fix: indent

app.py

@@ -92,8 +92,6 @@ def display_streamlit_sidebar():
     max_new_tokens = form.slider(label="max_new_tokens", min_value=32, max_value=512, step=1, value=st.session_state["max_new_tokens"])
     repetition_penalty = form.slider(label="repetition_penalty", min_value=0.0, max_value=5.0, step=0.01, value=st.session_state["repetition_penalty"])
 
-    form.write('Best practice is to only modify temperature or top_p, not both')
-
     submitted = form.form_submit_button("Start session")
     if submitted and not st.session_state['session_started']:
         print('Parameters updated...')
@@ -122,7 +120,7 @@ def init_session_state():
     st.session_state["top_k"] = 40
     st.session_state["max_new_tokens"] = 512
     st.session_state["repetition_penalty"] = 1.1
-    st.session_state["system_message"] = "You are a medical expert that provides answers for a medically trained audience"  
+    st.session_state["system_prompt"] = "You are a medical expert that provides answers for a medically trained audience"
 
 def get_genders():
     return ['Male', 'Female']
@@ -156,135 +154,6 @@ def display_session_overview():
         st.write("Total compute time (ms): " + str(total_time))
 
 
-def remove_question(question_id):
-    st.session_state["questions"] = [value for value in st.session_state["questions"] if
-                                     str(value["question_id"]) != str(question_id)]
-    st.rerun()
-
-
-def get_prompt_from_lead_symptom(df_config, df_prompt, lead_symptom, lang, fallback=True):
-    de_lead_symptom = lead_symptom
-
-    if lang != "DE":
-        df_lead_symptom = df_config[df_config[lang + ": Symptome"] == lead_symptom]
-        de_lead_symptom = df_lead_symptom["DE: Symptome"].iloc[0]
-        print("DE lead symptom: " + de_lead_symptom)
-
-    for i, row in df_prompt.iterrows():
-        if de_lead_symptom in row["Questionnaire"]:
-            return row["Prompt"]
-
-    warning_text = "No guidelines found for lead symptom " + lead_symptom + " (" + de_lead_symptom + ")"
-    if fallback:
-        st.toast(warning_text + ", using generic prompt", icon='🚨')
-        return st.session_state["system_prompt"]
-    st.toast(warning_text, icon='🚨')
-
-    return ""
-
-
-def get_scenarios(df):
-    return [v for v in df.columns.values if v.startswith('TLC') or v.startswith('GP')]
-
-
-def get_gender_age_from_test_scenario(test_scenario):
-    try:
-        result = re.search(r"([FM])(\d+)", test_scenario)
-        res_age = int(result.group(2))
-        gender = result.group(1)
-        res_gender = None
-        if gender == "M":
-            res_gender = "Male"
-        elif gender == "F":
-            res_gender = "Female"
-        else:
-            raise Exception('Unexpected gender')
-
-        return res_gender, res_age
-
-    except:
-        st.error("Unable to extract name, gender; using 30M as default")
-        return "Male", 30
-
-def get_freetext_to_reco(reco_freetext_cased, emg_class_enabled=False):
-    reco_freetext = ""
-    if reco_freetext_cased:
-        reco_freetext = reco_freetext_cased.lower()
-
-    if reco_freetext.startswith('treat remotely') or reco_freetext.startswith('telecare'):
-        return 'TELECARE'
-    if reco_freetext.startswith('treat ad-real') or reco_freetext.startswith('gp') \
-            or reco_freetext.startswith('general practitioner'):
-        return 'GP'
-    if reco_freetext.startswith('emergency') or reco_freetext.startswith('emg') \
-            or reco_freetext.startswith('urgent'):
-        if emg_class_enabled:
-            return 'EMERGENCY'
-        return 'GP'
-
-    if "gp" in reco_freetext or 'general practitioner' in reco_freetext \
-            or "nicht über tele" in reco_freetext or 'durch einen arzt erford' in reco_freetext \
-            or "persönliche untersuchung erfordert" in reco_freetext:
-        return 'GP'
-
-    if ("telecare" in reco_freetext or 'telemed' in reco_freetext or
-            'can be treated remotely' in reco_freetext):
-        return 'TELECARE'
-
-    if ('emergency' in reco_freetext or 'urgent' in reco_freetext or
-            'not be treated remotely' in reco_freetext or "nicht tele" in reco_freetext):
-        return 'GP'
-
-    warning_msg = 'Cannot extract reco from LLM text: ' + reco_freetext
-    st.toast(warning_msg)
-    print(warning_msg)
-    return 'TRIAGE_IMPOSSIBLE'
-
-
-def get_structured_reco(row, index, emg_class_enabled):
-    freetext_reco_col_name = "llm_reco_freetext_" + str(index)
-    freetext_reco = row[freetext_reco_col_name].lower()
-    return get_freetext_to_reco(freetext_reco, emg_class_enabled)
-
-
-def add_expected_dispo(row, emg_class_enabled):
-    disposition = row["disposition"]
-    if disposition == "GP" or disposition == "TELECARE":
-        return disposition
-    if disposition == "EMERGENCY":
-        if emg_class_enabled:
-            return "EMERGENCY"
-        return "GP"
-
-    raise Exception("Missing disposition for row " + str(row.name) + " with summary " + row["case_summary"])
-
-
-def get_test_scenarios(df):
-    res = []
-    for col in df.columns.values:
-        if str(col).startswith('GP') or str(col).startswith('TLC'):
-            res.append(col)
-    return res
-
-
-def get_transcript(df, test_scenario, lang):
-    transcript = ""
-    for i, row in df.iterrows():
-        transcript += "\nDoctor: " + row[lang + ": Fragen"]
-        transcript += ", Patient: " + str(row[test_scenario])
-    return transcript
-
-
-def get_expected_from_scenario(test_scenario):
-    reco = test_scenario.split('_')[0]
-    if reco == "GP":
-        return "GP"
-    elif reco == "TLC":
-        return "TELECARE"
-    else:
-        raise Exception('Unexpected reco: ' + reco)
-
-
 def plot_report(title, expected, predicted, display_labels):
     st.markdown('#### ' + title)
     conf_matrix = confusion_matrix(expected, predicted, labels=display_labels)
@@ -312,49 +181,23 @@ def plot_report(title, expected, predicted, display_labels):
         pass
 
 
-def get_complete_prompt(generic_prompt, guidelines_prompt):
-    complete_prompt = ""
-    if generic_prompt:
-        complete_prompt += generic_prompt
-
-    if generic_prompt and guidelines_prompt:
-        complete_prompt += ".\n\n"
-
-    if guidelines_prompt:
-        complete_prompt += guidelines_prompt
-
-    return complete_prompt
-
-
-def run_command(args):
-    """Run command, transfer stdout/stderr back into Streamlit and manage error"""
-    cmd = ' '.join(args)
-    result = subprocess.run(cmd, capture_output=True, text=True)
-    print(result)
-
-def get_diarized_f_path(audio_f_name):
-    # TODO p2: Quick hack, cleaner with os or regexes
-    base_name = audio_f_name.split('.')[0]
-    return DATA_FOLDER + base_name + ".txt"
-
-
 def get_prompt_format(model_name):
     if model_name == "TheBloke/Llama-2-13B-chat-GPTQ":
         return '''[INST] <<SYS>>
-        You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe.  Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.
-        <</SYS>>
-        {prompt}[/INST]
+                You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe.  Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.
+                <</SYS>>
+                {prompt}[/INST]
 
-        '''
+                '''
     if model_name == "TheBloke/Llama-2-7B-Chat-GPTQ":
         return "[INST] <<SYS>>＼nYou are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe.  Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.＼n<</SYS>>＼n{prompt}[/INST]"
     
     if model_name == "TheBloke/meditron-7B-GPTQ" or model_name == "TheBloke/meditron-70B-GPTQ":
         return '''<|im_start|>system
-        {system_message}<|im_end|>
-        <|im_start|>user
-        {prompt}<|im_end|>
-        <|im_start|>assistant'''
+                {system_message}<|im_end|>
+                <|im_start|>user
+                {prompt}<|im_end|>
+                <|im_start|>assistant'''
     
     return ""
 
@@ -370,7 +213,7 @@ def display_llm_output():
 
     prompt_format_str = get_prompt_format(st.session_state["model_name_or_path"])
     prompt_format = form.text_area('Prompt format', value=prompt_format_str)
-    system_prompt = form.text_area('System prompt', value=st.session_state["system_prompt"])
+    system_prompt = ""#form.text_area('System prompt', value=st.session_state["system_prompt"])
     prompt = form.text_area('Prompt', value=st.session_state["prompt"])
 
     submitted = form.form_submit_button('Submit')