QCRI
/

AliShahroor commited on
Commit
4bf4f99
·
verified ·
1 Parent(s): 3676651

add results

Browse files
Files changed (1) hide show
  1. README.md +78 -0
README.md CHANGED
@@ -71,6 +71,84 @@ response = tokenizer.decode(output[0], skip_special_tokens=True)
71
  print(response)
72
  ```
73
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
74
  ## Paper
75
  For an in-depth understanding, refer to our paper: [**LlamaLens: Specialized Multilingual LLM for Analyzing News and Social Media Content**](https://arxiv.org/pdf/2410.15308).
76
 
 
71
  print(response)
72
  ```
73
 
74
+
75
+ ## Results
76
+
77
+ Below, we present the performance of **LlamaLens** compared to existing SOTA (if available) and the Llama-Instruct baseline, The “Delta” column here is
78
+ calculated as **(LLamalens – SOTA)**.
79
+
80
+ ---
81
+
82
+ ## Arabic
83
+
84
+ | **Task** | **Dataset** | **Metric** | **SOTA** | **Llama-instruct** | **LLamalens** | **Delta** (LLamalens - SOTA) |
85
+ |------------------------|---------------------------|-----------:|--------:|--------------------:|--------------:|------------------------------:|
86
+ | News Summarization | xlsum | R-2 | 0.137 | 0.034 | 0.075 | -0.062 |
87
+ | News Genre | ASND | Ma-F1 | 0.770 | 0.587 | 0.938 | 0.168 |
88
+ | News Genre | SANADAkhbarona | Acc | 0.940 | 0.784 | 0.922 | -0.018 |
89
+ | News Genre | SANADAlArabiya | Acc | 0.974 | 0.893 | 0.986 | 0.012 |
90
+ | News Genre | SANADAlkhaleej | Acc | 0.986 | 0.865 | 0.967 | -0.019 |
91
+ | News Genre | UltimateDataset | Ma-F1 | 0.970 | 0.376 | 0.883 | -0.087 |
92
+ | News Credibility | NewsCredibility | Acc | 0.899 | 0.455 | 0.494 | -0.405 |
93
+ | Emotion | Emotional-Tone | W-F1 | 0.658 | 0.358 | 0.748 | 0.090 |
94
+ | Emotion | NewsHeadline | Acc | 1.000 | 0.406 | 0.551 | -0.449 |
95
+ | Sarcasm | ArSarcasm-v2 | F1_Pos | 0.584 | 0.477 | 0.307 | -0.277 |
96
+ | Sentiment | ar_reviews_100k | F1_Pos | – | 0.343 | 0.665 | – |
97
+ | Sentiment | ArSAS | Acc | 0.920 | 0.603 | 0.795 | -0.125 |
98
+ | Stance | stance | Ma-F1 | 0.767 | 0.608 | 0.936 | 0.169 |
99
+ | Stance | Mawqif-Arabic-Stance | Ma-F1 | 0.789 | 0.764 | 0.867 | 0.078 |
100
+ | Att.worthiness | CT22Attentionworthy | W-F1 | 0.412 | 0.158 | 0.544 | 0.132 |
101
+ | Checkworthiness | CT24_T1 | F1_Pos | 0.569 | 0.404 | 0.877 | 0.308 |
102
+ | Claim | CT22Claim | Acc | 0.703 | 0.581 | 0.778 | 0.075 |
103
+ | Factuality | Arafacts | Mi-F1 | 0.850 | 0.210 | 0.534 | -0.316 |
104
+ | Factuality | COVID19Factuality | W-F1 | 0.831 | 0.492 | 0.781 | -0.050 |
105
+ | Propaganda | ArPro | Mi-F1 | 0.767 | 0.597 | 0.762 | -0.005 |
106
+ | Cyberbullying | ArCyc_CB | Acc | 0.863 | 0.766 | 0.753 | -0.110 |
107
+ | Harmfulness | CT22Harmful | F1_Pos | 0.557 | 0.507 | 0.508 | -0.049 |
108
+ | Hate Speech | annotated-hatetweets-4 | W-F1 | 0.630 | 0.257 | 0.549 | -0.081 |
109
+ | Hate Speech | OSACT4SubtaskB | Mi-F1 | 0.950 | 0.819 | 0.802 | -0.148 |
110
+ | Offensive | ArCyc_OFF | Ma-F1 | 0.878 | 0.489 | 0.652 | -0.226 |
111
+ | Offensive | OSACT4SubtaskA | Ma-F1 | 0.905 | 0.782 | 0.899 | -0.006 |
112
+
113
+ ---
114
+
115
+ ## English
116
+
117
+ | **Task** | **Dataset** | **Metric** | **SOTA** | **Llama-instruct** | **LLamalens** | **Delta** (LLamalens - SOTA) |
118
+ |----------------------|---------------------------|-----------:|--------:|--------------------:|--------------:|------------------------------:|
119
+ | News Summarization | xlsum | R-2 | 0.152 | 0.074 | 0.141 | -0.011 |
120
+ | News Genre | CNN_News_Articles | Acc | 0.940 | 0.644 | 0.915 | -0.025 |
121
+ | News Genre | News_Category | Ma-F1 | 0.769 | 0.970 | 0.505 | -0.264 |
122
+ | News Genre | SemEval23T3-ST1 | Mi-F1 | 0.815 | 0.687 | 0.241 | -0.574 |
123
+ | Subjectivity | CT24_T2 | Ma-F1 | 0.744 | 0.535 | 0.508 | -0.236 |
124
+ | Emotion | emotion | Ma-F1 | 0.790 | 0.353 | 0.878 | 0.088 |
125
+ | Sarcasm | News-Headlines | Acc | 0.897 | 0.668 | 0.956 | 0.059 |
126
+ | Sentiment | NewsMTSC | Ma-F1 | 0.817 | 0.628 | 0.627 | -0.190 |
127
+ | Checkworthiness | CT24_T1 | F1_Pos | 0.753 | 0.404 | 0.877 | 0.124 |
128
+ | Claim | claim-detection | Mi-F1 | – | 0.545 | 0.915 | – |
129
+ | Factuality | News_dataset | Acc | 0.920 | 0.654 | 0.946 | 0.026 |
130
+ | Factuality | Politifact | W-F1 | 0.490 | 0.121 | 0.290 | -0.200 |
131
+ | Propaganda | QProp | Ma-F1 | 0.667 | 0.759 | 0.851 | 0.184 |
132
+ | Cyberbullying | Cyberbullying | Acc | 0.907 | 0.175 | 0.847 | -0.060 |
133
+ | Offensive | Offensive_Hateful | Mi-F1 | – | 0.692 | 0.805 | – |
134
+ | Offensive | offensive_language | Mi-F1 | 0.994 | 0.646 | 0.884 | -0.110 |
135
+ | Offensive & Hate | hate-offensive-speech | Acc | 0.945 | 0.602 | 0.924 | -0.021 |
136
+
137
+ ---
138
+
139
+ ## Hindi
140
+
141
+ | **Task** | **Dataset** | **Metric** | **SOTA** | **Llama-instruct** | **LLamalens** | **Delta** (LLamalens - SOTA) |
142
+ |------------------------|------------------------|-----------:|--------:|--------------------:|--------------:|------------------------------:|
143
+ | NLI | NLI_dataset | W-F1 | 0.646 | 0.633 | 0.655 | 0.009 |
144
+ | News Summarization | xlsum | R-2 | 0.136 | 0.078 | 0.117 | -0.019 |
145
+ | Sentiment | Sentiment Analysis | Acc | 0.697 | 0.552 | 0.669 | -0.028 |
146
+ | Factuality | fake-news | Mi-F1 | – | 0.759 | 0.713 | – |
147
+ | Hate Speech | hate-speech-detection | Mi-F1 | 0.639 | 0.750 | 0.994 | 0.355 |
148
+ | Hate Speech | Hindi-Hostility | W-F1 | 0.841 | 0.469 | 0.720 | -0.121 |
149
+ | Offensive | Offensive Speech | Mi-F1 | 0.723 | 0.621 | 0.847 | 0.124 |
150
+ | Cyberbullying | MC_Hinglish1 | Acc | 0.609 | 0.233 | 0.587 | -0.022 |
151
+
152
  ## Paper
153
  For an in-depth understanding, refer to our paper: [**LlamaLens: Specialized Multilingual LLM for Analyzing News and Social Media Content**](https://arxiv.org/pdf/2410.15308).
154