From 21187c27c90acbec7f912b8af4feaec154de960f Mon Sep 17 00:00:00 2001 From: drbh Date: Tue, 27 Aug 2024 13:31:08 -0400 Subject: [PATCH] fix: bump minijinja version and add test for llama 3.1 tools (#2463) * fix: support tojson and avoid message indexing issue in template * fix: prefer minijinja native methods and prefer workspace level dependency * fix: adjust comment typo --- Cargo.lock | 4 ++-- Cargo.toml | 2 ++ backends/v3/Cargo.toml | 4 ++-- router/Cargo.toml | 4 ++-- router/src/infer/chat_template.rs | 37 +++++++++++++++++++++++++++++-- 5 files changed, 43 insertions(+), 8 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index aa5cb642..02e91bc1 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -2169,9 +2169,9 @@ dependencies = [ [[package]] name = "minijinja" -version = "2.1.0" +version = "2.2.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "45f7e8e35b6c7b169bf40b0176d2c79291ab8ee53290b84e0668ab21d841aa9d" +checksum = "6d7d3e3a3eece1fa4618237ad41e1de855ced47eab705cec1c9a920e1d1c5aad" dependencies = [ "serde", "serde_json", diff --git a/Cargo.toml b/Cargo.toml index 8bf75b90..79fda15d 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -29,6 +29,8 @@ tokenizers = { version = "0.19.1", features = ["http"] } hf-hub = { version = "0.3.1", features = ["tokio"] } metrics = { version = "0.23.0" } metrics-exporter-prometheus = { version = "0.15.1", features = [] } +minijinja = { version = "2.2.0", features = ["json"] } +minijinja-contrib = { version = "2.0.2", features = ["pycompat"] } [profile.release] incremental = true diff --git a/backends/v3/Cargo.toml b/backends/v3/Cargo.toml index 06a44bec..69dad072 100644 --- a/backends/v3/Cargo.toml +++ b/backends/v3/Cargo.toml @@ -53,8 +53,8 @@ utoipa-swagger-ui = { version = "6.0.0", features = ["axum"] } init-tracing-opentelemetry = { version = "0.14.1", features = [ "opentelemetry-otlp", ] } -minijinja = { version = "2.0.2" } -minijinja-contrib = { version = "2.0.2", features = ["pycompat"] } +minijinja = { workspace = true } +minijinja-contrib = { workspace = true } futures-util = "0.3.30" regex = "1.10.3" once_cell = "1.19.0" diff --git a/router/Cargo.toml b/router/Cargo.toml index 45acab8e..5c328e8a 100644 --- a/router/Cargo.toml +++ b/router/Cargo.toml @@ -46,8 +46,8 @@ ngrok = { version = "0.13.1", features = ["axum"], optional = true } init-tracing-opentelemetry = { version = "0.14.1", features = [ "opentelemetry-otlp", ] } -minijinja = { version = "2.0.2", features = ["json"] } -minijinja-contrib = { version = "2.0.2", features = ["pycompat"] } +minijinja = { workspace = true } +minijinja-contrib = { workspace = true } futures-util = "0.3.30" regex = "1.10.3" once_cell = "1.19.0" diff --git a/router/src/infer/chat_template.rs b/router/src/infer/chat_template.rs index bfa9421c..a736fc12 100644 --- a/router/src/infer/chat_template.rs +++ b/router/src/infer/chat_template.rs @@ -1,9 +1,8 @@ -use std::collections::HashSet; - use crate::infer::InferError; use crate::{ChatTemplateInputs, Message, MessageChunk, TextMessage, TokenizerConfigToken, Tool}; use minijinja::{Environment, ErrorKind, Template}; use minijinja_contrib::pycompat; +use std::collections::HashSet; /// Raise a exception (custom function) used in the chat templates pub(crate) fn raise_exception(err_text: String) -> Result { @@ -872,4 +871,38 @@ mod tests { let expected = "[INST] I'd like to show off how chat templating works! [/INST]Great! How can I help you today? [INST] Just testing\n---\n[{\"type\":\"function\",\"function\":{\"description\":\"Get the current weather\",\"name\":\"get_current_weather\",\"arguments\":{\"properties\":{\"format\":{\"description\":\"The temperature unit to use. Infer this from the users location.\",\"enum\":[\"celsius\",\"fahrenheit\"],\"type\":\"string\"},\"location\":{\"description\":\"The city and state, e.g. San Francisco, CA\",\"type\":\"string\"}},\"required\":[\"location\",\"format\"],\"type\":\"object\"}}}]\nThis default prompt will be used [/INST]".to_string(); assert_eq!(result.unwrap(), expected); } + + #[test] + fn test_chat_template_with_custom_tool_template() { + // chat template from meta-llama/Meta-Llama-3.1-8B-Instruct + let ct = ChatTemplate::new( + "{{- bos_token }}\n{%- if not tools_in_user_message is defined %}\n {%- set tools_in_user_message = true %}\n{%- endif %}\n{%- if not date_string is defined %}\n {%- set date_string = \"26 Jul 2024\" %}\n{%- endif %}\n{%- if not tools is defined %}\n {%- set tools = none %}\n{%- endif %}\n\n{#- This block extracts the system message, so we can slot it into the right place. #}\n{%- if messages[0]['role'] == 'system' %}\n {%- set system_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n{%- else %}\n {%- set system_message = \"\" %}\n{%- endif %}\n\n{#- System message + builtin tools #}\n{{- \"<|start_header_id|>system<|end_header_id|>\\n\\n\" }}\n{%- if builtin_tools is defined or tools is not none %}\n {{- \"Environment: ipython\\n\" }}\n{%- endif %}\n{%- if builtin_tools is defined %}\n {{- \"Tools: \" + builtin_tools | reject('equalto', 'code_interpreter') | join(\", \") + \"\\n\\n\"}}\n{%- endif %}\n{{- \"Cutting Knowledge Date: December 2023\\n\" }}\n{{- \"Today Date: \" + date_string + \"\\n\\n\" }}\n{%- if tools is not none and not tools_in_user_message %}\n {{- \"You have access to the following functions. To call a function, please respond with JSON for a function call.\" }}\n {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n {{- \"Do not use variables.\\n\\n\" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- \"\\n\\n\" }}\n {%- endfor %}\n{%- endif %}\n{{- system_message }}\n{{- \"<|eot_id|>\" }}\n\n{#- Custom tools are passed in a user message with some extra guidance #}\n{%- if tools_in_user_message and not tools is none %}\n {#- Extract the first user message so we can plug it in here #}\n {%- if messages | length != 0 %}\n {%- set first_user_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n {%- else %}\n {{- raise_exception(\"Cannot put tools in the first user message when there's no first user message!\") }}\n{%- endif %}\n {{- '<|start_header_id|>user<|end_header_id|>\\n\\n' -}}\n {{- \"Given the following functions, please respond with a JSON for a function call \" }}\n {{- \"with its proper arguments that best answers the given prompt.\\n\\n\" }}\n {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n {{- \"Do not use variables.\\n\\n\" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- \"\\n\\n\" }}\n {%- endfor %}\n {{- first_user_message + \"<|eot_id|>\"}}\n{%- endif %}\n\n{%- for message in messages %}\n {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}\n {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\\n\\n'+ message['content'] | trim + '<|eot_id|>' }}\n {%- elif 'tool_calls' in message %}\n {%- if not message.tool_calls|length == 1 %}\n {{- raise_exception(\"This model only supports single tool-calls at once!\") }}\n {%- endif %}\n {%- set tool_call = message.tool_calls[0].function %}\n {%- if builtin_tools is defined and tool_call.name in builtin_tools %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n {{- \"<|python_tag|>\" + tool_call.name + \".call(\" }}\n {%- for arg_name, arg_val in tool_call.arguments | items %}\n {{- arg_name + '=\"' + arg_val + '\"' }}\n {%- if not loop.last %}\n {{- \", \" }}\n {%- endif %}\n {%- endfor %}\n {{- \")\" }}\n {%- else %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n {{- '{\"name\": \"' + tool_call.name + '\", ' }}\n {{- '\"parameters\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- \"}\" }}\n {%- endif %}\n {%- if builtin_tools is defined %}\n {#- This means we're in ipython mode #}\n {{- \"<|eom_id|>\" }}\n {%- else %}\n {{- \"<|eot_id|>\" }}\n {%- endif %}\n {%- elif message.role == \"tool\" or message.role == \"ipython\" %}\n {{- \"<|start_header_id|>ipython<|end_header_id|>\\n\\n\" }}\n {%- if message.content is mapping or message.content is iterable %}\n {{- message.content | tojson }}\n {%- else %}\n {{- message.content }}\n {%- endif %}\n {{- \"<|eot_id|>\" }}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' }}\n{%- endif %}\n".to_string(), + Some(TokenizerConfigToken::String("".to_string())), + Some(TokenizerConfigToken::String("".to_string())), + ); + let msgs: Vec = vec![ + Message { + name: None, + role: "system".to_string(), + content: MessageContent::SingleText( + "Youre a helpful assistant! Answer the users question best you can." + .to_string(), + ), + }, + Message { + name: None, + role: "user".to_string(), + content: MessageContent::SingleText( + "What is the weather like in Brooklyn, New York?".to_string(), + ), + }, + ]; + let tools_string = r#"[{"type": "function","function": {"name": "get_current_weather","description": "Get the current weather","parameters": {"type": "object","properties": {"location": {"type": "string","description": "The city and state, e.g. San Francisco, CA"},"format": {"type": "string","enum": ["celsius", "fahrenheit"],"description": "The temperature unit to use. Infer this from the users location."}},"required": ["location", "format"]}}}]"#.to_string(); + let tools: Vec = serde_json::from_str(&tools_string).unwrap(); + let tool_prompt = "This default prompt will be used".to_string(); + let tools_and_prompt = Some((tools, tool_prompt)); + let result = ct.apply(None, msgs, tools_and_prompt); + let expected = "<|start_header_id|>system<|end_header_id|>\n\nEnvironment: ipython\nCutting Knowledge Date: December 2023\nToday Date: 26 Jul 2024\n\nYoure a helpful assistant! Answer the users question best you can.<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nGiven the following functions, please respond with a JSON for a function call with its proper arguments that best answers the given prompt.\n\nRespond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.Do not use variables.\n\n{\n \"function\": {\n \"arguments\": {\n \"properties\": {\n \"format\": {\n \"description\": \"The temperature unit to use. Infer this from the users location.\",\n \"enum\": [\n \"celsius\",\n \"fahrenheit\"\n ],\n \"type\": \"string\"\n },\n \"location\": {\n \"description\": \"The city and state, e.g. San Francisco, CA\",\n \"type\": \"string\"\n }\n },\n \"required\": [\n \"location\",\n \"format\"\n ],\n \"type\": \"object\"\n },\n \"description\": \"Get the current weather\",\n \"name\": \"get_current_weather\"\n },\n \"type\": \"function\"\n}\n\nWhat is the weather like in Brooklyn, New York?\n---\nThis default prompt will be used<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n".to_string(); + assert_eq!(result.unwrap(), expected); + } }