ValiantLabs/Fireplace-34b

Fireplace-34b is a chat and function calling model built on the Yi-34b architecture.

• Built on Yi-34b architecture, using abacusai/Smaug-34B-v0.1 as the base model.
• Adds function calling capabilities to a high-performance chat-focused base model.
• For a smaller, code-focused Fireplace, try Fireplace-13b!

Version

This is Version 1.0 of Fireplace-34b.

The current version of Fireplace-34b uses abacusai/Smaug-34B-v0.1 trained on glaive-function-calling-v2 using Llama 2's [INST] chat format.

Unlike our smaller code-first Fireplace-13b model, Fireplace-34b is built for all-purpose chat performance, with function calling as a secondary feature. We'll look to continue this balance with future upgrades to Fireplace-34b.

Fireplace is part of our Build Tools campaign, to deliver helpful open source capabilities for users and creators.

We're excited to continue to bring Fireplace to larger model architectures, to maximize baseline model capability and function-calling performance!

Prompting Guide

Fireplace-34b specializes in function calling, but is well suited to general chat and offers generally flexible chat and chat-instruct capabilities.

The recommended format for Fireplace-34b is based on the Llama 2 chat format:

[INST] << SYS >> You are Fireplace, an expert code assistant. << /SYS >> Hi, I need to write a cover letter. Can you help me with that? [/INST]

For function calling in Fireplace-34b, the recommended format is to deliver the function(s) in a system message and then proceed with chat:

[INST] << SYS >> You are Fireplace, an expert code assistant with access to the following functions. Use them if required - { "name": "convert_currency", "description": "Convert currency from one type to another", "parameters": { "type": "object", "properties": { "amount": { "type": "number", "description": "The amount to be converted" }, "from_currency": { "type": "string", "description": "The currency to convert from" }, "to_currency": { "type": "string", "description": "The currency to convert to" } }, "required": [ "amount", "from_currency", "to_currency" ] } } { "name": "generate_random_password", "description": "Generate a random password", "parameters": { "type": "object", "properties": { "length": { "type": "integer", "description": "The length of the password" } }, "required": [ "length" ] } } << /SYS >> Hi, I need to convert 500 USD to Euros. Can you help me with that? [/INST]

Assistant will deliver function call responses between <functioncall> and <|endoftext|>:

<functioncall> {"name": "convert_currency", "arguments": '{"amount": 500, "from_currency": "USD", "to_currency": "EUR"}'} <|endoftext|>

(Please note that <|endoftext|> is not an EOS/EOT token, it is used to indicate the end of function call responses specifically.)

For recommended handling of function call responses, append a "FUNCTION RESPONSE: " to the existing chat history:

[INST] FUNCTION RESPONSE: {"converted_amount": 425.50, "from_currency": "USD", "to_currency": "EUR"} [/INST]

The model may be subject to errors and limitations, including those of the base model and dataset. We offer Fireplace-34b as open source for all to use. The user is responsible for all outputs.

Fireplace is created by Valiant Labs.

Try our flagship chat model, Shining Valiant!

Try Esper-70b, our DevOps-first code model!

Follow us on X for updates on our models!

We care about open source. For everyone to use.

We encourage others to finetune further from our models.