Adding tweeteval classifier

.ipynb_checkpoints/README-checkpoint.md

@@ -0,0 +1,87 @@
+# Twitter-roBERTa-base
+
+This is a roBERTa-base model trained on ~58M tweets and finetuned for the hate detection task at Semeval 2019. 
+For full description: [_TweetEval_ benchmark (Findings of EMNLP 2020)](https://arxiv.org/pdf/2010.12421.pdf). 
+To evaluate this and other models on Twitter-specific data, please refer to the [Tweeteval official repository](https://github.com/cardiffnlp/tweeteval).
+
+## Example of classification
+
+```python
+from transformers import AutoModelForSequenceClassification
+from transformers import TFAutoModelForSequenceClassification
+from transformers import AutoTokenizer
+import numpy as np
+from scipy.special import softmax
+import csv
+import urllib.request
+
+# Tasks:
+# emoji, emotion, hate, irony, offensive, sentiment
+# stance/abortion, stance/atheism, stance/climate, stance/feminist, stance/hillary
+
+task='hate'
+MODEL = f"cardiffnlp/twitter-roberta-base-{task}"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL)
+
+# download label mapping
+labels=[]
+mapping_link = f"https://raw.githubusercontent.com/cardiffnlp/tweeteval/main/datasets/{task}/mapping.txt"
+with urllib.request.urlopen(mapping_link) as f:
+    html = f.read().decode('utf-8').split("\n")
+    spamreader = csv.reader(html[:-1], delimiter='\t')
+labels = [row[1] for row in spamreader]
+
+# PT
+model = AutoModelForSequenceClassification.from_pretrained(MODEL)
+model.save_pretrained(MODEL)
+
+text = "Good night 😊"
+encoded_input = tokenizer(text, return_tensors='pt')
+output = model(**encoded_input)
+scores = output[0][0].detach().numpy()
+scores = softmax(scores)
+
+# # TF
+# model = TFAutoModelForSequenceClassification.from_pretrained(MODEL)
+# model.save_pretrained(MODEL)
+
+# text = "Good night 😊"
+# encoded_input = tokenizer(text, return_tensors='tf')
+# output = model(encoded_input)
+# scores = output[0][0].numpy()
+# scores = softmax(scores)
+
+ranking = np.argsort(scores)
+ranking = ranking[::-1]
+for i in range(scores.shape[0]):
+    l = labels[ranking[i]]
+    s = scores[ranking[i]]
+    print(f"{i+1}) {l} {np.round(float(s), 4)}")
+
+```
+
+Output: 
+
+```
+1) 😘 0.2637
+2) ❤️ 0.1952
+3) 💕 0.1171
+4) ✨ 0.0927
+5) 😊 0.0756
+6) 💜 0.046
+7) 💙 0.0444
+8) 😍 0.0272
+9) 😉 0.0228
+10) 😎 0.0198
+11) 😜 0.0166
+12) 😂 0.0132
+13) 😁 0.0131
+14) ☀ 0.0112
+15) 🎄 0.009
+16) 💯 0.009
+17) 🔥 0.008
+18) 📷 0.0057
+19) 🇺🇸 0.005
+20) 📸 0.0048
+```

README.md

@@ -0,0 +1,87 @@
+# Twitter-roBERTa-base
+
+This is a roBERTa-base model trained on ~58M tweets and finetuned for the hate detection task at Semeval 2019. 
+For full description: [_TweetEval_ benchmark (Findings of EMNLP 2020)](https://arxiv.org/pdf/2010.12421.pdf). 
+To evaluate this and other models on Twitter-specific data, please refer to the [Tweeteval official repository](https://github.com/cardiffnlp/tweeteval).
+
+## Example of classification
+
+```python
+from transformers import AutoModelForSequenceClassification
+from transformers import TFAutoModelForSequenceClassification
+from transformers import AutoTokenizer
+import numpy as np
+from scipy.special import softmax
+import csv
+import urllib.request
+
+# Tasks:
+# emoji, emotion, hate, irony, offensive, sentiment
+# stance/abortion, stance/atheism, stance/climate, stance/feminist, stance/hillary
+
+task='hate'
+MODEL = f"cardiffnlp/twitter-roberta-base-{task}"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL)
+
+# download label mapping
+labels=[]
+mapping_link = f"https://raw.githubusercontent.com/cardiffnlp/tweeteval/main/datasets/{task}/mapping.txt"
+with urllib.request.urlopen(mapping_link) as f:
+    html = f.read().decode('utf-8').split("\n")
+    spamreader = csv.reader(html[:-1], delimiter='\t')
+labels = [row[1] for row in spamreader]
+
+# PT
+model = AutoModelForSequenceClassification.from_pretrained(MODEL)
+model.save_pretrained(MODEL)
+
+text = "Good night 😊"
+encoded_input = tokenizer(text, return_tensors='pt')
+output = model(**encoded_input)
+scores = output[0][0].detach().numpy()
+scores = softmax(scores)
+
+# # TF
+# model = TFAutoModelForSequenceClassification.from_pretrained(MODEL)
+# model.save_pretrained(MODEL)
+
+# text = "Good night 😊"
+# encoded_input = tokenizer(text, return_tensors='tf')
+# output = model(encoded_input)
+# scores = output[0][0].numpy()
+# scores = softmax(scores)
+
+ranking = np.argsort(scores)
+ranking = ranking[::-1]
+for i in range(scores.shape[0]):
+    l = labels[ranking[i]]
+    s = scores[ranking[i]]
+    print(f"{i+1}) {l} {np.round(float(s), 4)}")
+
+```
+
+Output: 
+
+```
+1) 😘 0.2637
+2) ❤️ 0.1952
+3) 💕 0.1171
+4) ✨ 0.0927
+5) 😊 0.0756
+6) 💜 0.046
+7) 💙 0.0444
+8) 😍 0.0272
+9) 😉 0.0228
+10) 😎 0.0198
+11) 😜 0.0166
+12) 😂 0.0132
+13) 😁 0.0131
+14) ☀ 0.0112
+15) 🎄 0.009
+16) 💯 0.009
+17) 🔥 0.008
+18) 📷 0.0057
+19) 🇺🇸 0.005
+20) 📸 0.0048
+```

config.json

@@ -0,0 +1,23 @@
+{
+  "_name_or_path": "tweeteval_new/roberta-base-rt-hate/",
+  "architectures": [
+    "RobertaForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "type_vocab_size": 1,
+  "vocab_size": 50265
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b1fb0237f8355b4f0d9cfc99d4be4ca13987b7641a0a05150ea353c6e60acf8
+size 498676425

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": "<mask>"}

tf_model.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4431998797c5c1de55464e508eb6d3e86c6589209ef97b14ae094a8c3b7e3a64
+size 501227232