Bump to exllamav2 0.2.3 and fix for API changes

backend/sessions.py

@@ -533,7 +533,9 @@ class Session:
                     gen_settings = gen_settings,
                     token_healing = p_healing,
                     abort_event = abort_event,
-                    banned_strings = banned_strings
+                    banned_strings = banned_strings,
+                    filters = gen_settings.filters,
+                    filter_prefer_eos = gen_settings.filters
                 )
                 if abort_event.is_set():
                     abort_event.clear()

requirements.txt

@@ -1,5 +1,5 @@
 torch>=2.1.0
 pynvml
-exllamav2>=0.2.0
+exllamav2>=0.2.3
 Flask>=2.3.2
 waitress>=2.1.2