Llm rxn fix (#52)

* notebook and LLM updates

* first merge llm

* llm fix and rxn fix

---------

Co-authored-by: Phil Downey <phildowney@pd-work-mac.local>

Author: PhillipDowney

README.md

@@ -15,7 +15,7 @@ urls = [
     'https://cps.foc-deepsearch.zurich.ibm.com',
     'https://rxn.app.accelerate.science',
     'https://sds.app.accelerate.science',
-    'https://platform.openai.com/account/api-keys'
+    
 ]
 for (var i=0; i< urls.length; i++) {
     window.open(urls[i], '_blank', 'width=1000,height=600');
@@ -369,15 +369,15 @@ To run a command in bash mode, prepend it with `openad` and make sure to escape
 
 To enable our AI assistant, you'll need either have access to [IBM BAM](https://bam.res.ibm.com/auth/signin) or to use a free open source LLM use [ollama](ollama.com).
 
-**Note:** Ollama will regire a 8gb GPU
+**Note:** Ollama will requires a 8gb GPU
 
 > **Note:** watsonx coming soon
 
 ## IBM BAM Setup
 For IBM BAM simply used your supplied API key if you have BAM access
 
 ### Run BAM LLM
-run `tell me` to be prompted for your OpenAI API credentials
+run `tell me` to be prompted for your BAM API credentials
 ```
 >> set llm bam
 >> tell me <enter prompt>

openad/app/main.py

@@ -57,7 +57,7 @@
 import inspect
 import importlib
 
-
+MAGIC_PROMPT = None
 PLUGIN_CLASS_LIST = []
 installed_packages = pkg_resources.working_set
 installed_packages_list = [
@@ -116,8 +116,8 @@ class RUNCMD(Cmd):
         False  # Signals the Refresh of the vector DB should be done due to changes in Workspace or Toolkits
     )
     refresh_train = False  # Signals Refreshing of the training repository for help should be done
-    llm_service = "OPENAI"  # set with OPENAI as default type until WatsonX or alternative available
-    llm_model = "gpt-3.5-turbo"
+    llm_service = "OLLAMA"  # set with OLLAMA as default type until WatsonX or alternative available
+    llm_model = "instructlab/granite-7b-lab"
     llm_models = SUPPORTED_TELL_ME_MODELS_SETTINGS
 
     # Load OpenAD Plugins into cmd_pointer
@@ -825,6 +825,8 @@ def error_first_word_grabber(error):
 # If the application is called with parameters, it executes the parameters.
 # If called without parameters, the command line enters the shell environment.
 # History is only kept for commands executed once in the shell.
+
+
 def api_remote(
     inp: str,
     api_context: dict = {"workspace": None, "toolkit": None},
@@ -842,16 +844,21 @@ def api_remote(
     - It is deliberate that the whole RUNCMD class object is not kept alive as there is no logical
       exit point for magic commands, unlike a command line.
     """
-
+    global MAGIC_PROMPT
     # GLOBAL_SETTINGS["display"] = "notebook"
+
     initialise()
 
     arguments = inp.split()
     inp = ""  # reset input after splitting into arguments
     a_space = ""  # reset a_space
 
     # setup for notebook mode
-    magic_prompt = RUNCMD()
+    if MAGIC_PROMPT is None:
+        magic_prompt = RUNCMD()
+        MAGIC_PROMPT = magic_prompt
+    else:
+        magic_prompt = MAGIC_PROMPT
 
     if api_context["workspace"] is None:
         api_context["workspace"] = magic_prompt.settings["workspace"]

openad/app/metadata.json

@@ -1,7 +1,7 @@
 {
 	"banner": "OPENAD",
 	"title": "Welcome to the Open Accelerated Discovery CLI",
-	"version": "0.3.2",
+	"version": "0.3.6",
 	"commands": {
 		"intro": "If this is your first time, let us help you getting started.",
 		"? toolkits": "Learn about the toolkits.",

openad/helpers/credentials.py

@@ -1,4 +1,5 @@
 """ Helpers for getting and Saving Credentials"""
+
 import pickle
 import readline
 import os
@@ -56,7 +57,3 @@ def write_credentials(registry: dict, location):
     with open(os.path.expanduser(location), "wb") as handle:
         pickle.dump(registry, handle)
     return True
-
-
-if __name__ == "__main__":
-    print(load_credentials("~/.openad/openai_api.json"))

openad/llm_assist/llm_interface.py

@@ -20,7 +20,8 @@
 
 # Constants
 TRAINING_LLM_DIR = "/prompt_train/"
-SUPPORTED_LLMS = ["WATSONX", "OPENAI", "BAM", "OLLAMA"]
+# SUPPORTED_LLMS = ["WATSONX", "OPENAI", "BAM", "OLLAMA"]
+SUPPORTED_LLMS = ["WATSONX", "BAM", "OLLAMA"]
 PROMPT_DIR = "~/.chat_embedded"
 STANDARD_FILE_TYPES_EMBED = ["*.txt", "*.ipynb", "*.run", "*.cdoc", "*.pdf"]
 EXTENDED_FILE_TYPES_EMBED = ["**/*.txt", "**/*.ipynb", "**/*.run", "**/*.cdoc", "**/*.pdf"]

openad/llm_assist/model_reference.py

@@ -52,8 +52,9 @@
 LOCAL_MODEL_KWARGS = {"device": LOCAL_EMBEDDINGS_DEVICE}
 LOCAL_ENCODE_KWARGS = {"normalize_embeddings": False}
 
-DEFAULT_TELL_ME_MODEL = "BAM"
-SUPPORTED_TELL_ME_MODELS = ["BAM", "OPENAI", "OLLAMA"]
+DEFAULT_TELL_ME_MODEL = "OLLAMA"
+# SUPPORTED_TELL_ME_MODELS = ["BAM", "OPENAI", "OLLAMA"]
+SUPPORTED_TELL_ME_MODELS = ["BAM", "OLLAMA"]
 OLLAMA_HOST = "http://0.0.0.0:11434"
 
 try:
@@ -83,7 +84,7 @@
 
 Answer the question based only on the following context: {context}  Question: {question} """,
         "settings": {
-            "temperature": 0.7,
+            "temperature": 0.5,
             "decoding_method": "greedy",
             "max_new_tokens": 2048,
             "min_new_tokens": 1,
@@ -95,7 +96,7 @@
     },
     "OLLAMA": {
         # "model": "granite-code:8b",
-        "model": "llama3",
+        "model": "instructlab/granite-7b-lab",
         "url": OLLAMA_HOST,
         "template": """You are a technical documentation writer and when responding follow the following rules:
                 - Format All Command Syntax, Clauses, Examples or Option Syntax in codeblock ipython Markdown
@@ -175,15 +176,15 @@ def get_tell_me_model(service: str, api_key: str):
 
             return model, SUPPORTED_TELL_ME_MODELS_SETTINGS[service]["template"]
         except Exception as e:  # pylint: disable=broad-exception-caught
-            output_error("Error Loading OPENAI Model see error Messsage : \n" + e, return_val=False)
+            output_error("Error Loading  Model see error Messsage : \n" + e, return_val=False)
             return None, None
     elif service == "OLLAMA":
         try:
             model = ChatOllama(model=SUPPORTED_TELL_ME_MODELS_SETTINGS[service]["model"], base_url=OLLAMA_HOST)
 
             return model, SUPPORTED_TELL_ME_MODELS_SETTINGS[service]["template"]
         except Exception as e:  # pylint: disable=broad-exception-caught
-            output_error("Error Loading OPENAI Model see error Messsage : \n" + e, return_val=False)
+            output_error("Error Loading  Model see error Messsage : \n" + e, return_val=False)
             return None, None
     elif service == "BAM":
         creds = Credentials(api_key=api_key, api_endpoint=SUPPORTED_TELL_ME_MODELS_SETTINGS[service]["url"])

openad/llm_assist/prime_chat.py

@@ -78,8 +78,8 @@ def __init__(
         document_types=document_types,
         db_dir_override=None,
         refresh_vector=False,
-        llm_model="gpt-4",
-        llm_service="OPENAI",
+        llm_model="nstructlab/granite-7b-lab",
+        llm_service="ollama",
     ):
         self.organisation = organisation
         self.target = target