")266        m = LogProbs(prompt)267        st.write("___")268        st.write(m)269        st.write("___")270    if submit_button3:271        print("----")272        st.write("___")273        st.write(BestProbs)274    if submit_button4:275      BestProbs5(prompt)