Merge pull request #22 from oddiz/gpt-3.5-turbo-16k

feat: added model gpt-3.5-turbo-16k
sean1832 · Oct 29, 2023 · 5590572 · 5590572
2 parents 31d08eb + 2fd1124
commit 5590572
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 3 deletions.
diff --git a/src/Components/sidebar.py b/src/Components/sidebar.py
@@ -110,12 +110,14 @@ def sidebar():
                 persona_sum = ""
 
         with st.expander('🔥 Advanced Options'):
-            model_options = ['gpt-3.5-turbo', 'gpt-4']
+            model_options = ['gpt-3.5-turbo','gpt-3.5-turbo-16k', 'gpt-4']
             model_index = model_options.index(_set_config(config_file, "MODEL", 'gpt-3.5-turbo'))
             model = st.selectbox("Model", options=model_options, index=model_index)
 
             if model == 'gpt-4':
                 max_chunk = 4000
+            elif model == 'gpt-3.5-turbo-16k':
+                max_chunk = 16000
             else:
                 max_chunk = 2500
             chunk_size = st.slider('Chunk Size (word count)', min_value=0, max_value=max_chunk, step=20,
@@ -137,7 +139,7 @@ def sidebar():
                                          value=_set_config(config_file, "PRESENCE_PENALTY", 0.0))
             if st_toggle_switch(label="Delay (free openAI API user)",
                                 default_value=_set_config(config_file, "ENABLE_DELAY", False)):
-                delay = st.slider('Delay (seconds)', min_value=0, max_value=5, step=1,
+                delay = st.slider('Delay (seconds)', min_value=0, max_value=60, step=1,
                                   value=_set_config(config_file, "DELAY_TIME", 1))
             else:
                 delay = 0

diff --git a/src/SumGPT.py b/src/SumGPT.py
@@ -80,8 +80,10 @@
             st.markdown('**Note:** To access GPT-4, please [join the waitlist](https://openai.com/waitlist/gpt-4-api)'
                         " if you haven't already received an invitation from OpenAI.")
             st.info("ℹ️️ Please keep in mind that GPT-4 is significantly **[more expensive](https://openai.com/pricing#language-models)** than GPT-3.5. ")
+        elif param.model == 'gpt-3.5-turbo-16k':
+            price = round(prompt_token * 0.000003 + completion_token *0.000004, 5)
         else:
-            price = round((prompt_token + completion_token) * 0.000002, 5)
+            price = round(prompt_token * 0.0000015 + completion_token * 0.000002 , 5)
         st.markdown(
             f"Price Prediction: `${price}` || Total Prompt: `{prompt_token}`, Total Completion: `{completion_token}`")
         # max tokens exceeded warning

diff --git a/src/util.py b/src/util.py
@@ -225,6 +225,8 @@ def exceeded_token_handler(param, chunks) -> bool:
     """Handles the case where the user has exceeded the number of tokens."""
     if param.model == 'gpt-4':
         max_token = 8100
+    elif param.model == 'gpt-3.5-turbo-16k':
+        max_token = 16385
     else:
         max_token = 4096
     info = GPT.misc.is_tokens_exceeded(param, chunks, max_token)