Merge pull request #498 from TransformerOptimus/gpt4-32k

Gpt4 32k

config_template.yaml

@@ -9,17 +9,11 @@ OPENAI_API_KEY: YOUR_OPEN_API_KEY
 OPENAI_API_BASE: https://api.openai.com/v1
 #OPENAI_API_BASE: "http://super__tgwui:5001/v1"
 
-# "gpt-3.5-turbo-0301": 4032, "gpt-4-0314": 8092, "gpt-3.5-turbo": 4032, "gpt-4": 8092, "llama":2048, "mpt-7b-storywriter":45000
+# "gpt-3.5-turbo-0301": 4032, "gpt-4-0314": 8092, "gpt-3.5-turbo": 4032, "gpt-4": 8092, "gpt-4-32k": 32768, "gpt-4-32k-0314": 32768, "llama":2048, "mpt-7b-storywriter":45000
 MODEL_NAME: "gpt-3.5-turbo-0301"
 MAX_TOOL_TOKEN_LIMIT: 800
 MAX_MODEL_TOKEN_LIMIT: 4032 # set to 2048 for llama
 
-# For running stable diffusion
-STABILITY_API_KEY: YOUR_STABILITY_API_KEY
-#Engine IDs that can be used: 'stable-diffusion-v1', 'stable-diffusion-v1-5','stable-diffusion-512-v2-0', 'stable-diffusion-768-v2-0','stable-diffusion-512-v2-1','stable-diffusion-768-v2-1','stable-diffusion-xl-beta-v2-2-2'
-ENGINE_ID: "stable-diffusion-xl-beta-v2-2-2"
-
-
 #DATABASE INFO
 # redis details
 DB_NAME: super_agi_main
@@ -56,7 +50,6 @@ FRONTEND_URL: "http://localhost:3000"
 #ENCRPYTION KEY
 ENCRYPTION_KEY: secret
 
-
 #WEAVIATE
 
 # If you are using docker or web hosted uncomment the next two lines and comment the third one
@@ -93,4 +86,9 @@ JIRA_USERNAME: YOUR_JIRA_EMAIL
 JIRA_API_TOKEN: YOUR_JIRA_API_TOKEN
 
 #SLACK
-SLACK_BOT_TOKEN: YOUR_SLACK_BOT_TOKEN
+SLACK_BOT_TOKEN: YOUR_SLACK_BOT_TOKEN
+
+# For running stable diffusion
+STABILITY_API_KEY: YOUR_STABILITY_API_KEY
+#Engine IDs that can be used: 'stable-diffusion-v1', 'stable-diffusion-v1-5','stable-diffusion-512-v2-0', 'stable-diffusion-768-v2-0','stable-diffusion-512-v2-1','stable-diffusion-768-v2-1','stable-diffusion-xl-beta-v2-2-2'
+ENGINE_ID: "stable-diffusion-xl-beta-v2-2-2"

gui/pages/Content/Agents/AgentCreate.js

@@ -35,7 +35,7 @@ export default function AgentCreate({sendAgentData, selectedProjectId, fetchAgen
   const [goals, setGoals] = useState(['Describe the agent goals here']);
   const [instructions, setInstructions] = useState(['']);
 
-  const models = ['gpt-4', 'gpt-3.5-turbo','gpt-3.5-turbo-16k']
+  const models = ['gpt-4', 'gpt-3.5-turbo','gpt-3.5-turbo-16k', 'gpt-4-32k']
   const [model, setModel] = useState(models[1]);
   const modelRef = useRef(null);
   const [modelDropdown, setModelDropdown] = useState(false);

superagi/helper/token_counter.py

@@ -21,7 +21,7 @@ def token_limit(model: str = "gpt-3.5-turbo-0301") -> int:
             int: The token limit.
         """
         try:
-            model_token_limit_dict = {"gpt-3.5-turbo-0301": 4032, "gpt-4-0314": 8092, "gpt-3.5-turbo": 4032, "gpt-4": 8092,"gpt-3.5-turbo-16k": 16184}
+            model_token_limit_dict = {"gpt-3.5-turbo-0301": 4032, "gpt-4-0314": 8092, "gpt-3.5-turbo": 4032, "gpt-4": 8092,"gpt-3.5-turbo-16k": 16184, "gpt-4-32k": 32768, "gpt-4-32k-0314": 32768}
             return model_token_limit_dict[model]
         except KeyError:
             logger.warning("Warning: model not found. Using cl100k_base encoding.")
@@ -43,7 +43,7 @@ def count_message_tokens(messages: List[BaseMessage], model: str = "gpt-3.5-turb
             int: The number of tokens in the messages.
         """
         try:
-            model_token_per_message_dict = {"gpt-3.5-turbo-0301": 4, "gpt-4-0314": 3, "gpt-3.5-turbo": 4, "gpt-4": 3,"gpt-3.5-turbo-16k":4}
+            model_token_per_message_dict = {"gpt-3.5-turbo-0301": 4, "gpt-4-0314": 3, "gpt-3.5-turbo": 4, "gpt-4": 3,"gpt-3.5-turbo-16k":4, "gpt-4-32k": 3, "gpt-4-32k-0314": 3}
             encoding = tiktoken.encoding_for_model(model)
         except KeyError:
             logger.warning("Warning: model not found. Using cl100k_base encoding.")

tests/unit_tests/helper/test_token_counter.py

@@ -0,0 +1,28 @@
+import pytest
+
+from typing import List
+from superagi.helper.token_counter import TokenCounter
+from superagi.types.common import BaseMessage
+from unittest.mock import MagicMock
+
+
+def test_token_limit():
+    assert TokenCounter.token_limit("gpt-3.5-turbo-0301") == 4032
+    assert TokenCounter.token_limit("gpt-4-0314") == 8092
+    assert TokenCounter.token_limit("gpt-3.5-turbo") == 4032
+    assert TokenCounter.token_limit("gpt-4") == 8092
+    assert TokenCounter.token_limit("gpt-3.5-turbo-16k") == 16184
+    assert TokenCounter.token_limit("gpt-4-32k") == 32768
+    assert TokenCounter.token_limit("gpt-4-32k-0314") == 32768
+    assert TokenCounter.token_limit("non_existing_model") == 8092
+
+
+def test_count_text_tokens():
+    # You might need to adjust the hardcoded values in the TokenCounter.count_text_tokens function
+    # and update the expected tokens accordingly if the function logic is changed.
+
+    text = "You are a helpful assistant."
+    assert TokenCounter.count_text_tokens(text) == 10
+
+    text = "What is your name?"
+    assert TokenCounter.count_text_tokens(text) == 9