Assistant Chat Completion

artitw · Sep 24, 2023 · 8ee1317 · 8ee1317
1 parent 749135e
commit 8ee1317
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 20 deletions.
diff --git a/README.md b/README.md
@@ -35,7 +35,7 @@ Transform texts in a hundred different [languages](https://github.com/artitw/tex
 </details>
 
 ## Colab Notebooks
-* Assistant (free private chatGPT alternative) [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1K6Kk80w9vjFZ7PL9dPRgVuOPuaWcY4ae?usp=sharing)
+* Assistant (free private ChatGPT LLM alternative) [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1K6Kk80w9vjFZ7PL9dPRgVuOPuaWcY4ae?usp=sharing)
 * Assistant with knowledge base [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1hkNgpSmmUA-mzUibqz25xq-E8KYOLuVx?usp=sharing)
 * STF-IDF multilingual search [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1RaWj5SqWvyC2SsCTGg8IAVcl9G5hOB50?usp=sharing)
 * All examples [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1LE_ifTpOGO5QJCKNQYtZe6c_tjbwnulR)
@@ -201,19 +201,16 @@ res = asst.transform([instructions])
 
 #OpenAI API format
 input_prompts = ["Hello, world!"]
-print(
-    asst.completion_tokens(input_prompts),
-    asst.completion(input_prompts)
-)
-#[13]
-#['Hello there! How can I help you today? If you have any questions or need assistance with something, feel free to ask.']
+asst.completion_tokens(input_prompts) #[13]
+asst.completion(input_prompts) #['Hello there! How can I help you today? If you have any questions or need assistance with something, feel free to ask.']
 
-results = asst.chat_completion([
+chat_history = [
     {"role": "user",  "content": "Hi"},
     {"role": "assistant", "content": "Hello, how are you?"},
     {"role": "user", "content": "What should I do today?"}
-])
-#{'role': 'assistant', 'content': '1. Make a list of things to be grateful for.\n2. Go outside and take a walk in nature.\n3. Practice mindfulness meditation.\n4. Connect with a loved one or friend.\n5. Do something kind for someone else.\n6. Engage in a creative activity like drawing or writing.\n7. Read an uplifting book or listen to motivational podcasts.'}
+]
+asst.chat_completion_tokens(chat_history) #31
+asst.chat_completion(chat_history) #{'role': 'assistant', 'content': '1. Make a list of things to be grateful for.\n2. Go outside and take a walk in nature.\n3. Practice mindfulness meditation.\n4. Connect with a loved one or friend.\n5. Do something kind for someone else.\n6. Engage in a creative activity like drawing or writing.\n7. Read an uplifting book or listen to motivational podcasts.'}
 ```
 - To use a dynamic knowledge base, see [![Q&A Assistant Demo](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1hkNgpSmmUA-mzUibqz25xq-E8KYOLuVx?usp=sharing)
 - To use with LangChain, see [![LangChain integration](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1K6Kk80w9vjFZ7PL9dPRgVuOPuaWcY4ae?usp=sharing)

diff --git a/demos/Text2Text<>LangChain.ipynb → demos/Text2Text_LLM.ipynb b/demos/Text2Text<>LangChain.ipynb → demos/Text2Text_LLM.ipynb
@@ -18,7 +18,7 @@
   "cells": [
     {
       "cell_type": "code",
-      "execution_count": 3,
+      "execution_count": null,
       "metadata": {
         "colab": {
           "base_uri": "https://localhost:8080/"
@@ -40,6 +40,31 @@
         "pip install -qq -U text2text"
       ]
     },
+    {
+      "cell_type": "code",
+      "source": [
+        "import text2text as t2t\n",
+        "asst = t2t.Assistant()\n",
+        "\n",
+        "#OpenAI API format\n",
+        "input_prompts = [\"Hello, world!\"]\n",
+        "asst.completion_tokens(input_prompts) #[13]\n",
+        "asst.completion(input_prompts) #['Hello there! How can I help you today? If you have any questions or need assistance with something, feel free to ask.']\n",
+        "\n",
+        "chat_history = [\n",
+        "    {\"role\": \"user\",  \"content\": \"Hi\"},\n",
+        "    {\"role\": \"assistant\", \"content\": \"Hello, how are you?\"},\n",
+        "    {\"role\": \"user\", \"content\": \"What should I do today?\"}\n",
+        "]\n",
+        "asst.chat_completion_tokens(chat_history) #31\n",
+        "asst.chat_completion(chat_history) #{'role': 'assistant', 'content': '1. Make a list of things to be grateful for.\\n2. Go outside and take a walk in nature.\\n3. Practice mindfulness meditation.\\n4. Connect with a loved one or friend.\\n5. Do something kind for someone else.\\n6. Engage in a creative activity like drawing or writing.\\n7. Read an uplifting book or listen to motivational podcasts.'}"
+      ],
+      "metadata": {
+        "id": "UqprqtKi33IE"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
     {
       "cell_type": "code",
       "source": [
@@ -60,7 +85,7 @@
         "id": "a8aELPRAo4HP",
         "outputId": "5296cfe9-2b9e-4929-aaa0-ce5a7d5b2420"
       },
-      "execution_count": 6,
+      "execution_count": null,
       "outputs": [
         {
           "output_type": "stream",

diff --git a/setup.py b/setup.py
@@ -5,7 +5,7 @@
 
 setuptools.setup(
   name="text2text",
-  version="1.3.3",
+  version="1.3.4",
   author="artitw",
   author_email="artitw@gmail.com",
   description="Text2Text: Crosslingual NLP/G toolkit",

diff --git a/text2text/assistant.py b/text2text/assistant.py
@@ -24,7 +24,7 @@ def __init__(self, **kwargs):
       quantize_config=None
     )
 
-  def preprocess(self, input_lines, retriever=None, **kwargs):
+  def completion_preprocess(self, input_lines, retriever=None, **kwargs):
     df = pd.DataFrame({"input_line": input_lines})
     if retriever:
       k = kwargs.get('k', 1)
@@ -34,13 +34,13 @@ def preprocess(self, input_lines, retriever=None, **kwargs):
     return df
 
   def completion_tokens(self, input_lines):
-    df = self.preprocess(input_lines)
+    df = self.completion_preprocess(input_lines)
     tok = self.__class__.tokenizer
     input_ids = tok(df["input_line"].tolist(), return_tensors="pt", padding=True).input_ids
     return [len(x) for x in input_ids]
 
   def transform(self, input_lines, retriever=None, **kwargs):
-    df = self.preprocess(input_lines, retriever, **kwargs)
+    df = self.completion_preprocess(input_lines, retriever, **kwargs)
     temperature = kwargs.get('temperature', 0.7)
     top_p = kwargs.get('top_p', 0.95)
     top_k = kwargs.get('top_k', 0)
@@ -68,12 +68,20 @@ def transform(self, input_lines, retriever=None, **kwargs):
 
   completion = transform
 
-  def chat_completion(self, input_lines, **kwargs):
-    chat_history = []
-    for line in input_lines:
-      chat_history.append(f'{line["role"].upper()}: {line["content"]}')
+  def chat_completion_preprocess(self, messages):
+    chat_history = [f'{line["role"].upper()}: {line["content"]}' for line in messages]
     chat_history.append("ASSISTANT: ")
     input_prompt = "\n".join(chat_history)
+    return input_prompt
+
+  def chat_completion_tokens(self, messages):
+    input_prompt = self.chat_completion_preprocess(messages)
+    tok = self.__class__.tokenizer
+    input_ids = tok([input_prompt], return_tensors="pt", padding=True).input_ids[0]
+    return len(input_ids)
+
+  def chat_completion(self, messages, **kwargs):
+    input_prompt = self.chat_completion_preprocess(messages)
 
     temperature = kwargs.get('temperature', 0.7)
     top_p = kwargs.get('top_p', 0.95)