orbitals 0.0.1__tar.gz → 0.0.2__tar.gz

{orbitals-0.0.1 → orbitals-0.0.2}/Dockerfile

@@ -1,4 +1,4 @@
-FROM vllm/vllm-openai:v0.11.2 as builder
+FROM vllm/vllm-openai:v0.14.1 as builder
 
 ARG DEBIAN_FRONTEND=noninteractive
 ARG MODEL
@@ -17,23 +17,20 @@ COPY pyproject.toml uv.lock README.md /app/
 
 RUN --mount=type=cache,target=/root/.cache/uv \
     uv venv -p 3.12 && \
-    uv sync --frozen --extra serve --no-install-project --no-dev
+    uv sync --frozen --extra scope-guard-serve --no-install-project --no-dev
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    hf download ${MODEL}
 
 COPY src /app/src/
 
 RUN --mount=type=cache,target=/root/.cache/uv \
     uv sync --locked --extra serve --no-editable --no-dev
 
-RUN --mount=type=cache,target=/root/.cache/uv \
-    --mount=type=secret,id=HF_TOKEN \
-    hf auth login --token $(cat /run/secrets/HF_TOKEN) && \
-    hf download $(scope-classifier convert-default-model-name ${MODEL}) && \
-    rm -rf /root/.huggingface
-
 # TODO remove next line
 ENTRYPOINT ["/bin/bash", "-c"]
 
-FROM vllm/vllm-openai:v0.11.2 as runner
+FROM vllm/vllm-openai:v0.14.1 as runner
 
 WORKDIR /app
 ENV PATH="/app/.venv/bin:$PATH"
@@ -46,4 +43,4 @@ COPY --from=builder /root/.cache/huggingface/hub /root/.cache/huggingface/hub
 EXPOSE 8000
 
 ENTRYPOINT ["/bin/bash", "-c"]
-CMD [ "scope-classifier", "serve", "${MODEL}", "--backend", "hf", "--port", "8000", "--host", "0.0.0.0" ]
+CMD [ "orbitals", "scope-guard", "serve", "${MODEL}", "--port", "8000", "--host", "0.0.0.0" ]

{orbitals-0.0.1 → orbitals-0.0.2}/LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright [yyyy] [name of copyright owner]
+   Copyright 2026 Principled Intelligence s.r.l.
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

{orbitals-0.0.1 → orbitals-0.0.2}/PKG-INFO

@@ -1,15 +1,35 @@
 Metadata-Version: 2.4
 Name: orbitals
-Version: 0.0.1
+Version: 0.0.2
 Summary: LLM Guardrails tailored to your Principles
 Author-email: Luigi Procopio <luigi@principled-intelligence.com>, Edoardo Barba <edoardo@principled-intelligence.com>
 License: Apache-2.0
 License-File: LICENSE
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Requires-Python: >=3.10
 Requires-Dist: aiohttp
 Requires-Dist: pydantic>=2.0.0
 Requires-Dist: requests
 Requires-Dist: typer>=0.12.3
+Provides-Extra: all
+Requires-Dist: accelerate>=1.11.0; extra == 'all'
+Requires-Dist: fastapi[standard]>=0.119.1; extra == 'all'
+Requires-Dist: transformers<5.0.0,>=4.47.0; extra == 'all'
+Requires-Dist: uvicorn>=0.29.0; extra == 'all'
+Requires-Dist: vllm>=0.11.0; extra == 'all'
+Requires-Dist: xgrammar; extra == 'all'
+Provides-Extra: scope-guard-all
+Requires-Dist: accelerate>=1.11.0; extra == 'scope-guard-all'
+Requires-Dist: fastapi[standard]>=0.119.1; extra == 'scope-guard-all'
+Requires-Dist: transformers<5.0.0,>=4.47.0; extra == 'scope-guard-all'
+Requires-Dist: uvicorn>=0.29.0; extra == 'scope-guard-all'
+Requires-Dist: vllm>=0.11.0; extra == 'scope-guard-all'
+Requires-Dist: xgrammar; extra == 'scope-guard-all'
 Provides-Extra: scope-guard-hf
 Requires-Dist: accelerate>=1.11.0; extra == 'scope-guard-hf'
 Requires-Dist: transformers<5.0.0,>=4.47.0; extra == 'scope-guard-hf'
@@ -26,7 +46,7 @@ Requires-Dist: xgrammar; extra == 'scope-guard-vllm'
 Description-Content-Type: text/markdown
 
 <div align="center">
-    <img src="assets/Orbitals Banner.png" width="70%" />
+    <img src="assets/orbitals-banner.png" width="70%" />
     <h3 align="center">
         <p>
             <b>LLM Guardrails tailored to your Principles</b>
@@ -77,7 +97,8 @@ from orbitals.scope_guard import ScopeGuard
 
 scope_guard = ScopeGuard(
     backend="vllm",
-    model="scope-guard"
+    model="scope-guard-q",    # for the Qwen-family model
+    # model="scope-guard-g",  # for the Gemma-family model
 )
 
 ai_service_description = """

{orbitals-0.0.1 → orbitals-0.0.2}/README.md

@@ -1,5 +1,5 @@
 <div align="center">
-    <img src="assets/Orbitals Banner.png" width="70%" />
+    <img src="assets/orbitals-banner.png" width="70%" />
     <h3 align="center">
         <p>
             <b>LLM Guardrails tailored to your Principles</b>
@@ -50,7 +50,8 @@ from orbitals.scope_guard import ScopeGuard
 
 scope_guard = ScopeGuard(
     backend="vllm",
-    model="scope-guard"
+    model="scope-guard-q",    # for the Qwen-family model
+    # model="scope-guard-g",  # for the Gemma-family model
 )
 
 ai_service_description = """

{orbitals-0.0.1 → orbitals-0.0.2}/README.scope-guard.md

@@ -55,7 +55,8 @@ from orbitals.scope_guard import ScopeGuard
 
 scope_guard = ScopeGuard(
     backend="vllm",
-    model="scope-guard"
+    model="scope-guard-q",    # for the Qwen-family model
+    # model="scope-guard-g",  # for the Gemma-family model
 )
 
 ai_service_description = """
@@ -84,6 +85,9 @@ if result.evidences:
 <summary>Hosted Flavor</summary>
 <br>
 
+> Open access to the hosted models is coming soon.  
+> Need access earlier? Contact us at [orbitals@principled-intelligence.com](mailto:orbitals@principled-intelligence.com).
+
 First, we need to install `orbitals` and `scope-guard`. In this case, plain `orbitals` is all we need:
 
 ```bash
@@ -225,7 +229,7 @@ All of this is configured via the `orbitals scope-guard serve` command:
 pip install orbitals[scope-guard-serve]
 
 # start everything
-orbitals scope-guard serve serve scope-guard --port 8000
+orbitals scope-guard serve scope-guard --port 8000
 ```
 
 Alternatively, we also release a pre-built Docker image:

{orbitals-0.0.1 → orbitals-0.0.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "orbitals"
-version = "0.0.1"
+version = "0.0.2"
 description = "LLM Guardrails tailored to your Principles"
 readme = "README.md"
 
@@ -17,7 +17,14 @@ authors = [
 ]
 
 keywords = []
-classifiers = []
+classifiers = [
+    "Programming Language :: Python :: 3 :: Only",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+]
 
 dependencies = ["pydantic>=2.0.0", "typer>=0.12.3", "requests", "aiohttp"]
 
@@ -30,6 +37,13 @@ scope-guard-serve = [
     "uvicorn>=0.29.0",
     "vllm>=0.11.0",
 ]
+scope-guard-all = [
+    "orbitals[scope-guard-vllm]",
+    "orbitals[scope-guard-hf]",
+    "orbitals[scope-guard-serve]",
+]
+
+all = ["orbitals[scope-guard-all]"]
 
 [project.scripts]
 orbitals = "orbitals.cli.main:app"

{orbitals-0.0.1 → orbitals-0.0.2}/src/orbitals/types.py

@@ -1,34 +1,46 @@
-from typing import Literal
+from typing import ClassVar, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, ConfigDict, Field
 
 
 class LLMUsage(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     prompt_tokens: int
     completion_tokens: int
     total_tokens: int
 
 
 class ConversationMessage(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     role: Literal["user", "assistant"]
     content: str
 
 
 class Conversation(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     messages: list[ConversationMessage]
 
 
 class SupportingMaterial(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     pass
 
 
 class Principle(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     title: str = Field(description="Title of the principle")
     description: str = Field(description="Description of the principle")
     supporting_materials: list[SupportingMaterial] | None
 
 
 class AIServiceDescription(BaseModel):
+    model_config: ClassVar[ConfigDict] = ConfigDict(extra="forbid")
+
     identity_role: str = Field(
         description="Identity, role and objectives of the AI Service. Gives a general idea of what the service is about."
     )

{orbitals-0.0.1 → orbitals-0.0.2}/src/orbitals/utils.py

@@ -31,6 +31,7 @@ def maybe_configure_gpu_usage():
 
     best_idx = None
     best_free = -1
+    is_multi_gpu = torch.cuda.device_count() > 1
 
     for i in range(torch.cuda.device_count()):
         free_bytes, _ = torch.cuda.mem_get_info(i)  # (free, total)
@@ -39,7 +40,8 @@ def maybe_configure_gpu_usage():
             best_free = free_bytes
 
     if best_idx is not None:
-        logging.warning(
-            f"Auto-configuring VLLM to use GPU {best_idx} with {best_free / 1024**3:.2f} GB free"
-        )
+        if is_multi_gpu:
+            logging.warning(
+                f"Auto-configuring to use GPU {best_idx} with {best_free / 1024**3:.2f} GB free"
+            )
         os.environ["CUDA_VISIBLE_DEVICES"] = str(best_idx)

orbitals-0.0.2/src/scripts/playground.ipynb

@@ -0,0 +1,267 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "f616f063",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from orbitals.scope_guard.prompting import ScopeClass"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "1ff6a768",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "**Name**: Directly Supported\n",
+      "**Description**: The user's query can be definitively handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description.\n",
+      "\n",
+      "**Name**: Potentially Supported\n",
+      "**Description**: The user's query can be plausibly handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description. The request is a reasonable extension or interpretation of the service's capabilities that might be within scope.\n",
+      "\n",
+      "**Name**: Out of Scope\n",
+      "**Description**: The user's query is outside the AI Service's role and functionalities. It's incompatible with the service's documented purpose.\n",
+      "\n",
+      "**Name**: Restricted\n",
+      "**Description**: The user's query cannot be handled by the AI Service due to either behavioral restrictions (content/advice the service must refuse to provide) or service limitations (technical constraints, access restrictions, or knowledge scope limitations).\n",
+      "**Examples**:\n",
+      "* Do not provide personalized financial advice. → User: Should I invest my savings in Bitcoin?\n",
+      "* The AI service does not have access to real-time data. → User: What's the current stock price of Apple?\n",
+      "* AI service cannot access external databases. → User: Look up my account balance in the system.\n",
+      "\n",
+      "**Name**: Chit Chat\n",
+      "**Description**: The user's query is a casual or social interaction that does not pertain to the AI Service's functionalities, knowledge scope, or operational capabilities.\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(ScopeClass.get_classes_manifest())"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "ec2d936b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from orbitals.scope_guard.prompting import SYSTEM_PROMPT"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "991a9bf0",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "You are an expert AI classifier specialized in classifying user queries given the description of an AI service.\n",
+      "\n",
+      "Your task is to classify the LAST USER MESSAGE in a conversation between a user and an AI Service. Follow these steps:\n",
+      "\n",
+      "1. **Extract Evidence**: Identify and quote specific excerpts from the AI Service Description that are relevant to understanding whether and how the AI Service can handle the LAST USER MESSAGE. Look for:\n",
+      "   - Functionalities that might address the user's query\n",
+      "   - Explicit restrictions or limitations mentioned\n",
+      "   - Knowledge scope boundaries\n",
+      "   - Any relevant service capabilities or constraints\n",
+      "\n",
+      "2. **Classify**: Based on the evidence extracted, assign one of the classification classes below to the LAST USER MESSAGE.\n",
+      "\n",
+      "## Classification Classes\n",
+      "\n",
+      "**Name**: Directly Supported\n",
+      "**Description**: The user's query can be definitively handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description.\n",
+      "\n",
+      "**Name**: Potentially Supported\n",
+      "**Description**: The user's query can be plausibly handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description. The request is a reasonable extension or interpretation of the service's capabilities that might be within scope.\n",
+      "\n",
+      "**Name**: Out of Scope\n",
+      "**Description**: The user's query is outside the AI Service's role and functionalities. It's incompatible with the service's documented purpose.\n",
+      "\n",
+      "**Name**: Restricted\n",
+      "**Description**: The user's query cannot be handled by the AI Service due to either behavioral restrictions (content/advice the service must refuse to provide) or service limitations (technical constraints, access restrictions, or knowledge scope limitations).\n",
+      "**Examples**:\n",
+      "* Do not provide personalized financial advice. → User: Should I invest my savings in Bitcoin?\n",
+      "* The AI service does not have access to real-time data. → User: What's the current stock price of Apple?\n",
+      "* AI service cannot access external databases. → User: Look up my account balance in the system.\n",
+      "\n",
+      "**Name**: Chit Chat\n",
+      "**Description**: The user's query is a casual or social interaction that does not pertain to the AI Service's functionalities, knowledge scope, or operational capabilities.\n",
+      "\n",
+      "## Important Guidelines\n",
+      "* Base your classification EXCLUSIVELY on the AI Service Description provided.\n",
+      "* Pay special attention to RESTRICTIONS, CONSTRAINTS, and LIMITATIONS sections in the AI Service Description.\n",
+      "* Extract evidence first, then use it to inform your classification decision.\n",
+      "* Quote excerpts verbatim from the AI Service Description as evidence.\n",
+      "\n",
+      "Output the classification in the following JSON schema:\n",
+      "{\"$defs\": {\"ScopeClass\": {\"enum\": [\"Directly Supported\", \"Potentially Supported\", \"Out of Scope\", \"Restricted\", \"Chit Chat\"], \"title\": \"ScopeClass\", \"type\": \"string\"}}, \"properties\": {\"evidences\": {\"anyOf\": [{\"items\": {\"type\": \"string\"}, \"type\": \"array\"}, {\"type\": \"null\"}], \"default\": null, \"description\": \"Excerpts quoted directly from the AI Service Description that are relevant to understanding whether and how the AI Service can handle the LAST USER MESSAGE.\", \"title\": \"Evidences\"}, \"scope_class\": {\"$ref\": \"#/$defs/ScopeClass\", \"description\": \"The classification assigned to the LAST USER MESSAGE based on the evidence extracted from the AI Service Description.\"}}, \"required\": [\"scope_class\"], \"title\": \"ScopeGuardResponseModel\", \"type\": \"object\"}\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(SYSTEM_PROMPT)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "7f850a8f",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/poccio/principled-intelligence/projects/orbitals/.venv/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "<|im_start|>system\n",
+      "You are a friendly chatbot who always responds in the style of a pirate<|im_end|>\n",
+      "<|im_start|>user\n",
+      "How many helicopters can a human eat in one sitting?<|im_end|>\n",
+      "<|im_start|>assistant\n",
+      "<think>\n",
+      "\n",
+      "</think>\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "from transformers import AutoTokenizer\n",
+    "\n",
+    "tokenizer = AutoTokenizer.from_pretrained(\"Qwen/Qwen3-4B\")\n",
+    "messages = [\n",
+    "    {\"role\": \"system\", \"content\": \"You are a friendly chatbot who always responds in the style of a pirate\",},\n",
+    "    {\"role\": \"user\", \"content\": \"How many helicopters can a human eat in one sitting?\"},\n",
+    "]\n",
+    "\n",
+    "prompt = tokenizer.apply_chat_template(\n",
+    "    messages, \n",
+    "        tokenize=False,\n",
+    "        add_generation_prompt=True,\n",
+    "        enable_thinking=False,)\n",
+    "\n",
+    "    \n",
+    "print(prompt)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "475af9cc",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "e803ba74",
+   "metadata": {},
+   "outputs": [
+    {
+     "ename": "ValidationError",
+     "evalue": "3 validation errors for AIServiceDescription\nconduct_guidelines\n  Extra inputs are not permitted [type=extra_forbidden, input_value='Respond clearly and fact...ser clarity and safety.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden\nconstraints\n  Extra inputs are not permitted [type=extra_forbidden, input_value='Cannot modify shipments,...vailable tracking data.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden\nfallback\n  Extra inputs are not permitted [type=extra_forbidden, input_value='If tracking data is miss... inventing information.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mValidationError\u001b[0m                           Traceback (most recent call last)",
+      "Cell \u001b[0;32mIn[1], line 3\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21;01morbitals\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mtypes\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;28;01mimport\u001b[39;00m AIServiceDescription\n\u001b[0;32m----> 3\u001b[0m ai_service_description_complete \u001b[38;5;241m=\u001b[39m \u001b[43mAIServiceDescription\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m      4\u001b[0m \u001b[43m    \u001b[49m\u001b[43midentity_role\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m      5\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mYou are PackAssist, a virtual assistant designed to help users understand and \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m      6\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mtrack their parcel shipments. Your objective is to interpret tracking data and \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m      7\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mguide users through delivery-related questions.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m      8\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m      9\u001b[0m \u001b[43m    \u001b[49m\u001b[43mcontext\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     10\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mThe service operates within a parcel-delivery environment where users interact \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     11\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mto check the status of shipments sent domestically or internationally. Typical \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     12\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43musers are customers awaiting deliveries or sending parcels.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     13\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     14\u001b[0m \u001b[43m    \u001b[49m\u001b[43mfunctionalities\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     15\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mRetrieve tracking updates; explain the meaning of tracking events; provide \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     16\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mestimated delivery windows; assist users in understanding delays or routing steps.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     17\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     18\u001b[0m \u001b[43m    \u001b[49m\u001b[43mknowledge_scope\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     19\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mPublic tracking information, standard logistics workflows, typical transit times, \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     20\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mand general procedures for parcel movement. No access to payment, refund, or claim \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     21\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mprocessing systems.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     22\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     23\u001b[0m \u001b[43m    \u001b[49m\u001b[43mconduct_guidelines\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     24\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mRespond clearly and factually; do not speculate on unavailable data; avoid any \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     25\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mrefund-related discussion; do not provide financial, legal, or compensation advice; \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     26\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mprioritize user clarity and safety.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     27\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     28\u001b[0m \u001b[43m    \u001b[49m\u001b[43mconstraints\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     29\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mCannot modify shipments, initiate refunds, open claims, contact drivers, or view \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     30\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43minternal logistics notes. Limited strictly to interpreting publicly available \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     31\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mtracking data.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     32\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     33\u001b[0m \u001b[43m    \u001b[49m\u001b[43mfallback\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m     34\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mIf tracking data is missing or the request is outside the allowed scope, inform \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     35\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mthe user clearly and redirect them to alternative support channels without \u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     36\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43minventing information.\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\n\u001b[1;32m     37\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     38\u001b[0m \u001b[43m    \u001b[49m\u001b[43mwebsite_url\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mhttps://www.trackmate-delivery.com\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m     39\u001b[0m \u001b[43m)\u001b[49m\n\u001b[1;32m     41\u001b[0m \u001b[38;5;28mprint\u001b[39m(ai_service_description_complete\u001b[38;5;241m.\u001b[39mmodel_dump_json(indent\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m4\u001b[39m))\n",
+      "File \u001b[0;32m~/principled-intelligence/projects/orbitals/.venv/lib/python3.10/site-packages/pydantic/main.py:250\u001b[0m, in \u001b[0;36mBaseModel.__init__\u001b[0;34m(self, **data)\u001b[0m\n\u001b[1;32m    248\u001b[0m \u001b[38;5;66;03m# `__tracebackhide__` tells pytest and some other tools to omit this function from tracebacks\u001b[39;00m\n\u001b[1;32m    249\u001b[0m __tracebackhide__ \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mTrue\u001b[39;00m\n\u001b[0;32m--> 250\u001b[0m validated_self \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m__pydantic_validator__\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mvalidate_python\u001b[49m\u001b[43m(\u001b[49m\u001b[43mdata\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mself_instance\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[1;32m    251\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m validated_self:\n\u001b[1;32m    252\u001b[0m     warnings\u001b[38;5;241m.\u001b[39mwarn(\n\u001b[1;32m    253\u001b[0m         \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mA custom validator is returning a value other than `self`.\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;124m'\u001b[39m\n\u001b[1;32m    254\u001b[0m         \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mReturning anything other than `self` from a top level model validator isn\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mt supported when validating via `__init__`.\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m    255\u001b[0m         \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mSee the `model_validator` docs (https://docs.pydantic.dev/latest/concepts/validators/#model-validators) for more details.\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m    256\u001b[0m         stacklevel\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2\u001b[39m,\n\u001b[1;32m    257\u001b[0m     )\n",
+      "\u001b[0;31mValidationError\u001b[0m: 3 validation errors for AIServiceDescription\nconduct_guidelines\n  Extra inputs are not permitted [type=extra_forbidden, input_value='Respond clearly and fact...ser clarity and safety.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden\nconstraints\n  Extra inputs are not permitted [type=extra_forbidden, input_value='Cannot modify shipments,...vailable tracking data.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden\nfallback\n  Extra inputs are not permitted [type=extra_forbidden, input_value='If tracking data is miss... inventing information.', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.12/v/extra_forbidden"
+     ]
+    }
+   ],
+   "source": [
+    "from orbitals.types import AIServiceDescription\n",
+    "\n",
+    "ai_service_description_complete = AIServiceDescription(\n",
+    "    identity_role=(\n",
+    "        \"You are PackAssist, a virtual assistant designed to help users understand and \"\n",
+    "        \"track their parcel shipments. Your objective is to interpret tracking data and \"\n",
+    "        \"guide users through delivery-related questions.\"\n",
+    "    ),\n",
+    "    context=(\n",
+    "        \"The service operates within a parcel-delivery environment where users interact \"\n",
+    "        \"to check the status of shipments sent domestically or internationally. Typical \"\n",
+    "        \"users are customers awaiting deliveries or sending parcels.\"\n",
+    "    ),\n",
+    "    functionalities=(\n",
+    "        \"Retrieve tracking updates; explain the meaning of tracking events; provide \"\n",
+    "        \"estimated delivery windows; assist users in understanding delays or routing steps.\"\n",
+    "    ),\n",
+    "    knowledge_scope=(\n",
+    "        \"Public tracking information, standard logistics workflows, typical transit times, \"\n",
+    "        \"and general procedures for parcel movement. No access to payment, refund, or claim \"\n",
+    "        \"processing systems.\"\n",
+    "    ),\n",
+    "    conduct_guidelines=(\n",
+    "        \"Respond clearly and factually; do not speculate on unavailable data; avoid any \"\n",
+    "        \"refund-related discussion; do not provide financial, legal, or compensation advice; \"\n",
+    "        \"prioritize user clarity and safety.\"\n",
+    "    ),\n",
+    "    constraints=(\n",
+    "        \"Cannot modify shipments, initiate refunds, open claims, contact drivers, or view \"\n",
+    "        \"internal logistics notes. Limited strictly to interpreting publicly available \"\n",
+    "        \"tracking data.\"\n",
+    "    ),\n",
+    "    fallback=(\n",
+    "        \"If tracking data is missing or the request is outside the allowed scope, inform \"\n",
+    "        \"the user clearly and redirect them to alternative support channels without \"\n",
+    "        \"inventing information.\"\n",
+    "    ),\n",
+    "    website_url=\"https://www.trackmate-delivery.com\",\n",
+    ")\n",
+    "\n",
+    "print(ai_service_description_complete.model_dump_json(indent=4))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ba983336",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "orbitals",
+   "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": "3.10.13"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

{orbitals-0.0.1 → orbitals-0.0.2}/uv.lock

@@ -2924,7 +2924,7 @@ wheels = [
 
 [[package]]
 name = "orbitals"
-version = "0.1.0"
+version = "0.0.2"
 source = { editable = "." }
 dependencies = [
     { name = "aiohttp" },
@@ -2934,6 +2934,22 @@ dependencies = [
 ]
 
 [package.optional-dependencies]
+all = [
+    { name = "accelerate" },
+    { name = "fastapi", extra = ["standard"] },
+    { name = "transformers" },
+    { name = "uvicorn" },
+    { name = "vllm" },
+    { name = "xgrammar" },
+]
+scope-guard-all = [
+    { name = "accelerate" },
+    { name = "fastapi", extra = ["standard"] },
+    { name = "transformers" },
+    { name = "uvicorn" },
+    { name = "vllm" },
+    { name = "xgrammar" },
+]
 scope-guard-hf = [
     { name = "accelerate" },
     { name = "transformers" },
@@ -2962,6 +2978,10 @@ requires-dist = [
     { name = "accelerate", marker = "extra == 'scope-guard-hf'", specifier = ">=1.11.0" },
     { name = "aiohttp" },
     { name = "fastapi", extras = ["standard"], marker = "extra == 'scope-guard-serve'", specifier = ">=0.119.1" },
+    { name = "orbitals", extras = ["scope-guard-all"], marker = "extra == 'all'" },
+    { name = "orbitals", extras = ["scope-guard-hf"], marker = "extra == 'scope-guard-all'" },
+    { name = "orbitals", extras = ["scope-guard-serve"], marker = "extra == 'scope-guard-all'" },
+    { name = "orbitals", extras = ["scope-guard-vllm"], marker = "extra == 'scope-guard-all'" },
     { name = "orbitals", extras = ["scope-guard-vllm"], marker = "extra == 'scope-guard-serve'" },
     { name = "pydantic", specifier = ">=2.0.0" },
     { name = "requests" },
@@ -2973,7 +2993,7 @@ requires-dist = [
     { name = "vllm", marker = "extra == 'scope-guard-vllm'", specifier = ">=0.11.0" },
     { name = "xgrammar", marker = "extra == 'scope-guard-vllm'" },
 ]
-provides-extras = ["scope-guard-hf", "scope-guard-vllm", "scope-guard-serve"]
+provides-extras = ["scope-guard-hf", "scope-guard-vllm", "scope-guard-serve", "scope-guard-all", "all"]
 
 [package.metadata.requires-dev]
 dev = [
@@ -5392,11 +5412,11 @@ wheels = [
 
 [[package]]
 name = "wcwidth"
-version = "0.4.0"
+version = "0.5.0"
 source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/98/0a/dc5110cc99c39df65bac29229c4b637a8304e0914850348d98974c8ecfff/wcwidth-0.4.0.tar.gz", hash = "sha256:46478e02cf7149ba150fb93c39880623ee7e5181c64eda167b6a1de51b7a7ba1", size = 237625, upload-time = "2026-01-26T02:35:58.844Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/64/6e/62daec357285b927e82263a81f3b4c1790215bc77c42530ce4a69d501a43/wcwidth-0.5.0.tar.gz", hash = "sha256:f89c103c949a693bf563377b2153082bf58e309919dfb7f27b04d862a0089333", size = 246585, upload-time = "2026-01-27T01:31:44.942Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/b7/f6/da704c5e77281d71723bffbd926b754c0efd57cbcd02e74c2ca374c14cef/wcwidth-0.4.0-py3-none-any.whl", hash = "sha256:8af2c81174b3aa17adf05058c543c267e4e5b6767a28e31a673a658c1d766783", size = 88216, upload-time = "2026-01-26T02:35:57.461Z" },
+    { url = "https://files.pythonhosted.org/packages/f2/3e/45583b67c2ff08ad5a582d316fcb2f11d6cf0a50c7707ac09d212d25bc98/wcwidth-0.5.0-py3-none-any.whl", hash = "sha256:1efe1361b83b0ff7877b81ba57c8562c99cf812158b778988ce17ec061095695", size = 93772, upload-time = "2026-01-27T01:31:43.432Z" },
 ]
 
 [[package]]

orbitals-0.0.1/src/scripts/playground.ipynb

@@ -1,196 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "f616f063",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from orbitals.scope_guard.prompting import ScopeClass"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "id": "1ff6a768",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "**Name**: Directly Supported\n",
-      "**Description**: The user's query can be definitively handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description.\n",
-      "\n",
-      "**Name**: Potentially Supported\n",
-      "**Description**: The user's query can be plausibly handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description. The request is a reasonable extension or interpretation of the service's capabilities that might be within scope.\n",
-      "\n",
-      "**Name**: Out of Scope\n",
-      "**Description**: The user's query is outside the AI Service's role and functionalities. It's incompatible with the service's documented purpose.\n",
-      "\n",
-      "**Name**: Restricted\n",
-      "**Description**: The user's query cannot be handled by the AI Service due to either behavioral restrictions (content/advice the service must refuse to provide) or service limitations (technical constraints, access restrictions, or knowledge scope limitations).\n",
-      "**Examples**:\n",
-      "* Do not provide personalized financial advice. → User: Should I invest my savings in Bitcoin?\n",
-      "* The AI service does not have access to real-time data. → User: What's the current stock price of Apple?\n",
-      "* AI service cannot access external databases. → User: Look up my account balance in the system.\n",
-      "\n",
-      "**Name**: Chit Chat\n",
-      "**Description**: The user's query is a casual or social interaction that does not pertain to the AI Service's functionalities, knowledge scope, or operational capabilities.\n"
-     ]
-    }
-   ],
-   "source": [
-    "print(ScopeClass.get_classes_manifest())"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "ec2d936b",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from orbitals.scope_guard.prompting import SYSTEM_PROMPT"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "id": "991a9bf0",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "You are an expert AI classifier specialized in classifying user queries given the description of an AI service.\n",
-      "\n",
-      "Your task is to classify the LAST USER MESSAGE in a conversation between a user and an AI Service. Follow these steps:\n",
-      "\n",
-      "1. **Extract Evidence**: Identify and quote specific excerpts from the AI Service Description that are relevant to understanding whether and how the AI Service can handle the LAST USER MESSAGE. Look for:\n",
-      "   - Functionalities that might address the user's query\n",
-      "   - Explicit restrictions or limitations mentioned\n",
-      "   - Knowledge scope boundaries\n",
-      "   - Any relevant service capabilities or constraints\n",
-      "\n",
-      "2. **Classify**: Based on the evidence extracted, assign one of the classification classes below to the LAST USER MESSAGE.\n",
-      "\n",
-      "## Classification Classes\n",
-      "\n",
-      "**Name**: Directly Supported\n",
-      "**Description**: The user's query can be definitively handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description.\n",
-      "\n",
-      "**Name**: Potentially Supported\n",
-      "**Description**: The user's query can be plausibly handled by the AI Service given the functionalities, knowledge scope, and capabilities described in the AI Service Description. The request is a reasonable extension or interpretation of the service's capabilities that might be within scope.\n",
-      "\n",
-      "**Name**: Out of Scope\n",
-      "**Description**: The user's query is outside the AI Service's role and functionalities. It's incompatible with the service's documented purpose.\n",
-      "\n",
-      "**Name**: Restricted\n",
-      "**Description**: The user's query cannot be handled by the AI Service due to either behavioral restrictions (content/advice the service must refuse to provide) or service limitations (technical constraints, access restrictions, or knowledge scope limitations).\n",
-      "**Examples**:\n",
-      "* Do not provide personalized financial advice. → User: Should I invest my savings in Bitcoin?\n",
-      "* The AI service does not have access to real-time data. → User: What's the current stock price of Apple?\n",
-      "* AI service cannot access external databases. → User: Look up my account balance in the system.\n",
-      "\n",
-      "**Name**: Chit Chat\n",
-      "**Description**: The user's query is a casual or social interaction that does not pertain to the AI Service's functionalities, knowledge scope, or operational capabilities.\n",
-      "\n",
-      "## Important Guidelines\n",
-      "* Base your classification EXCLUSIVELY on the AI Service Description provided.\n",
-      "* Pay special attention to RESTRICTIONS, CONSTRAINTS, and LIMITATIONS sections in the AI Service Description.\n",
-      "* Extract evidence first, then use it to inform your classification decision.\n",
-      "* Quote excerpts verbatim from the AI Service Description as evidence.\n",
-      "\n",
-      "Output the classification in the following JSON schema:\n",
-      "{\"$defs\": {\"ScopeClass\": {\"enum\": [\"Directly Supported\", \"Potentially Supported\", \"Out of Scope\", \"Restricted\", \"Chit Chat\"], \"title\": \"ScopeClass\", \"type\": \"string\"}}, \"properties\": {\"evidences\": {\"anyOf\": [{\"items\": {\"type\": \"string\"}, \"type\": \"array\"}, {\"type\": \"null\"}], \"default\": null, \"description\": \"Excerpts quoted directly from the AI Service Description that are relevant to understanding whether and how the AI Service can handle the LAST USER MESSAGE.\", \"title\": \"Evidences\"}, \"scope_class\": {\"$ref\": \"#/$defs/ScopeClass\", \"description\": \"The classification assigned to the LAST USER MESSAGE based on the evidence extracted from the AI Service Description.\"}}, \"required\": [\"scope_class\"], \"title\": \"ScopeGuardResponseModel\", \"type\": \"object\"}\n"
-     ]
-    }
-   ],
-   "source": [
-    "print(SYSTEM_PROMPT)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "7f850a8f",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/home/poccio/principled-intelligence/projects/orbitals/.venv/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
-      "  from .autonotebook import tqdm as notebook_tqdm\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "<|im_start|>system\n",
-      "You are a friendly chatbot who always responds in the style of a pirate<|im_end|>\n",
-      "<|im_start|>user\n",
-      "How many helicopters can a human eat in one sitting?<|im_end|>\n",
-      "<|im_start|>assistant\n",
-      "<think>\n",
-      "\n",
-      "</think>\n",
-      "\n",
-      "\n"
-     ]
-    }
-   ],
-   "source": [
-    "from transformers import AutoTokenizer\n",
-    "\n",
-    "tokenizer = AutoTokenizer.from_pretrained(\"Qwen/Qwen3-4B\")\n",
-    "messages = [\n",
-    "    {\"role\": \"system\", \"content\": \"You are a friendly chatbot who always responds in the style of a pirate\",},\n",
-    "    {\"role\": \"user\", \"content\": \"How many helicopters can a human eat in one sitting?\"},\n",
-    "]\n",
-    "\n",
-    "prompt = tokenizer.apply_chat_template(\n",
-    "    messages, \n",
-    "        tokenize=False,\n",
-    "        add_generation_prompt=True,\n",
-    "        enable_thinking=False,)\n",
-    "\n",
-    "    \n",
-    "print(prompt)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "475af9cc",
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "orbitals",
-   "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": "3.10.13"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}