ahmed.ibrahim39699
/
JanetBackEnd
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

working_backend

This commit is contained in:
ahmed531998 2023-04-10 00:47:26 +02:00
parent 7e1bf49d0d
commit 79489fbf68
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
ResponseGenerator.py
View File

@ -81,7 +81,7 @@ class ResponseGenerator:
return "I am sorry, I cannot answer to this kind of language"
elif action == "ConvGen":
gen_kwargs = {"length_penalty": 2.5, "num_beams":2, "max_length": 30}
gen_kwargs = {"length_penalty": 2.5, "num_beams":2, "max_length": 30, "repetition_penalty": 2.5, "temperature": 2}
answer = self.generators['chat']('history: '+ consec_history + ' ' + utterance + ' persona: ' + 'I am Janet. My name is Janet. I am an AI developed by CNR to help VRE users.' , **gen_kwargs)[0]['generated_text']
return answer
@ -147,7 +147,7 @@ class ResponseGenerator:
gen_seq = 'question: '+utterance+" context: "+content
#handle return random 2 answers
gen_kwargs = {"length_penalty": 0.5, "num_beams":2, "max_length": 60}
gen_kwargs = {"length_penalty": 0.5, "num_beams":2, "max_length": 60, "repetition_penalty": 2.5, "temperature": 2}
answer = self.generators['qa'](gen_seq, **gen_kwargs)[0]['generated_text']
return str(answer)
@ -165,7 +165,7 @@ class ResponseGenerator:
answer = ""
for i, row in df.iterrows():
gen_seq = 'summarize: '+row['content']
gen_kwargs = {"length_penalty": 1.5, "num_beams":6, "max_length": 120}
gen_kwargs = {"length_penalty": 1.5, "num_beams":6, "max_length": 120, "repetition_penalty": 2.5, "temperature": 2}
answer = self.generators['summ'](gen_seq, **gen_kwargs)[0]['generated_text'] + ' '
return answer
elif action == "Clarify":
@ -176,6 +176,6 @@ class ResponseGenerator:
if len(self.dataset) == 0:
return 'Please specify the title, the topic of the dataset of interest.'
else:
gen_kwargs = {"length_penalty": 2.5, "num_beams":8, "max_length": 120}
gen_kwargs = {"length_penalty": 2.5, "num_beams":8, "max_length": 120, "repetition_penalty": 2.5, "temperature": 2}
question = self.generators['amb']('question: '+ utterance + ' context: ' + consec_history , **gen_kwargs)[0]['generated_text']
return question