feat: Dynamically scrape Ollama model names

.github/workflows/release-python-package.yaml

@@ -71,10 +71,13 @@ jobs:
       - name: Write release notes
         run: |
           python -m pip install -e .
-          python -m pip install pre-commit
           llamabot configure api-key --api-key="${{ secrets.OPENAI_API_KEY }}"
           llamabot configure default-model --model-name="${{ secrets.OPENAI_DEFAULT_MODEL }}"
           llamabot git write-release-notes
+
+      - name: Commit release notes
+        run: |
+          python -m pip install pre-commit
           pre-commit run --all-files || true
           git add .
           git commit -m "Add release notes for ${{ env.version_number }}"

.pre-commit-config.yaml

@@ -31,3 +31,11 @@ repos:
           - id: ruff
             args: [--fix, --exit-non-zero-on-fix]
             types_or: [python]
+    - repo: local
+      hooks:
+      - id: autoupdate-ollama-models
+        name: Autoupdate Ollama Models
+        entry: python scripts/autoupdate_ollama_models.py
+        language: python
+        additional_dependencies: ["beautifulsoup4", "lxml", "requests"]
+        pass_filenames: false

MANIFEST.in

@@ -1,3 +1,4 @@
 # Some references to help:
 # https://wiki.python.org/moin/Distutils/Tutorial
 # https://stackoverflow.com/a/24727824/1274908
+include llamabot/bot/ollama_model_names.txt

docs/releases/v0.0.86.md

@@ -27,4 +27,4 @@ This version includes several enhancements and updates to the codebase, includin
 
 ### Chore
 
-- Updated the versions of pre-commit hooks for pre-commit-hooks, black, and ruff-pre-commit. It also replaces the darglint hook with pydoclint for better documentation linting. (9cc49022) (Eric Ma)
+- Updated the versions of pre-commit hooks for pre-commit-hooks, black, and ruff-pre-commit. It also replaces the darglint hook with pydoclint for better documentation linting. (9cc49022) (Eric Ma)

llamabot/bot/model_dispatcher.py

@@ -13,41 +13,21 @@
 from time import sleep
 from loguru import logger
 from functools import partial
+from pathlib import Path
+from functools import lru_cache
 
-# get this list from: https://ollama.ai/library
-ollama_model_keywords = [
-    "mistral",
-    "llama2",
-    "codellama",
-    "vicuna",
-    "orca-mini",
-    "llama2-uncensored",
-    "wizard-vicuna-uncensored",
-    "nous-hermes",
-    "phind-codellama",
-    "mistral-openorca",
-    "wizardcoder",
-    "wizard-math",
-    "llama2-chinese",
-    "stable-beluga",
-    "codeup",
-    "everythinglm",
-    "medllama2",
-    "wizardlm-uncensored",
-    "zephyr",
-    "falcon",
-    "wizard-vicuna",
-    "open-orca-platypus2",
-    "starcoder",
-    "samantha-mistral",
-    "openhermes2-mistral",
-    "wizardlm",
-    "sqlcoder",
-    "dolphin2.1-mistral",
-    "nexusraven",
-    "dolphin2.2-mistral",
-    "codebooga",
-]
+
+@lru_cache(maxsize=128)
+def ollama_model_keywords() -> list:
+    """Return ollama model keywords.
+
+    This is stored within a the `ollama_model_names.txt` file
+    that is distributed with this package.
+
+    :returns: The list of model names.
+    """
+    with open(Path(__file__).parent / "ollama_model_names.txt") as f:
+        return [line.strip() for line in f.readlines()]
 
 
 def create_model(
@@ -82,7 +62,7 @@ def create_model(
     # We use a `partial` here to ensure that we have the correct way of specifying
     # a model name between ChatOpenAI and ChatOllama.
     ModelClass = partial(ChatOpenAI, model_name=model_name)
-    if model_name.split(":")[0] in ollama_model_keywords:
+    if model_name.split(":")[0] in ollama_model_keywords():
         ModelClass = partial(ChatOllama, model=model_name)
 
     return ModelClass(

llamabot/bot/ollama_model_names.txt

@@ -0,0 +1,31 @@
+mistral
+llama2
+codellama
+vicuna
+orca-mini
+llama2-uncensored
+wizard-vicuna-uncensored
+nous-hermes
+phind-codellama
+mistral-openorca
+wizardcoder
+wizard-math
+llama2-chinese
+stable-beluga
+codeup
+everythinglm
+medllama2
+wizardlm-uncensored
+zephyr
+falcon
+wizard-vicuna
+open-orca-platypus2
+starcoder
+samantha-mistral
+openhermes2-mistral
+wizardlm
+sqlcoder
+dolphin2.1-mistral
+nexusraven
+dolphin2.2-mistral
+codebooga

mkdocs.yaml

@@ -11,6 +11,7 @@ theme:
   features:
     - instant
     - tabs
+    - content.code.copy
   language: en
 
 # We customize the navigation by hand to control the order

scratch_notebooks/scrape_ollama_models.ipynb

@@ -0,0 +1,78 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%load_ext autoreload\n",
+    "%autoreload 2\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from bs4 import BeautifulSoup\n",
+    "import requests\n",
+    "\n",
+    "# Your HTML snippet\n",
+    "response = requests.get(\"https://ollama.ai/library\")\n",
+    "if response.status_code == 200:\n",
+    "\n",
+    "    html_content = response.text\n",
+    "\n",
+    "    # Parse the HTML snippet with BeautifulSoup\n",
+    "    soup = BeautifulSoup(html_content, \"lxml\")\n",
+    "\n",
+    "    # Find all h2 tags that contain the model names\n",
+    "    model_names = [h2.text for h2 in soup.find_all(\"h2\")]\n",
+    "\n",
+    "    # Print out the model names\n",
+    "    for name in model_names:\n",
+    "        model_name = (name.strip('\\n').strip(' ').strip('\\n'))\n",
+    "        print(f'\"{model_name}\",')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "name\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "llamabot",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

scripts/autoupdate_ollama_models.py

@@ -0,0 +1,21 @@
+"""Automatically update the list of Ollama models in llamabot/bot/ollama_model_names.txt"""
+from bs4 import BeautifulSoup
+import requests
+
+# Your HTML snippet
+response = requests.get("https://ollama.ai/library")
+if response.status_code == 200:
+    html_content = response.text
+
+    # Parse the HTML snippet with BeautifulSoup
+    soup = BeautifulSoup(html_content, "lxml")
+
+    # Find all h2 tags that contain the model names
+    model_names = [
+        h2.text.strip("\n").strip(" ").strip("\n") for h2 in soup.find_all("h2")
+    ]
+
+# Write model names to llamabot/bot/ollama_model_names.txt
+with open("llamabot/bot/ollama_model_names.txt", "w") as f:
+    f.write("\n".join(model_names))
+    f.write("\n")