From 68b79939ee5b834b68ca922994a45208019bde3c Mon Sep 17 00:00:00 2001
From: Hadrien <ketsapiwiq@protonmail.com>
Date: Mon, 20 May 2024 14:36:58 +0200
Subject: [PATCH 1/4] began work on llama

---
 .gitignore                                                 | 3 ++-
 .../persona/prompt_template/gpt_structure.py               | 7 ++++---
 reverie/backend_server/reverie.py                          | 5 +++--
 3 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/.gitignore b/.gitignore
index 5ef9005b6b..64b4a2c04b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,5 +1,6 @@
 reverie/backend_server/utils.py
 frontend_server/settings/local.py
+environment/frontend_server/storage/
 storage/*
 static_dirs/*
 
@@ -99,4 +100,4 @@ ENV/
 .spyderproject
 
 # Rope project settings
-.ropeproject
\ No newline at end of file
+.ropeproject
diff --git a/reverie/backend_server/persona/prompt_template/gpt_structure.py b/reverie/backend_server/persona/prompt_template/gpt_structure.py
index f9c4718949..5e91268671 100644
--- a/reverie/backend_server/persona/prompt_template/gpt_structure.py
+++ b/reverie/backend_server/persona/prompt_template/gpt_structure.py
@@ -12,6 +12,7 @@
 from utils import *
 
 openai.api_key = openai_api_key
+openai.api_base = "http://localhost:11434/v1"
 
 def temp_sleep(seconds=0.1):
   time.sleep(seconds)
@@ -20,7 +21,7 @@ def ChatGPT_single_request(prompt):
   temp_sleep()
 
   completion = openai.ChatCompletion.create(
-    model="gpt-3.5-turbo", 
+    model="llama3:8b-instruct-q8_0", 
     messages=[{"role": "user", "content": prompt}]
   )
   return completion["choices"][0]["message"]["content"]
@@ -46,7 +47,7 @@ def GPT4_request(prompt):
 
   try: 
     completion = openai.ChatCompletion.create(
-    model="gpt-4", 
+    model="llama3:8b-instruct-q8_0", 
     messages=[{"role": "user", "content": prompt}]
     )
     return completion["choices"][0]["message"]["content"]
@@ -273,7 +274,7 @@ def safe_generate_response(prompt,
   return fail_safe_response
 
 
-def get_embedding(text, model="text-embedding-ada-002"):
+def get_embedding(text, model="nomic-embed-text"):
   text = text.replace("\n", " ")
   if not text: 
     text = "this is blank"
diff --git a/reverie/backend_server/reverie.py b/reverie/backend_server/reverie.py
index 2d753d1029..8486bc3816 100644
--- a/reverie/backend_server/reverie.py
+++ b/reverie/backend_server/reverie.py
@@ -600,12 +600,13 @@ def open_server(self):
 
 if __name__ == '__main__':
   # rs = ReverieServer("base_the_ville_isabella_maria_klaus", 
-  #                    "July1_the_ville_isabella_maria_klaus-step-3-1")
+  #                    "Yo")
   # rs = ReverieServer("July1_the_ville_isabella_maria_klaus-step-3-20", 
   #                    "July1_the_ville_isabella_maria_klaus-step-3-21")
   # rs.open_server()
 
-  origin = input("Enter the name of the forked simulation: ").strip()
+  origin = input("Enter the name of the forked simulation [base_the_ville_isabella_maria_klaus]: ").strip()
+  if origin == "": origin = "base_the_ville_isabella_maria_klaus"
   target = input("Enter the name of the new simulation: ").strip()
 
   rs = ReverieServer(origin, target)

From 0a7438ae57c61a34139063c42e0174f62f2f8a5b Mon Sep 17 00:00:00 2001
From: Hadrien <ketsapiwiq@protonmail.com>
Date: Mon, 20 May 2024 15:24:37 +0200
Subject: [PATCH 2/4] fix embeddings

---
 .../backend_server/persona/prompt_template/gpt_structure.py   | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/reverie/backend_server/persona/prompt_template/gpt_structure.py b/reverie/backend_server/persona/prompt_template/gpt_structure.py
index 5e91268671..8e48fd6b63 100644
--- a/reverie/backend_server/persona/prompt_template/gpt_structure.py
+++ b/reverie/backend_server/persona/prompt_template/gpt_structure.py
@@ -273,13 +273,13 @@ def safe_generate_response(prompt,
       print ("~~~~")
   return fail_safe_response
 
+import ollama
 
 def get_embedding(text, model="nomic-embed-text"):
   text = text.replace("\n", " ")
   if not text: 
     text = "this is blank"
-  return openai.Embedding.create(
-          input=[text], model=model)['data'][0]['embedding']
+  return ollama.embeddings(model=model,prompt=text)
 
 
 if __name__ == '__main__':

From 434a4721b51a27cb1f886e9324d6f9e24ddc0670 Mon Sep 17 00:00:00 2001
From: Hadrien <ketsapiwiq@protonmail.com>
Date: Mon, 20 May 2024 15:59:22 +0200
Subject: [PATCH 3/4] use phi3 for small model

---
 .../persona/prompt_template/run_gpt_prompt.py | 102 +++++++++---------
 1 file changed, 52 insertions(+), 50 deletions(-)

diff --git a/reverie/backend_server/persona/prompt_template/run_gpt_prompt.py b/reverie/backend_server/persona/prompt_template/run_gpt_prompt.py
index b25615e707..16ce0cc02f 100644
--- a/reverie/backend_server/persona/prompt_template/run_gpt_prompt.py
+++ b/reverie/backend_server/persona/prompt_template/run_gpt_prompt.py
@@ -16,6 +16,8 @@
 from persona.prompt_template.gpt_structure import *
 from persona.prompt_template.print_prompt import *
 
+model = "phi3:latest"
+
 def get_random_alphanumeric(i=6, j=6): 
   """
   Returns a random alpha numeric strength that has the length of somewhere
@@ -66,7 +68,7 @@ def get_fail_safe():
     fs = 8
     return fs
 
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 5, 
+  gpt_param = {"engine": model, "max_tokens": 5, 
              "temperature": 0.8, "top_p": 1, "stream": False,
              "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   prompt_template = "persona/prompt_template/v2/wake_up_hour_v1.txt"
@@ -138,7 +140,7 @@ def get_fail_safe():
 
 
   
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 500, 
+  gpt_param = {"engine": model, "max_tokens": 500, 
                "temperature": 1, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/daily_planning_v6.txt"
@@ -244,7 +246,7 @@ def get_fail_safe():
   #   return True
 
   # print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 10") ########
-  # gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  # gpt_param = {"engine": model, "max_tokens": 15, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v3_ChatGPT/generate_hourly_schedule_v2.txt" ########
@@ -265,7 +267,7 @@ def get_fail_safe():
   # # ChatGPT Plugin ===========================================================
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0.5, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   prompt_template = "persona/prompt_template/v2/generate_hourly_schedule_v2.txt"
@@ -426,7 +428,7 @@ def get_fail_safe():
     fs = ["asleep"]
     return fs
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 1000, 
+  gpt_param = {"engine": model, "max_tokens": 1000, 
              "temperature": 0, "top_p": 1, "stream": False,
              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/task_decomp_v3.txt"
@@ -583,7 +585,7 @@ def get_fail_safe():
   #   return True 
 
   # print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 20") ########
-  # gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  # gpt_param = {"engine": model, "max_tokens": 15, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v3_ChatGPT/action_location_sector_v2.txt" ########
@@ -602,7 +604,7 @@ def get_fail_safe():
 
 
 
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v1/action_location_sector_v1.txt"
@@ -699,7 +701,7 @@ def get_fail_safe():
     fs = ("kitchen")
     return fs
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v1/action_location_object_vMar11.txt"
@@ -755,7 +757,7 @@ def get_fail_safe():
     fs = ("bed")
     return fs
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v1/action_object_v2.txt"
@@ -825,7 +827,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
     return True
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 4") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/generate_pronunciatio_v1.txt" ########
@@ -844,7 +846,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 15, 
+  # gpt_param = {"engine": model, "max_tokens": 15, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   # prompt_template = "persona/prompt_template/v2/generate_pronunciatio_v1.txt"
@@ -915,7 +917,7 @@ def get_fail_safe(persona):
   #   return True 
 
   # print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 5") ########
-  # gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  # gpt_param = {"engine": model, "max_tokens": 15, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v3_ChatGPT/generate_event_triple_v1.txt" ########
@@ -933,7 +935,7 @@ def get_fail_safe(persona):
 
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 30, 
+  gpt_param = {"engine": model, "max_tokens": 30, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   prompt_template = "persona/prompt_template/v2/generate_event_triple_v1.txt"
@@ -1001,7 +1003,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
     return True 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 6") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/generate_obj_event_v1.txt" ########
@@ -1018,7 +1020,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 30, 
+  # gpt_param = {"engine": model, "max_tokens": 30, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   # prompt_template = "persona/prompt_template/v2/generate_obj_event_v1.txt"
@@ -1066,7 +1068,7 @@ def get_fail_safe(act_game_object):
     fs = (act_game_object, "is", "idle")
     return fs
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 30, 
+  gpt_param = {"engine": model, "max_tokens": 30, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": ["\n"]}
   prompt_template = "persona/prompt_template/v2/generate_event_triple_v1.txt"
@@ -1206,7 +1208,7 @@ def get_fail_safe(main_act_dur, truncated_act_dur):
 
     return ret
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 1000, 
+  gpt_param = {"engine": model, "max_tokens": 1000, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/new_decomp_schedule_v1.txt"
@@ -1320,7 +1322,7 @@ def get_fail_safe():
 
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 20, 
+  gpt_param = {"engine": model, "max_tokens": 20, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/decide_to_talk_v2.txt"
@@ -1418,7 +1420,7 @@ def get_fail_safe():
     return fs
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 20, 
+  gpt_param = {"engine": model, "max_tokens": 20, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/decide_to_react_v1.txt"
@@ -1561,7 +1563,7 @@ def get_fail_safe(init_persona, target_persona):
     return convo
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 1000, 
+  gpt_param = {"engine": model, "max_tokens": 1000, 
                "temperature": 0.7, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/create_conversation_v2.txt"
@@ -1626,7 +1628,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 11") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/summarize_conversation_v1.txt" ########
@@ -1642,7 +1644,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
   # ChatGPT Plugin ===========================================================
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  # gpt_param = {"engine": model, "max_tokens": 50, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/summarize_conversation_v1.txt"
@@ -1696,7 +1698,7 @@ def __func_validate(gpt_response, prompt=""):
   def get_fail_safe(): 
     return []
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/get_keywords_v1.txt"
@@ -1741,7 +1743,7 @@ def __func_validate(gpt_response, prompt=""):
   def get_fail_safe(): 
     return ""
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 40, 
+  gpt_param = {"engine": model, "max_tokens": 40, 
                "temperature": 0.7, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/keyword_to_thoughts_v1.txt"
@@ -1796,7 +1798,7 @@ def __func_validate(gpt_response, prompt=""):
   def get_fail_safe(): 
     return ""
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 40, 
+  gpt_param = {"engine": model, "max_tokens": 40, 
                "temperature": 0.7, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/convo_to_thoughts_v1.txt"
@@ -1879,7 +1881,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 7") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/poignancy_event_v1.txt" ########
@@ -1897,7 +1899,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 3, 
+  # gpt_param = {"engine": model, "max_tokens": 3, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/poignancy_event_v1.txt"
@@ -1950,7 +1952,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 8") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/poignancy_thought_v1.txt" ########
@@ -1967,7 +1969,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 3, 
+  # gpt_param = {"engine": model, "max_tokens": 3, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/poignancy_thought_v1.txt"
@@ -2022,7 +2024,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 9") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/poignancy_chat_v1.txt" ########
@@ -2040,7 +2042,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 3, 
+  # gpt_param = {"engine": model, "max_tokens": 3, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/poignancy_chat_v1.txt"
@@ -2098,7 +2100,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 12") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/generate_focal_pt_v1.txt" ########
@@ -2118,7 +2120,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 150, 
+  gpt_param = {"engine": model, "max_tokens": 150, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/generate_focal_pt_v1.txt"
@@ -2169,7 +2171,7 @@ def get_fail_safe(n):
 
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 150, 
+  gpt_param = {"engine": model, "max_tokens": 150, 
                "temperature": 0.5, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/insight_and_evidence_v1.txt"
@@ -2225,7 +2227,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 17") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/summarize_chat_ideas_v1.txt" ########
@@ -2242,7 +2244,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 150, 
+  # gpt_param = {"engine": model, "max_tokens": 150, 
   #              "temperature": 0.5, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/summarize_chat_ideas_v1.txt"
@@ -2293,7 +2295,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 18") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/summarize_chat_relationship_v2.txt" ########
@@ -2309,7 +2311,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
   # ChatGPT Plugin ===========================================================
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 150, 
+  # gpt_param = {"engine": model, "max_tokens": 150, 
   #              "temperature": 0.5, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/summarize_chat_relationship_v1.txt"
@@ -2421,7 +2423,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
   # print ("HERE JULY 23 -- ----- ") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/agent_chat_v1.txt" ########
@@ -2442,7 +2444,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 2000, 
+  # gpt_param = {"engine": model, "max_tokens": 2000, 
   #              "temperature": 0.7, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/agent_chat_v1.txt"
@@ -2502,7 +2504,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
       return False 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 16") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/summarize_ideas_v1.txt" ########
@@ -2518,7 +2520,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
   # ChatGPT Plugin ===========================================================
 
 
-  # gpt_param = {"engine": "text-davinci-003", "max_tokens": 150, 
+  # gpt_param = {"engine": model, "max_tokens": 150, 
   #              "temperature": 0.5, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v2/summarize_ideas_v1.txt"
@@ -2576,7 +2578,7 @@ def get_fail_safe():
   #     return False 
 
   # print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 15") ########
-  # gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  # gpt_param = {"engine": model, "max_tokens": 15, 
   #              "temperature": 0, "top_p": 1, "stream": False,
   #              "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   # prompt_template = "persona/prompt_template/v3_ChatGPT/generate_next_convo_line_v1.txt" ########
@@ -2593,7 +2595,7 @@ def get_fail_safe():
 
 
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 250, 
+  gpt_param = {"engine": model, "max_tokens": 250, 
                "temperature": 1, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/generate_next_convo_line_v1.txt"
@@ -2633,7 +2635,7 @@ def __func_validate(gpt_response, prompt=""):
   def get_fail_safe(): 
     return "..."
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/whisper_inner_thought_v1.txt"
@@ -2670,7 +2672,7 @@ def __func_validate(gpt_response, prompt=""):
   def get_fail_safe(): 
     return "..."
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/planning_thought_on_convo_v1.txt"
@@ -2721,7 +2723,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
 
 
   print ("asdhfapsh8p9hfaiafdsi;ldfj as DEBUG 15") ########
-  gpt_param = {"engine": "text-davinci-002", "max_tokens": 15, 
+  gpt_param = {"engine": model, "max_tokens": 15, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v3_ChatGPT/memo_on_convo_v1.txt" ########
@@ -2736,7 +2738,7 @@ def __chat_func_validate(gpt_response, prompt=""): ############
     return output, [output, prompt, gpt_param, prompt_input, fail_safe]
   # ChatGPT Plugin ===========================================================
 
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   prompt_template = "persona/prompt_template/v2/memo_on_convo_v1.txt"
@@ -2790,7 +2792,7 @@ def get_fail_safe():
                         __chat_func_validate, __chat_func_clean_up, verbose)
   print (output)
   
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   return output, [output, prompt, gpt_param, prompt_input, fail_safe]
@@ -2905,7 +2907,7 @@ def get_fail_safe():
                         __chat_func_validate, __chat_func_clean_up, verbose)
   print (output)
   
-  gpt_param = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_param = {"engine": model, "max_tokens": 50, 
                "temperature": 0, "top_p": 1, "stream": False,
                "frequency_penalty": 0, "presence_penalty": 0, "stop": None}
   return output, [output, prompt, gpt_param, prompt_input, fail_safe]

From d421a5af568ad5abcbc325ef6f4b1e90d512f717 Mon Sep 17 00:00:00 2001
From: Hadrien <ketsapiwiq@protonmail.com>
Date: Mon, 20 May 2024 16:00:03 +0200
Subject: [PATCH 4/4] fix llama prompt + add longer llama model

---
 .../persona/prompt_template/gpt_structure.py  | 20 ++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/reverie/backend_server/persona/prompt_template/gpt_structure.py b/reverie/backend_server/persona/prompt_template/gpt_structure.py
index 8e48fd6b63..27c9e0496c 100644
--- a/reverie/backend_server/persona/prompt_template/gpt_structure.py
+++ b/reverie/backend_server/persona/prompt_template/gpt_structure.py
@@ -21,7 +21,8 @@ def ChatGPT_single_request(prompt):
   temp_sleep()
 
   completion = openai.ChatCompletion.create(
-    model="llama3:8b-instruct-q8_0", 
+    model="llama3-gradient:8b-instruct-1048k-q8_0",
+    # model="llama3:8b-instruct-q8_0", 
     messages=[{"role": "user", "content": prompt}]
   )
   return completion["choices"][0]["message"]["content"]
@@ -47,7 +48,7 @@ def GPT4_request(prompt):
 
   try: 
     completion = openai.ChatCompletion.create(
-    model="llama3:8b-instruct-q8_0", 
+    model="llama3-gradient:8b-instruct-1048k-q8_0",
     messages=[{"role": "user", "content": prompt}]
     )
     return completion["choices"][0]["message"]["content"]
@@ -72,7 +73,7 @@ def ChatGPT_request(prompt):
   # temp_sleep()
   try: 
     completion = openai.ChatCompletion.create(
-    model="gpt-3.5-turbo", 
+    model="llama3-gradient:8b-instruct-1048k-q8_0",
     messages=[{"role": "user", "content": prompt}]
     )
     return completion["choices"][0]["message"]["content"]
@@ -209,9 +210,9 @@ def GPT_request(prompt, gpt_parameter):
   """
   temp_sleep()
   try: 
-    response = openai.Completion.create(
+    response = openai.ChatCompletion.create(
                 model=gpt_parameter["engine"],
-                prompt=prompt,
+                messages=[{"role": "system", "content": prompt}],
                 temperature=gpt_parameter["temperature"],
                 max_tokens=gpt_parameter["max_tokens"],
                 top_p=gpt_parameter["top_p"],
@@ -220,9 +221,10 @@ def GPT_request(prompt, gpt_parameter):
                 stream=gpt_parameter["stream"],
                 stop=gpt_parameter["stop"],)
     return response.choices[0].text
-  except: 
-    print ("TOKEN LIMIT EXCEEDED")
-    return "TOKEN LIMIT EXCEEDED"
+  except Exception as e: 
+    # print ("TOKEN LIMIT EXCEEDED")
+    print(str(e))
+    return "NO GENERATION"
 
 
 def generate_prompt(curr_input, prompt_lib_file): 
@@ -283,7 +285,7 @@ def get_embedding(text, model="nomic-embed-text"):
 
 
 if __name__ == '__main__':
-  gpt_parameter = {"engine": "text-davinci-003", "max_tokens": 50, 
+  gpt_parameter = {"engine": model, "max_tokens": 50, 
                    "temperature": 0, "top_p": 1, "stream": False,
                    "frequency_penalty": 0, "presence_penalty": 0, 
                    "stop": ['"']}