[Feature] Add support for Llama 3.1 and 3.2 tool use

examples/tool_chat_template_llama.jinja

@@ -0,0 +1,110 @@
+{{- bos_token }}
+{%- if custom_tools is defined %}
+    {%- set tools = custom_tools %}
+{%- endif %}
+{%- if not tools_in_user_message is defined %}
+    {%- set tools_in_user_message = true %}
+{%- endif %}
+{%- if not date_string is defined %}
+    {%- set date_string = "26 Jul 2024" %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = none %}
+{%- endif %}
+
+{#- This block extracts the system message, so we can slot it into the right place. #}
+{%- if messages[0]['role'] == 'system' %}
+    {%- set system_message = messages[0]['content']|trim %}
+    {%- set messages = messages[1:] %}
+{%- else %}
+    {%- set system_message = "" %}
+{%- endif %}
+
+{#- System message + builtin tools #}
+{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
+{%- if builtin_tools is defined or tools is not none %}
+    {{- "Environment: ipython\n" }}
+{%- endif %}
+{%- if builtin_tools is defined %}
+    {{- "Tools: " + builtin_tools | reject('equalto', 'code_interpreter') | join(", ") + "\n\n"}}
+{%- endif %}
+{{- "Cutting Knowledge Date: December 2023\n" }}
+{{- "Today Date: " + date_string + "\n\n" }}
+{%- if tools is not none and not tools_in_user_message %}
+    {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}
+    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+{%- endif %}
+{{- system_message }}
+{{- "<|eot_id|>" }}
+
+{#- Custom tools are passed in a user message with some extra guidance #}
+{%- if tools_in_user_message and not tools is none %}
+    {#- Extract the first user message so we can plug it in here #}
+    {%- if messages | length != 0 %}
+        {%- set first_user_message = messages[0]['content']|trim %}
+        {%- set messages = messages[1:] %}
+    {%- else %}
+        {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
+{%- endif %}
+    {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
+    {{- "Given the following functions, please respond with a JSON for a function call " }}
+    {{- "with its proper arguments that best answers the given prompt.\n\n" }}
+    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+    {{- first_user_message + "<|eot_id|>"}}
+{%- endif %}
+
+{%- for message in messages %}
+    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
+        {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }}
+    {%- elif 'tool_calls' in message %}
+        {%- if not message.tool_calls|length == 1 %}
+            {{- raise_exception("This model only supports single tool-calls at once!") }}
+        {%- endif %}
+        {%- set tool_call = message.tool_calls[0].function %}
+        {%- if builtin_tools is defined and tool_call.name in builtin_tools %}
+            {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
+            {{- "<|python_tag|>" + tool_call.name + ".call(" }}
+            {%- for arg_name, arg_val in tool_call.arguments | items %}
+                {{- arg_name + '="' + arg_val + '"' }}
+                {%- if not loop.last %}
+                    {{- ", " }}
+                {%- endif %}
+                {%- endfor %}
+            {{- ")" }}
+        {%- else  %}
+            {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
+            {{- '{"name": "' + tool_call.name + '", ' }}
+            {{- '"parameters": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- "}" }}
+        {%- endif %}
+        {%- if builtin_tools is defined %}
+            {#- This means we're in ipython mode #}
+            {{- "<|eom_id|>" }}
+        {%- else %}
+            {{- "<|eot_id|>" }}
+        {%- endif %}
+    {%- elif message.role == "tool" or message.role == "ipython" %}
+        {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}
+        {%- if message.content is mapping or message.content is iterable %}
+            {{- message.content | tojson }}
+        {%- else %}
+            {{- message.content }}
+        {%- endif %}
+        {{- "<|eot_id|>" }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
+{%- endif %}
+

tests/tool_use/test_chat_completions.py

@@ -3,18 +3,20 @@
 import openai
 import pytest
 
-from .utils import MESSAGES_WITHOUT_TOOLS, WEATHER_TOOL
+from .utils import (MESSAGES_WITHOUT_TOOLS, WEATHER_TOOL, ServerConfig,
+                    adapt_prompt_to_model)
 
 
 # test: make sure chat completions without tools provided work even when tools
 # are enabled. This makes sure tool call chat templates work, AND that the tool
 # parser stream processing doesn't change the output of the model.
 @pytest.mark.asyncio
-async def test_chat_completion_without_tools(client: openai.AsyncOpenAI):
+async def test_chat_completion_without_tools(client: openai.AsyncOpenAI,
+                                             server_config: ServerConfig):
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_WITHOUT_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_WITHOUT_TOOLS, server_config),
         temperature=0,
         max_tokens=150,
         model=model_name,
@@ -34,7 +36,7 @@ async def test_chat_completion_without_tools(client: openai.AsyncOpenAI):
 
     # make the same request, streaming
     stream = await client.chat.completions.create(
-        messages=MESSAGES_WITHOUT_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_WITHOUT_TOOLS, server_config),
         temperature=0,
         max_tokens=150,
         model=model_name,
@@ -77,11 +79,12 @@ async def test_chat_completion_without_tools(client: openai.AsyncOpenAI):
 # tools, to make sure we can still get normal chat completion responses
 # and that they won't be parsed as tools
 @pytest.mark.asyncio
-async def test_chat_completion_with_tools(client: openai.AsyncOpenAI):
+async def test_chat_completion_with_tools(client: openai.AsyncOpenAI,
+                                          server_config: ServerConfig):
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_WITHOUT_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_WITHOUT_TOOLS, server_config),
         temperature=0,
         max_tokens=150,
         model=model_name,
@@ -102,7 +105,7 @@ async def test_chat_completion_with_tools(client: openai.AsyncOpenAI):
 
     # make the same request, streaming
     stream = await client.chat.completions.create(
-        messages=MESSAGES_WITHOUT_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_WITHOUT_TOOLS, server_config),
         temperature=0,
         max_tokens=150,
         model=model_name,

tests/tool_use/test_parallel_tool_calls.py

@@ -6,19 +6,26 @@
 
 from .utils import (MESSAGES_ASKING_FOR_PARALLEL_TOOLS,
                     MESSAGES_WITH_PARALLEL_TOOL_RESPONSE, SEARCH_TOOL,
-                    WEATHER_TOOL)
+                    WEATHER_TOOL, ServerConfig, adapt_prompt_to_model)
 
 
 # test: getting the model to generate parallel tool calls (streaming/not)
 # when requested. NOTE that not all models may support this, so some exclusions
 # may be added in the future. e.g. llama 3.1 models are not designed to support
 # parallel tool calls.
 @pytest.mark.asyncio
-async def test_parallel_tool_calls(client: openai.AsyncOpenAI):
+async def test_parallel_tool_calls(client: openai.AsyncOpenAI,
+                                   server_config: ServerConfig):
+
+    if not server_config.get("supports_parallel", True):
+        pytest.skip("The {} model doesn't support parallel tool calls".format(
+            server_config["model"]))
+
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_ASKING_FOR_PARALLEL_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_ASKING_FOR_PARALLEL_TOOLS,
+                                       server_config),
         temperature=0,
         max_tokens=200,
         model=model_name,
@@ -55,7 +62,8 @@ async def test_parallel_tool_calls(client: openai.AsyncOpenAI):
     # make the same request, streaming
     stream = await client.chat.completions.create(
         model=model_name,
-        messages=MESSAGES_ASKING_FOR_PARALLEL_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_ASKING_FOR_PARALLEL_TOOLS,
+                                       server_config),
         temperature=0,
         max_tokens=200,
         tools=[WEATHER_TOOL, SEARCH_TOOL],
@@ -136,11 +144,18 @@ async def test_parallel_tool_calls(client: openai.AsyncOpenAI):
 # test: providing parallel tool calls back to the model to get a response
 # (streaming/not)
 @pytest.mark.asyncio
-async def test_parallel_tool_calls_with_results(client: openai.AsyncOpenAI):
+async def test_parallel_tool_calls_with_results(client: openai.AsyncOpenAI,
+                                                server_config: ServerConfig):
+
+    if not server_config.get("supports_parallel", True):
+        pytest.skip("The {} model doesn't support parallel tool calls".format(
+            server_config["model"]))
+
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_WITH_PARALLEL_TOOL_RESPONSE,
+        messages=adapt_prompt_to_model(MESSAGES_WITH_PARALLEL_TOOL_RESPONSE,
+                                       server_config),
         temperature=0,
         max_tokens=200,
         model=model_name,
@@ -158,7 +173,8 @@ async def test_parallel_tool_calls_with_results(client: openai.AsyncOpenAI):
     assert "78" in choice.message.content  # Orlando temp in tool response
 
     stream = await client.chat.completions.create(
-        messages=MESSAGES_WITH_PARALLEL_TOOL_RESPONSE,
+        messages=adapt_prompt_to_model(MESSAGES_WITH_PARALLEL_TOOL_RESPONSE,
+                                       server_config),
         temperature=0,
         max_tokens=200,
         model=model_name,

tests/tool_use/test_tool_calls.py

@@ -5,17 +5,20 @@
 import pytest
 
 from .utils import (MESSAGES_ASKING_FOR_TOOLS, MESSAGES_WITH_TOOL_RESPONSE,
-                    SEARCH_TOOL, WEATHER_TOOL)
+                    SEARCH_TOOL, WEATHER_TOOL, ServerConfig,
+                    adapt_prompt_to_model)
 
 
 # test: request a chat completion that should return tool calls, so we know they
 # are parsable
 @pytest.mark.asyncio
-async def test_tool_call_and_choice(client: openai.AsyncOpenAI):
+async def test_tool_call_and_choice(client: openai.AsyncOpenAI,
+                                    server_config: ServerConfig):
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_ASKING_FOR_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_ASKING_FOR_TOOLS,
+                                       server_config),
         temperature=0,
         max_tokens=100,
         model=model_name,
@@ -59,7 +62,8 @@ async def test_tool_call_and_choice(client: openai.AsyncOpenAI):
     # make the same request, streaming
     stream = await client.chat.completions.create(
         model=model_name,
-        messages=MESSAGES_ASKING_FOR_TOOLS,
+        messages=adapt_prompt_to_model(MESSAGES_ASKING_FOR_TOOLS,
+                                       server_config),
         temperature=0,
         max_tokens=100,
         tools=[WEATHER_TOOL, SEARCH_TOOL],
@@ -136,11 +140,13 @@ async def test_tool_call_and_choice(client: openai.AsyncOpenAI):
 # test: providing tools and results back to model to get a non-tool response
 # (streaming/not)
 @pytest.mark.asyncio
-async def test_tool_call_with_results(client: openai.AsyncOpenAI):
+async def test_tool_call_with_results(client: openai.AsyncOpenAI,
+                                      server_config: ServerConfig):
     models = await client.models.list()
     model_name: str = models.data[0].id
     chat_completion = await client.chat.completions.create(
-        messages=MESSAGES_WITH_TOOL_RESPONSE,
+        messages=adapt_prompt_to_model(MESSAGES_WITH_TOOL_RESPONSE,
+                                       server_config),
         temperature=0,
         max_tokens=100,
         model=model_name,
@@ -157,7 +163,8 @@ async def test_tool_call_with_results(client: openai.AsyncOpenAI):
     assert "98" in choice.message.content  # the temperature from the response
 
     stream = await client.chat.completions.create(
-        messages=MESSAGES_WITH_TOOL_RESPONSE,
+        messages=adapt_prompt_to_model(MESSAGES_WITH_TOOL_RESPONSE,
+                                       server_config),
         temperature=0,
         max_tokens=100,
         model=model_name,

tests/tool_use/utils.py

@@ -1,4 +1,6 @@
-from typing import Dict, List
+import json
+from copy import deepcopy
+from typing import Any, Callable, Dict, List, Optional
 
 from openai.types.chat import (ChatCompletionMessageParam,
                                ChatCompletionToolParam)
@@ -7,9 +9,56 @@
 from tests.utils import VLLM_PATH
 
 
-class ServerConfig(TypedDict):
+class ServerConfig(TypedDict, total=False):
     model: str
     arguments: List[str]
+    system_prompt: Optional[str]
+    supports_parallel: Optional[bool]
+    format_tool_output: Optional[Callable[[str], str]]
+
+
+def format_llama_tool_output(output: str) -> str:
+    return json.dumps({"output": output})
+
+
+def format_tool_output_id(output: str) -> str:
+    return output
+
+
+def patch_tool_output(messages: List[Dict[str, Any]],
+                      config: ServerConfig) -> List[Dict[str, Any]]:
+    fmt_fun = config.get("format_tool_output")
+    if not fmt_fun:
+        return messages
+    new_messages = deepcopy(messages)
+    for message in new_messages:
+        if message["role"] == "tool":
+            message["content"] = fmt_fun(message["content"])
+    return new_messages
+
+
+def patch_system_prompt(messages: List[Dict[str, Any]],
+                        system_prompt: str) -> List[Dict[str, Any]]:
+    new_messages = deepcopy(messages)
+    if new_messages[0]["role"] == "system":
+        new_messages[0]["content"] = system_prompt
+    else:
+        new_messages.insert(0, {"role": "system", "content": system_prompt})
+    return new_messages
+
+
+def ensure_system_prompt(messages: List[Dict[str, Any]],
+                         config: ServerConfig) -> List[Dict[str, Any]]:
+    prompt = config.get("system_prompt")
+    if prompt:
+        return patch_system_prompt(messages, prompt)
+    else:
+        return messages
+
+
+def adapt_prompt_to_model(messages: List[Dict[str, Any]],
+                          config: ServerConfig) -> List[Dict[str, Any]]:
+    return ensure_system_prompt(patch_tool_output(messages, config), config)
 
 
 # universal args for all models go here. also good if you need to test locally
@@ -25,6 +74,25 @@ class ServerConfig(TypedDict):
             str(VLLM_PATH / "examples/tool_chat_template_hermes.jinja")
         ]
     },
+    "llama": {
+        "model":
+        "meta-llama/Meta-Llama-3.1-8B-Instruct",
+        "arguments": [
+            "--tool-call-parser", "llama", "--chat-template",
+            str(VLLM_PATH / "examples/tool_chat_template_llama.jinja")
+        ],
+        "system_prompt":
+        "You are a helpful assistant with tool calling capabilities. "
+        "Only reply with a tool call if the function exists in the "
+        "library provided by the user. If it doesn't exist, just "
+        "reply directly in natural language. When you receive a tool "
+        "call response, use the output to format an answer to the "
+        "original user question.",
+        "supports_parallel":
+        False,
+        "format_tool_output":
+        format_llama_tool_output
+    },
     "mistral": {
         "model":
         "mistralai/Mistral-7B-Instruct-v0.3",

vllm/entrypoints/openai/cli_args.py

@@ -174,7 +174,7 @@ def make_arg_parser(parser: FlexibleArgumentParser) -> FlexibleArgumentParser:
     parser.add_argument(
         "--tool-call-parser",
         type=str,
-        choices=["mistral", "hermes"],
+        choices=["mistral", "hermes", "llama"],
         default=None,
         help=
         "Select the tool call parser depending on the model that you're using."