asahi417 commited on
Commit
cd4d367
·
1 Parent(s): 9e6d8d6
data/tweet_hate_balance_test0_seed0/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ee74c8688b442266fb541e7a20c526fb5d716a2a7d6a9792e74e8ee9049e1294
3
- size 237307
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2da6d943a79e0ca6d66c1e8326da0b0daf42b53ae47569b01b9af54672e2de8f
3
+ size 291732
data/tweet_hate_balance_test0_seed0/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d89fe0aa8aa8636c0f3f3b07b457346a808c21f83bd29c4fdaa49d0b173e779e
3
- size 66066
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b98e89bd496a345570938fae97ae061f8daaa894c4e265b5341be22496752bf
3
+ size 76084
data/tweet_hate_balance_test0_seed1/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ec347c9483d896d9202373d0ec8ef8c542e9fa3e68bd2335e8b11061bb7fc45b
3
- size 234457
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e12a8289d71cfde1bd33f360ed19ee0d4170a4188182552958ec1d8370553e95
3
+ size 288664
data/tweet_hate_balance_test0_seed1/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9a645569eaed7df4621b12c6d1787dd416366180dcf892d981061af300b4c625
3
- size 67000
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e91ba5934a98e3dcd4b7ca1f1bf517874bf1ffef03a007e150a6bcdf61e98c7
3
+ size 76325
data/tweet_hate_balance_test0_seed2/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9f6f18356758223aba52f3b16564afadb6cce949d24e6a94a67f113f8f1722b7
3
- size 235688
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3626c1b044c36377420dfa85dfa90039405f7f8cba8d74d7cab58dfb07b7016b
3
+ size 290136
data/tweet_hate_balance_test0_seed2/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c30d5763a474d97fb88e5118cd006df2713dfe79d30923ef3bce4e3eaad99c8f
3
- size 68249
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e404c29af3690234566f478c1222f958cad44610e54e4b4aaf241487b93a965b
3
+ size 78226
data/tweet_hate_balance_test1_seed0/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:68682ac32892c1035801f2ab70fbc67207e918b1e9f3efad0e0f9a7f01c68298
3
- size 237191
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc9e87d836f11dd9299e32ea47e0e2784ba9429886bcc42aea1382f5adbf01ff
3
+ size 291859
data/tweet_hate_balance_test1_seed0/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:69950e69a5f31ff64bcd9b7b8ae120f69daff9696b47f47cc3a7bc81c413d320
3
- size 65388
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae2f50faade3808c072ca527049081658f07618f9060700cffa98c765630d8fc
3
+ size 76128
data/tweet_hate_balance_test1_seed1/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2e1aeb7de399c4365dc04770099609a25a1dc04c132237a1e593fe1ed4438ed
3
- size 234551
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba9aca6ff3afc523d219c3f7966a93ee5a6cc75a7610dbee9f534b23bb5f47a2
3
+ size 289033
data/tweet_hate_balance_test1_seed1/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c4da0adbfef49e9bf034ad5c563b0bdb340d5ec1b1a54e2ee00f28338f309c6c
3
- size 67091
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d38d1f5afa509654fd3b7134137a0d05e0743f852f12017a3227e010f99047a5
3
+ size 76864
data/tweet_hate_balance_test1_seed2/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ce351678af0f47fd3531542347546d600a315bd26b85e7d348a4a24c050a84c9
3
- size 234763
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58af673485bc6761b65e644192c5c6dc9de819f00b250d5e9746ed03070690c5
3
+ size 290295
data/tweet_hate_balance_test1_seed2/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6a96d25d7817ea108eaee70d6e8f7e9375ec44b2f24bc8089cd7a0a5b92fd91c
3
- size 67791
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59214336e48b3949416781e43bca77d688e655a7439bae52d204ecc21b00d950
3
+ size 77733
data/tweet_hate_balance_test2_seed0/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:55211d9187834d7a5fae69fe602b0bc091c7ce805aba158f48182f14654ee727
3
- size 235362
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe6292739dda03866b2b54d4e45c1002b0c9988bcf195df9cf17c0ddcafc25c5
3
+ size 290563
data/tweet_hate_balance_test2_seed0/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3e653d2a29d9f58fbb09ce71279498a8015debe535d575b0885016a1fa28b3e9
3
- size 66531
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29ce18a389fba779c7113e564faa4b8c1dbbc7264501800d2a06ee57048e3a69
3
+ size 76330
data/tweet_hate_balance_test2_seed1/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e7554cbb8cc50f9e926c5e365cdaa0177fc818293ab52f58c179f0b5311c5525
3
- size 233823
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b2c06a76d54d53c45a7df7a4cc77f2c9157f0ca5a4650f5bde14f13442607f4
3
+ size 288134
data/tweet_hate_balance_test2_seed1/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f6e3288d55960cfb3c1e1f483e111ef668b9ae35825dd740b9946fe54c552144
3
- size 66826
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebbdf86fea3afb3df85b5f7bb1a0eea24b71c0651adca14817310409f7267c1e
3
+ size 76898
data/tweet_hate_balance_test2_seed2/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:375ba99f71d530778f268ad21a9696464f63fa2fa730882ab17f815b7797f7d8
3
- size 232849
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6a7fac6e2e756cc5d34fd37bb22cc75740c05fca7ce197579c3fbd847e8e7c9
3
+ size 288973
data/tweet_hate_balance_test2_seed2/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:34d5b18e3b1a54f2c8597df8e7da35aa244fa7f3f77349e80d91eb7447b02ea1
3
- size 68113
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7e6b5b1c9692e01540683242b32f1bcdd471d555baff291eb827a53c759c7e2
3
+ size 77829
data/tweet_hate_balance_test3_seed0/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bd22189bc7e9017c71d2d665e0b4176fae86865cfbe99c651619f00e8e559fe4
3
- size 238221
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6cf1b415bdf7bb18b9f2f27d10cad2fb87f324a8eef7f902907c5676de5deb3e
3
+ size 292162
data/tweet_hate_balance_test3_seed0/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:712617427f44f86d08a5f0c4b818da24949f12359234d4796fd2e29370f1fda0
3
- size 66470
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2dbc08964d055f321d8931f0b901eb063d39471e47319e18bf0c8a1b0c3245e6
3
+ size 77106
data/tweet_hate_balance_test3_seed1/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3c82b8c70c3c28054f64ec0e33cdfc00538a6ac8d42965238f4a638bbbee38b8
3
- size 237066
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80948fd771616aa7fbe99091cfaf5eff7f455d7d6deafe58f3fbaff5ba126622
3
+ size 288698
data/tweet_hate_balance_test3_seed1/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dbd69e47a1ada75b2207d95053ade9f079c958eae94697435fa2db56dabd2f29
3
- size 65788
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b22170e691e3eaa936b309285cd9c55ef1c1102a0fd6f57eb239645226a15e6
3
+ size 76747
data/tweet_hate_balance_test3_seed2/train.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e42c31ca186b19c6b736a160f48bfe40f3d5976b08b64774c7ae0d28ef5b8755
3
- size 234090
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06436ce3a32b66fed8cbcbd7d72671d7c27c18e2c70cd77a76ff6f968a917c73
3
+ size 289822
data/tweet_hate_balance_test3_seed2/validation.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1f5ba6afada2e16da5837e63cbc4603723c8015e8c3f04f90c9714f601bea144
3
- size 68723
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d3f8f0b20cf24ec927ffcdee5d62d72d3f3d0d6909e03c6cf2dc597bee8ec25
3
+ size 77757
experiments/main.sh CHANGED
@@ -2,41 +2,59 @@
2
  MODEL="roberta-base"
3
  MODEL="vinai/bertweet-base"
4
  MODEL="jhu-clsp/bernice"
5
- MODEL="roberta-large"
6
- MODEL="vinai/bertweet-large"
7
  MODEL="cardiffnlp/twitter-roberta-base-2019-90m"
8
  MODEL="cardiffnlp/twitter-roberta-base-dec2020"
9
  MODEL="cardiffnlp/twitter-roberta-base-2021-124m"
10
- MODEL="cardiffnlp/twitter-roberta-base-2022-154m"
 
11
  MODEL="cardiffnlp/twitter-roberta-large-2022-154m"
12
 
 
 
 
13
  # EMOJI
14
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_temporal"
15
- rm -rf ckpt/emoji*${MODEL##*/}*
16
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed0"
17
- rm -rf ckpt/emoji*${MODEL##*/}*
18
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed0"
19
- rm -rf ckpt/emoji*${MODEL##*/}*
20
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed0"
21
- rm -rf ckpt/emoji*${MODEL##*/}*
22
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed0"
23
- rm -rf ckpt/emoji*${MODEL##*/}*
24
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed1"
25
- rm -rf ckpt/emoji*${MODEL##*/}*
26
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed1"
27
- rm -rf ckpt/emoji*${MODEL##*/}*
28
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed1"
29
- rm -rf ckpt/emoji*${MODEL##*/}*
30
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed1"
31
- rm -rf ckpt/emoji*${MODEL##*/}*
32
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed2"
33
- rm -rf ckpt/emoji*${MODEL##*/}*
34
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed2"
35
- rm -rf ckpt/emoji*${MODEL##*/}*
36
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed2"
37
- rm -rf ckpt/emoji*${MODEL##*/}*
38
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed2"
39
- rm -rf ckpt/emoji*${MODEL##*/}*
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
40
 
41
  # HATE
42
  python model_finetuning_hate.py -m "${MODEL}" -d "hate_temporal"
 
2
  MODEL="roberta-base"
3
  MODEL="vinai/bertweet-base"
4
  MODEL="jhu-clsp/bernice"
 
 
5
  MODEL="cardiffnlp/twitter-roberta-base-2019-90m"
6
  MODEL="cardiffnlp/twitter-roberta-base-dec2020"
7
  MODEL="cardiffnlp/twitter-roberta-base-2021-124m"
8
+ MODEL="roberta-large"
9
+ MODEL="vinai/bertweet-large"
10
  MODEL="cardiffnlp/twitter-roberta-large-2022-154m"
11
 
12
+ # todo
13
+ MODEL="cardiffnlp/twitter-roberta-base-2022-154m"
14
+
15
  # EMOJI
16
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_temporal"
 
17
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed0"
 
18
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed0"
 
19
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed0"
 
20
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed0"
 
21
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed1"
 
22
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed1"
 
23
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed1"
 
24
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed1"
 
25
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random0_seed2"
 
26
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random1_seed2"
 
27
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random2_seed2"
 
28
  python model_finetuning_emoji.py -m "${MODEL}" -d "emoji_random3_seed2"
29
+
30
+ # HATE
31
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_temporal"
32
+ rm -rf ckpt/hate*${MODEL##*/}*
33
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random0_seed0"
34
+ rm -rf ckpt/hate*${MODEL##*/}*
35
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random1_seed0"
36
+ rm -rf ckpt/hate*${MODEL##*/}*
37
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random2_seed0"
38
+ rm -rf ckpt/hate*${MODEL##*/}*
39
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random3_seed0"
40
+ rm -rf ckpt/hate*${MODEL##*/}*
41
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random0_seed1"
42
+ rm -rf ckpt/hate*${MODEL##*/}*
43
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random1_seed1"
44
+ rm -rf ckpt/hate*${MODEL##*/}*
45
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random2_seed1"
46
+ rm -rf ckpt/hate*${MODEL##*/}*
47
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random3_seed1"
48
+ rm -rf ckpt/hate*${MODEL##*/}*
49
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random0_seed2"
50
+ rm -rf ckpt/hate*${MODEL##*/}*
51
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random1_seed2"
52
+ rm -rf ckpt/hate*${MODEL##*/}*
53
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random2_seed2"
54
+ rm -rf ckpt/hate*${MODEL##*/}*
55
+ python model_finetuning_hate.py -m "${MODEL}" -d "hate_balance_random3_seed2"
56
+ rm -rf ckpt/hate*${MODEL##*/}*
57
+
58
 
59
  # HATE
60
  python model_finetuning_hate.py -m "${MODEL}" -d "hate_temporal"
process/tweet_hate_balance.py CHANGED
@@ -34,15 +34,17 @@ def sampler(chunk_index, r_seed):
34
 
35
  label_train_tmp_dist = dict(zip(*np.unique(df_train_tmp["gold_label_binary"], return_counts=True)))
36
  assert label_train_tmp_dist[0] < label_train_dist[0] and label_train_tmp_dist[1] < label_train_dist[1]
37
- df_train_tmp_p = df_train_positive.sample(label_train_dist[0] - label_train_tmp_dist[0], random_state=r_seed)
38
- df_train_tmp_n = df_train_positive.sample(label_train_dist[1] - label_train_tmp_dist[1], random_state=r_seed)
39
  df_train_tmp = pd.concat([df_train_tmp, df_train_tmp_p, df_train_tmp_n])
 
40
 
41
  label_val_tmp_dist = dict(zip(*np.unique(df_val_tmp["gold_label_binary"], return_counts=True)))
42
  assert label_val_tmp_dist[0] < label_val_dist[0] and label_val_tmp_dist[1] < label_val_dist[1]
43
- df_val_tmp_p = df_val_positive.sample(label_val_dist[0] - label_val_tmp_dist[0], random_state=r_seed)
44
- df_val_tmp_n = df_val_positive.sample(label_val_dist[1] - label_val_tmp_dist[1], random_state=r_seed)
45
  df_val_tmp = pd.concat([df_val_tmp, df_val_tmp_p, df_val_tmp_n])
 
46
  return list(df_train_tmp.T.to_dict().values()), list(df_val_tmp.T.to_dict().values())
47
 
48
  for n in range(4):
 
34
 
35
  label_train_tmp_dist = dict(zip(*np.unique(df_train_tmp["gold_label_binary"], return_counts=True)))
36
  assert label_train_tmp_dist[0] < label_train_dist[0] and label_train_tmp_dist[1] < label_train_dist[1]
37
+ df_train_tmp_p = df_train_positive.sample(label_train_dist[1] - label_train_tmp_dist[1], random_state=r_seed)
38
+ df_train_tmp_n = df_train_negative.sample(label_train_dist[0] - label_train_tmp_dist[0], random_state=r_seed)
39
  df_train_tmp = pd.concat([df_train_tmp, df_train_tmp_p, df_train_tmp_n])
40
+ assert dict(zip(*np.unique(df_train_tmp["gold_label_binary"], return_counts=True))) == label_train_dist
41
 
42
  label_val_tmp_dist = dict(zip(*np.unique(df_val_tmp["gold_label_binary"], return_counts=True)))
43
  assert label_val_tmp_dist[0] < label_val_dist[0] and label_val_tmp_dist[1] < label_val_dist[1]
44
+ df_val_tmp_p = df_val_positive.sample(label_val_dist[1] - label_val_tmp_dist[1], random_state=r_seed)
45
+ df_val_tmp_n = df_val_negative.sample(label_val_dist[0] - label_val_tmp_dist[0], random_state=r_seed)
46
  df_val_tmp = pd.concat([df_val_tmp, df_val_tmp_p, df_val_tmp_n])
47
+ assert dict(zip(*np.unique(df_val_tmp["gold_label_binary"], return_counts=True))) == label_val_dist
48
  return list(df_train_tmp.T.to_dict().values()), list(df_val_tmp.T.to_dict().values())
49
 
50
  for n in range(4):
tweet_temporal_shift.py CHANGED
@@ -2,7 +2,7 @@
2
  import json
3
  import datasets
4
 
5
- _VERSION = "1.1.3"
6
  _TWEET_TEMPORAL_DESCRIPTION = """"""
7
  _TWEET_TEMPORAL_CITATION = """"""
8
  _TWEET_TOPIC_DESCRIPTION = """
 
2
  import json
3
  import datasets
4
 
5
+ _VERSION = "1.1.4"
6
  _TWEET_TEMPORAL_DESCRIPTION = """"""
7
  _TWEET_TEMPORAL_CITATION = """"""
8
  _TWEET_TOPIC_DESCRIPTION = """