Upload 3 files

app.py

@@ -0,0 +1,109 @@
+import streamlit as st
+import os
+import torch
+from transformers import BertTokenizer
+from transformers import BertForSequenceClassification
+
+
+@st.cache_resource  # кэширование
+def get_model(unzip_root: str='./'):
+    """
+        unzip_root ~ в тестовой среде будет произведена операция `unzip archive.zip` с переданным архивом и в эту функцию будет передан путь до `realpath .`
+    """
+    checkpoint_path = os.path.join(unzip_root, "model.pth")
+    checkpoint = torch.load(checkpoint_path, map_location="cpu")
+
+    model_path = 'cointegrated/rubert-tiny'
+    model = BertForSequenceClassification.from_pretrained(model_path)
+    out_features = model.bert.encoder.layer[1].output.dense.out_features
+    model.classifier = torch.nn.Linear(out_features, len(dict_tetm_to_int))
+
+    model.load_state_dict(checkpoint)
+    return model
+
+
+@st.cache_resource  # кэширование
+def get_tokenizer():
+    tokenizer_path = 'cointegrated/rubert-tiny'
+    tokenizer = BertTokenizer.from_pretrained(tokenizer_path)
+    return tokenizer
+
+
+@st.cache_resource  # кэширование
+def get_vocab(unzip_root: str='./'):
+    """
+        unzip_root ~ в тестовой среде будет произведена операция `unzip archive.zip` с переданным архивом и в эту функцию будет передан путь до `realpath .`
+    """
+    path = os.path.join(unzip_root, "vocab.tsv")
+
+    with open(path, 'r') as f:
+      size_dict = int(f.readline())
+      dict_tetm_to_int = dict()
+      for _ in range(size_dict):
+        key = f.readline()[:-1]
+        dict_tetm_to_int[key] = int(f.readline())
+
+      size_dict = int(f.readline())
+      dict_int_to_term = dict()
+      for _ in range(size_dict):
+        key = int(f.readline())
+        dict_int_to_term[key] = f.readline()[:-1]
+
+    return dict_tetm_to_int, dict_int_to_term
+
+
+softmax = torch.nn.Softmax(dim=1)
+dict_tetm_to_int, dict_int_to_term = get_vocab()
+model = get_model()
+tokenizer = get_tokenizer()
+
+
+def predict(text, device='cpu'):
+    encoding = tokenizer.encode_plus(
+        text,
+        add_special_tokens=True,
+        max_length=512,
+        return_token_type_ids=False,
+        truncation=True,
+        padding='max_length',
+        return_attention_mask=True,
+        return_tensors='pt',
+    )
+    
+    out = {
+          'text': text,
+          'input_ids': encoding['input_ids'].flatten(),
+          'attention_mask': encoding['attention_mask'].flatten()
+      }
+    
+    input_ids = out["input_ids"].to(device)
+    attention_mask = out["attention_mask"].to(device)
+    
+    outputs = model(
+        input_ids=input_ids.unsqueeze(0),
+        attention_mask=attention_mask.unsqueeze(0)
+    )
+    
+    out = softmax(outputs.logits)
+    prediction = torch.argsort(outputs.logits, dim=1, descending=True).cpu()[0]
+    sum_answer = 0
+    answer = []
+    idx = 0
+    while sum_answer < 0.95:
+      sum_answer += out[0][idx].item()
+      answer.append(dict_int_to_term[prediction[idx].item()])
+      idx += 1
+
+    return answer
+
+
+st.title("We will help you determine what topic your article belongs to:)")
+st.header("Please enter a title and/or introduction")
+
+
+title = st.text_input(label="Title", value="")
+abstract = st.text_input(label="Abstract", value="")
+if(st.button('Show result')):
+    predict = ' '.join(predict(title.title() + ' ' + abstract.title()))
+    result = 'Suggested answer:\n' + predict
+    st.success(result)

model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b719d3530078cfa62d0f0bb8f203a117cfda27c808e236800614a0d53b619df
+size 47331569

vocab.tsv

@@ -0,0 +1,562 @@
+140
+cs.AI
+0
+cs.CL
+1
+cs.CV
+2
+cs.NE
+3
+stat.ML
+4
+cs.LG
+5
+physics.soc-ph
+6
+stat.AP
+7
+cs.RO
+8
+cs.SE
+9
+cs.MA
+10
+math.OC
+11
+cs.IR
+12
+cond-mat.dis-nn
+13
+stat.ME
+14
+physics.chem-ph
+15
+cs.DC
+16
+stat.CO
+17
+q-bio.NC
+18
+cs.GT
+19
+cs.MM
+20
+cs.CG
+21
+cs.CR
+22
+cs.HC
+23
+cs.GL
+24
+eess.AS
+25
+cs.SD
+26
+math.DS
+27
+cs.GR
+28
+math.NA
+29
+cs.CY
+30
+physics.data-an
+31
+math.ST
+32
+stat.TH
+33
+cs.IT
+34
+math.IT
+35
+astro-ph.IM
+36
+astro-ph.GA
+37
+cs.SI
+38
+cs.DB
+39
+cs.LO
+40
+nlin.AO
+41
+cs.PF
+42
+cs.ET
+43
+eess.IV
+44
+cs.AR
+45
+cs.SY
+46
+cs.CC
+47
+q-bio.BM
+48
+q-bio.QM
+49
+cs.NI
+50
+cs.DS
+51
+cond-mat.stat-mech
+52
+cs.NA
+53
+cs.DM
+54
+eess.SP
+55
+cs.MS
+56
+physics.med-ph
+57
+physics.optics
+58
+q-fin.CP
+59
+cs.FL
+60
+cs.SC
+61
+q-fin.EC
+62
+q-fin.TR
+63
+cond-mat.mes-hall
+64
+math.PR
+65
+q-fin.RM
+66
+nlin.CD
+67
+cs.CE
+68
+math.AT
+69
+stat.OT
+70
+physics.ao-ph
+71
+math.SP
+72
+cs.PL
+73
+math.AP
+74
+math.FA
+75
+physics.geo-ph
+76
+q-bio.TO
+77
+physics.comp-ph
+78
+cs.DL
+79
+math.CO
+80
+physics.flu-dyn
+81
+math.MG
+82
+astro-ph.EP
+83
+q-bio.CB
+84
+math.RA
+85
+astro-ph.CO
+86
+cond-mat.mtrl-sci
+87
+q-fin.ST
+88
+q-bio.GN
+89
+nlin.CG
+90
+nlin.PS
+91
+math.HO
+92
+q-fin.GN
+93
+math.LO
+94
+math.CT
+95
+cs.CL, cs.AI, math.CT
+96
+q-bio.PE
+97
+astro-ph.SR
+98
+q-fin.PM
+99
+physics.bio-ph
+100
+math.AG
+101
+cs.OH
+102
+math.DG
+103
+astro-ph.HE
+104
+econ.EM
+105
+math.QA
+106
+q-bio.SC
+107
+math.GM
+108
+q-bio.MN
+109
+math.GT
+110
+math.AC
+111
+math.CA
+112
+cond-mat.str-el
+113
+math.GN
+114
+aaai.org
+115
+cond-mat.supr-con
+116
+q-bio.OT
+117
+physics.ins-det
+118
+physics.app-ph
+119
+math.RT
+120
+math.MP
+121
+A.m
+122
+H.m
+123
+physics.class-ph
+124
+q-fin.PR
+125
+physics.space-ph
+126
+physics.gen-ph
+127
+cond-mat.other
+128
+math.GR
+129
+cond-mat.quant-gas
+130
+math.OA
+131
+physics.hist-ph
+132
+math.NT
+133
+cs.OS
+134
+cond-mat.soft
+135
+q-bio.BM, q-bio.MN, q-bio.NC, nlin.AO, nlin.CD
+136
+nlin.AO, nlin.CD, q-bio.NC, physics.bio-ph, cond-mat.dis-nn
+137
+physics.pop-ph
+138
+math.CV
+139
+140
+0
+cs.AI
+1
+cs.CL
+2
+cs.CV
+3
+cs.NE
+4
+stat.ML
+5
+cs.LG
+6
+physics.soc-ph
+7
+stat.AP
+8
+cs.RO
+9
+cs.SE
+10
+cs.MA
+11
+math.OC
+12
+cs.IR
+13
+cond-mat.dis-nn
+14
+stat.ME
+15
+physics.chem-ph
+16
+cs.DC
+17
+stat.CO
+18
+q-bio.NC
+19
+cs.GT
+20
+cs.MM
+21
+cs.CG
+22
+cs.CR
+23
+cs.HC
+24
+cs.GL
+25
+eess.AS
+26
+cs.SD
+27
+math.DS
+28
+cs.GR
+29
+math.NA
+30
+cs.CY
+31
+physics.data-an
+32
+math.ST
+33
+stat.TH
+34
+cs.IT
+35
+math.IT
+36
+astro-ph.IM
+37
+astro-ph.GA
+38
+cs.SI
+39
+cs.DB
+40
+cs.LO
+41
+nlin.AO
+42
+cs.PF
+43
+cs.ET
+44
+eess.IV
+45
+cs.AR
+46
+cs.SY
+47
+cs.CC
+48
+q-bio.BM
+49
+q-bio.QM
+50
+cs.NI
+51
+cs.DS
+52
+cond-mat.stat-mech
+53
+cs.NA
+54
+cs.DM
+55
+eess.SP
+56
+cs.MS
+57
+physics.med-ph
+58
+physics.optics
+59
+q-fin.CP
+60
+cs.FL
+61
+cs.SC
+62
+q-fin.EC
+63
+q-fin.TR
+64
+cond-mat.mes-hall
+65
+math.PR
+66
+q-fin.RM
+67
+nlin.CD
+68
+cs.CE
+69
+math.AT
+70
+stat.OT
+71
+physics.ao-ph
+72
+math.SP
+73
+cs.PL
+74
+math.AP
+75
+math.FA
+76
+physics.geo-ph
+77
+q-bio.TO
+78
+physics.comp-ph
+79
+cs.DL
+80
+math.CO
+81
+physics.flu-dyn
+82
+math.MG
+83
+astro-ph.EP
+84
+q-bio.CB
+85
+math.RA
+86
+astro-ph.CO
+87
+cond-mat.mtrl-sci
+88
+q-fin.ST
+89
+q-bio.GN
+90
+nlin.CG
+91
+nlin.PS
+92
+math.HO
+93
+q-fin.GN
+94
+math.LO
+95
+math.CT
+96
+cs.CL, cs.AI, math.CT
+97
+q-bio.PE
+98
+astro-ph.SR
+99
+q-fin.PM
+100
+physics.bio-ph
+101
+math.AG
+102
+cs.OH
+103
+math.DG
+104
+astro-ph.HE
+105
+econ.EM
+106
+math.QA
+107
+q-bio.SC
+108
+math.GM
+109
+q-bio.MN
+110
+math.GT
+111
+math.AC
+112
+math.CA
+113
+cond-mat.str-el
+114
+math.GN
+115
+aaai.org
+116
+cond-mat.supr-con
+117
+q-bio.OT
+118
+physics.ins-det
+119
+physics.app-ph
+120
+math.RT
+121
+math.MP
+122
+A.m
+123
+H.m
+124
+physics.class-ph
+125
+q-fin.PR
+126
+physics.space-ph
+127
+physics.gen-ph
+128
+cond-mat.other
+129
+math.GR
+130
+cond-mat.quant-gas
+131
+math.OA
+132
+physics.hist-ph
+133
+math.NT
+134
+cs.OS
+135
+cond-mat.soft
+136
+q-bio.BM, q-bio.MN, q-bio.NC, nlin.AO, nlin.CD
+137
+nlin.AO, nlin.CD, q-bio.NC, physics.bio-ph, cond-mat.dis-nn
+138
+physics.pop-ph
+139
+math.CV