Merge pull request #94 from fractalego/prompt-as-list

Prompt as list

README.md

@@ -1,4 +1,4 @@
-# WAFL 0.0.80 [![Tests](https://github.com/fractalego/wafl/actions/workflows/development-tests1.yml/badge.svg)](https://github.com/fractalego/wafl/actions/workflows/development-tests1.yml)[![Docs](https://readthedocs.org/projects/wafl/badge/?version=latest)](https://wafl.readthedocs.io/en/latest/)
+# WAFL 0.0.90 [![Tests](https://github.com/fractalego/wafl/actions/workflows/development-tests1.yml/badge.svg)](https://github.com/fractalego/wafl/actions/workflows/development-tests1.yml)[![Docs](https://readthedocs.org/projects/wafl/badge/?version=latest)](https://wafl.readthedocs.io/en/latest/)
 
 Introduction
 ============
@@ -56,13 +56,6 @@ wafl-llm start
 ```
 which will use the default models and start the server on port 8080.
 
-#### Docker
-A docker image can be used to run it as in the following:
-
-```bash
-$ docker run -p8080:8080 --env NVIDIA_DISABLE_REQUIRE=1 --gpus all fractalego/wafl-llm:0.80
-```
-
 The interface side has a `config.json` file that needs to be filled with the IP address of the LLM side.
 The default is localhost.
 Alternatively, you can run the LLM side by cloning [this repository](https://github.com/fractalego/wafl-llm).
@@ -73,6 +66,11 @@ Running WAFL
 This document contains a few examples of how to use the `wafl` CLI.
 There are four modes in which to run the system
 
+
+### $ wafl run
+Starts all the available interfaces of the chatbot at the same time.
+
+
 ## $ wafl run-audio
 
 This is the main mode of operation. It will run the system in a loop, waiting for the user to speak a command.
@@ -82,7 +80,7 @@ The default name is "computer", but you can change it to whatever you want.
 
 ## $ wafl run-server
 
-It runs a local web server that listens for HTTP requests on port 8889.
+It runs a local web server that listens for HTTP requests on port 8090.
 The server will act as a chatbot, executing commands and returning the result as defined in the rules.
 
 

documentation/source/configuration.rst

@@ -49,8 +49,7 @@ These settings regulate the following:
 
     * "frontend_port" is the port where the web frontend is running. The default is 8090.
 
-    * "llm_model" is the configuration to connect to the LLM model in the backend. The default is "localhost:8080".
-       The "temperature" parameter is used to set the temperature for the LLM model. The default is 0.4.
+    * "llm_model" is the configuration to connect to wafl-llm in the backend. The default url is "localhost:8080". The "temperature" parameter is used to set the temperature for the LLM model. The default is 0.4.
 
     * "listener_model" is the configuration to connect to the listener model in the backend. The default is "localhost:8080".
 

documentation/source/index.rst

@@ -3,7 +3,7 @@
    You can adapt this file completely to your liking, but it should at least
    contain the root `toctree` directive.
 
-Welcome to WAFL's 0.0.80 documentation!
+Welcome to WAFL's 0.0.90 documentation!
 =======================================
 
 .. toctree::

documentation/source/installation.rst

@@ -24,9 +24,10 @@ After installing the requirements, you can initialize the interface by running t
 .. code-block:: bash
 
     $ wafl init
+    $ wafl run
 
-which creates a `config.json` file that you can edit to change the default settings.
-A standard rule file is also created as `wafl.rules`.
+The first command creates a set of template files, including a configuration in `config.json` that you can edit to change the default settings.
+The second command starts the audio interface as well as a web server on port 8090 by default.
 Please see the examples in the following chapters.
 
 
@@ -42,13 +43,6 @@ In order to quickly run the LLM side, you can use the following installation com
 
     which will use the default models and start the server on port 8080.
 
-Alternatively, a Docker image can be used to run it as in the following:
-
-.. code-block:: bash
-
-    $ docker run -p8080:8080 --env NVIDIA_DISABLE_REQUIRE=1 --gpus all fractalego/wafl-llm:0.80
-
-
 The interface side has a `config.json` file that needs to be filled with the IP address of the LLM side.
 The default is localhost.
 

documentation/source/running_WAFL.rst

@@ -3,6 +3,12 @@ Running WAFL
 This document contains a few examples of how to use the `wafl` CLI.
 There are four modes in which to run the system
 
+$ wafl run
+----------
+Starts the available interfaces of the chatbot at the same time.
+This is equivalent to running run-audio and run-server in parallel (see below).
+
+
 $ wafl run-audio
 ----------------
 
@@ -14,7 +20,7 @@ The default name is "computer", but you can change it to whatever you want.
 $ wafl run-server
 -----------------
 
-It runs a local web server that listens for HTTP requests on port 8889.
+It runs a local web server that listens for HTTP requests on port 8090.
 The server will act as a chatbot, executing commands and returning the result as defined in the rules.
 
 

license.txt

@@ -1,4 +1,4 @@
-Copyright (c) 2023 [email protected]
+Copyright (c) 2024 [email protected]
 
 Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
 

tests/config.json

@@ -7,7 +7,8 @@
   "max_recursion": 2,
   "llm_model": {
     "model_host": "localhost",
-    "model_port": 8080
+    "model_port": 8080,
+    "temperature": 0.4
   },
   "listener_model": {
     "model_host": "localhost",

tests/test_closing_conversation.py

@@ -30,7 +30,7 @@ def test__thank_you_closes_conversation(self):
         )
         try:
             asyncio.run(conversation_events.process_next())
-
+            print(interface.get_utterances_list())
         except CloseConversation:
             self.assertTrue(True)
             return

tests/test_connection.py

@@ -3,9 +3,8 @@
 
 from unittest import TestCase
 from wafl.config import Configuration
-from wafl.connectors.bridges.llm_chitchat_answer_bridge import LLMChitChatAnswerBridge
+from wafl.connectors.prompt_template import PromptCreator
 from wafl.connectors.remote.remote_llm_connector import RemoteLLMConnector
-from wafl.speaker.fairseq_speaker import FairSeqSpeaker
 
 _path = os.path.dirname(__file__)
 
@@ -16,8 +15,10 @@ def test__connection_to_generative_model_can_generate_text(self):
         connector = RemoteLLMConnector(config.get_value("llm_model"))
         prediction = asyncio.run(
             connector.predict(
-                'Generate a full paragraph based on this chapter title "The first contact". '
-                "The theme of the paragraph is space opera. "
+                PromptCreator.create_from_one_instruction(
+                    'Generate a full paragraph based on this chapter title "The first contact".'
+                    "The theme of the paragraph is space opera. "
+                )
             )
         )
         assert len(prediction) > 0
@@ -34,7 +35,9 @@ def test__connection_to_generative_model_can_generate_text_within_tags(self):
 <result>
                 """.strip()
 
-        prediction = asyncio.run(connector.predict(prompt))
+        prediction = asyncio.run(
+            connector.predict(PromptCreator.create_from_one_instruction(prompt))
+        )
         print(prediction)
         assert len(prediction) > 0
 
@@ -43,6 +46,7 @@ def test__connection_to_generative_model_can_generate_a_python_list(self):
         connector = RemoteLLMConnector(config.get_value("llm_model"))
         connector._num_prediction_tokens = 200
         prompt = "Generate a Python list of 4 chapters names for a space opera book. The output needs to be a python list of strings: "
-        prediction = asyncio.run(connector.predict(prompt))
-        print(prediction)
+        prediction = asyncio.run(
+            connector.predict(PromptCreator.create_from_one_instruction(prompt))
+        )
         assert len(prediction) > 0

tests/test_prompts.py

@@ -0,0 +1,62 @@
+import os
+from unittest import TestCase
+
+from wafl.connectors.prompt_template import PromptCreator
+from wafl.interface.conversation import Conversation, Utterance
+
+_path = os.path.dirname(__file__)
+
+
+class TestPrompts(TestCase):
+    def test_utterance(self):
+        utterance = Utterance(
+            text="Hello", speaker="user", timestamp="2022-01-01T00:00:00"
+        )
+        self.assertEqual(
+            utterance.to_dict(),
+            {"text": "Hello", "speaker": "user", "timestamp": "2022-01-01T00:00:00"},
+        )
+
+    def test_conversation(self):
+        utterance1 = Utterance(text="Hello", speaker="user", timestamp=2)
+        utterance2 = Utterance(text="Hi", speaker="bot", timestamp=1)
+        conversation = Conversation(utterances=[utterance1, utterance2])
+        self.assertEqual(
+            conversation.to_dict(),
+            [
+                {
+                    "text": "Hello",
+                    "speaker": "user",
+                    "timestamp": 2,
+                },
+                {
+                    "text": "Hi",
+                    "speaker": "bot",
+                    "timestamp": 1,
+                },
+            ],
+        )
+
+    def test_prompt(self):
+        utterance1 = Utterance(text="Hello", speaker="user", timestamp=2)
+        utterance2 = Utterance(text="Hi", speaker="bot", timestamp=1)
+        conversation = Conversation(utterances=[utterance1, utterance2])
+        prompt = PromptCreator.create(system_prompt="Hello", conversation=conversation)
+        self.assertEqual(
+            prompt.to_dict(),
+            {
+                "system_prompt": "Hello",
+                "conversation": [
+                    {
+                        "text": "Hello",
+                        "speaker": "user",
+                        "timestamp": 2,
+                    },
+                    {
+                        "text": "Hi",
+                        "speaker": "bot",
+                        "timestamp": 1,
+                    },
+                ],
+            },
+        )

tests/test_rules.py

@@ -39,8 +39,8 @@ def test__rules_can_be_triggered(self):
             interface=interface,
         )
         asyncio.run(conversation_events.process_next())
-        expected = "bot: the horse is tall"
-        self.assertEqual(expected, interface.get_utterances_list()[-1])
+        expected = "The horse is tall"
+        self.assertIn(expected, interface.get_utterances_list()[-1])
 
     def test__rules_are_not_always_triggered(self):
         interface = DummyInterface(

tests/test_voice.py

@@ -5,7 +5,6 @@
 
 from unittest import TestCase
 from wafl.config import Configuration
-from wafl.interface.voice_interface import VoiceInterface
 from wafl.events.conversation_events import ConversationEvents
 from wafl.interface.dummy_interface import DummyInterface
 from wafl.listener.whisper_listener import WhisperListener
@@ -31,7 +30,8 @@ def test__activation(self):
         interface.activate()
         asyncio.run(conversation_events.process_next(activation_word="computer"))
         asyncio.run(conversation_events.process_next(activation_word="computer"))
-        assert interface.get_utterances_list()[-1] == "bot: I hear you"
+        print(interface.get_utterances_list())
+        assert "I hear you" in interface.get_utterances_list()[-1]
 
     def test__no_activation(self):
         interface = DummyInterface(to_utter=["my name is bob"])

todo.txt

@@ -1,3 +1,5 @@
+* substitute utterances in base_interface with the conversation class
+
 * add config file for model names
     - llm model name
     - whisper model name
@@ -6,9 +8,9 @@
 /* let user decide port for frontend
 /* update docs about port
 /* push new version
-* update pypi with wafl and wafl-llm
-* clean code for llm eval and make it public
-* update huggingface readme
+/* update pypi with wafl and wafl-llm
+/* clean code for llm eval and make it public
+/* update huggingface readme
 * read overleaf paper