@@ -13,28 +13,28 @@ def test_infill_without_input_extra():
1313 global server
1414 server .start ()
1515 res = server .make_request ("POST" , "/infill" , data = {
16- "prompt " : "Complete this " ,
17- "input_prefix " : "#include <cstdio> \n #include \" llama.h \" \n \n int main() { \n int n_threads = llama_" ,
16+ "input_prefix " : "#include <cstdio> \n #include \" llama.h \" \n \n int main() { \n " ,
17+ "prompt " : " int n_threads = llama_" ,
1818 "input_suffix" : "}\n " ,
1919 })
2020 assert res .status_code == 200
21- assert match_regex ("(One|day|she|saw|big|scary|bird )+" , res .body ["content" ])
21+ assert match_regex ("(Ann|small|shiny )+" , res .body ["content" ])
2222
2323
2424def test_infill_with_input_extra ():
2525 global server
2626 server .start ()
2727 res = server .make_request ("POST" , "/infill" , data = {
28- "prompt" : "Complete this" ,
2928 "input_extra" : [{
3029 "filename" : "llama.h" ,
3130 "text" : "LLAMA_API int32_t llama_n_threads();\n "
3231 }],
33- "input_prefix" : "#include <cstdio>\n #include \" llama.h\" \n \n int main() {\n int n_threads = llama_" ,
32+ "input_prefix" : "#include <cstdio>\n #include \" llama.h\" \n \n int main() {\n " ,
33+ "prompt" : " int n_threads = llama_" ,
3434 "input_suffix" : "}\n " ,
3535 })
3636 assert res .status_code == 200
37- assert match_regex ("(help|find|band )+" , res .body ["content" ])
37+ assert match_regex ("(Dad|excited|park )+" , res .body ["content" ])
3838
3939
4040@pytest .mark .parametrize ("input_extra" , [
@@ -65,12 +65,12 @@ def test_with_qwen_model():
6565 server .model_hf_file = "qwen2.5-coder-1.5b-iq3_xxs-imat.gguf"
6666 server .start (timeout_seconds = 600 )
6767 res = server .make_request ("POST" , "/infill" , data = {
68- # "prompt": "Complete this", # FIXME: add more complicated prompt when format_infill is fixed
6968 "input_extra" : [{
7069 "filename" : "llama.h" ,
7170 "text" : "LLAMA_API int32_t llama_n_threads();\n "
7271 }],
73- "input_prefix" : "#include <cstdio>\n #include \" llama.h\" \n \n int main() {\n int n_threads = llama_" ,
72+ "input_prefix" : "#include <cstdio>\n #include \" llama.h\" \n \n int main() {\n " ,
73+ "prompt" : " int n_threads = llama_" ,
7474 "input_suffix" : "}\n " ,
7575 })
7676 assert res .status_code == 200
0 commit comments