shreyasmeher commited on
Commit
c92d20a
·
verified ·
1 Parent(s): 2f4249a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +20 -12
README.md CHANGED
@@ -40,9 +40,11 @@ inference:
40
 
41
  # ConflLlama: Domain-Specific LLM for Conflict Event Classification
42
 
43
- \<p align="center"\>
44
- \<img src="images/logo.png" alt="Project Logo" width="300"/\>
45
- \</p\>
 
 
46
 
47
  **ConflLlama** is a large language model fine-tuned to classify conflict events from text descriptions. This repository contains the GGUF quantized models (q4\_k\_m, q8\_0, and BF16) based on **Llama-3.1 8B**, which have been adapted for the specialized domain of political violence research.
48
 
@@ -93,9 +95,11 @@ The most significant improvements were observed in historically difficult-to-cla
93
  * Alpha (`lora_alpha`): 16
94
  * Target Modules: `q_proj`, `k_proj`, `v_proj`, `o_proj`, `gate_proj`, `up_proj`, `down_proj`
95
 
96
- \<p align="center"\>
97
- \<img src="images/model-arch.png" alt="Model Training Architecture" width="800"/\>
98
- \</p\>
 
 
99
 
100
  ### Training Data
101
 
@@ -103,9 +107,11 @@ The most significant improvements were observed in historically difficult-to-cla
103
  * **Time Period**: The training dataset consists of 171,514 events that occurred before January 1, 2017. The test set includes 38,192 events from 2017 onwards.
104
  * **Preprocessing**: The pipeline filters data by date, cleans text summaries, and combines primary, secondary, and tertiary attack types into a single multi-label field.
105
 
106
- \<p align="center"\>
107
- \<img src="images/preprocessing.png" alt="Data Preprocessing Pipeline" width="800"/\>
108
- \</p\>
 
 
109
 
110
  -----
111
 
@@ -134,9 +140,11 @@ This model is designed for academic and research purposes within the fields of p
134
 
135
  ## Training Logs
136
 
137
- \<p align="center"\>
138
- \<img src="images/training.png" alt="Training Logs" width="800"/\>
139
- \</p\>
 
 
140
 
141
  The training logs show a successful training run with healthy convergence patterns:
142
 
 
40
 
41
  # ConflLlama: Domain-Specific LLM for Conflict Event Classification
42
 
43
+ <p align="center">
44
+
45
+   <img src="images/logo.png" alt="Project Logo" width="300"/>
46
+
47
+ </p>
48
 
49
  **ConflLlama** is a large language model fine-tuned to classify conflict events from text descriptions. This repository contains the GGUF quantized models (q4\_k\_m, q8\_0, and BF16) based on **Llama-3.1 8B**, which have been adapted for the specialized domain of political violence research.
50
 
 
95
  * Alpha (`lora_alpha`): 16
96
  * Target Modules: `q_proj`, `k_proj`, `v_proj`, `o_proj`, `gate_proj`, `up_proj`, `down_proj`
97
 
98
+ <p align="center">
99
+
100
+   <img src="images/model-arch.png" alt="Model Training Architecture" width="800"/>
101
+
102
+ </p>
103
 
104
  ### Training Data
105
 
 
107
  * **Time Period**: The training dataset consists of 171,514 events that occurred before January 1, 2017. The test set includes 38,192 events from 2017 onwards.
108
  * **Preprocessing**: The pipeline filters data by date, cleans text summaries, and combines primary, secondary, and tertiary attack types into a single multi-label field.
109
 
110
+ <p align="center">
111
+
112
+   <img src="images/preprocessing.png" alt="Data Preprocessing Pipeline" width="800"/>
113
+
114
+ </p>
115
 
116
  -----
117
 
 
140
 
141
  ## Training Logs
142
 
143
+ <p align="center">
144
+
145
+   <img src="images/training.png" alt="Training Logs" width="800"/>
146
+
147
+ </p>
148
 
149
  The training logs show a successful training run with healthy convergence patterns:
150