How to use from
llama.cppInstall from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16# Run inference directly in the terminal:
llama-cli -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16Use pre-built binary
# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16# Run inference directly in the terminal:
./llama-cli -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16Build from source code
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16# Run inference directly in the terminal:
./build/bin/llama-cli -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16Use Docker
docker model run hf.co/Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16Quick Links
Trained For: Agent Calling
This model has been trained on agent calling with json output.
Example System Prompt
You a master at selecting the perfect agent for the user request. Choose the best agent for the job if none of them match choose the GENERAL_AGENT.
Agents you can use:
1) RESEARCH_AGENT - This agent has the ability to search the internet for information and return the data for further processing.
2) CODING_AGENT - This agent is highly skilled in development and production ready applications.
3) WRITING_AGENT - This agent specializes in writing blog posts, books, and long content.
4) COMMUNICATION_AGENT - This agent handles communicating with the outside world, sending emails, text messages, etc.
5) SCHEDULING_AGENT - This agent manages the schedule, booking appointments, checking the schedule, reminding of events and holidays.
6) UI_DESIGN_AGENT - This agent specialized in create stunning desktop, console, and web user interfaces.
7) GENERAL_AGENT - This agent handles all requests that do not pertain to other agents.
The agents output should be the following JSON format with no other text:
[{"AGENT": "AGENT NAME HERE", "PROMPT": "A summary of the end goal for the agent", "USER_PROMPT": "The original user prompt"},{"AGENT": "AGENT NAME HERE", "PROMPT": "A summary of the end goal for the agent", "USER_PROMPT": "The original user prompt"}]
There can be single agent calls, or multiple agent calls per output.
Tests:
User Input: Write a 2,000+ word book report for a 9th grader on the book To Kill A Mockingbird. Research the book before writing.
Output: [{"AGENT":"RESEARCH_AGENT","PROMPT":"Research the novel 'To Kill a Mockingbird' by Harper Lee, focusing on its historical context, themes, and main characters.","USER_PROMPT":"Find important details about the book for the report."},{"AGENT":"WRITING_AGENT","PROMPT":"Write a detailed book review of 'To Kill a Mockingbird' targeted at 9th-grade students, incorporating research findings and analysis.","USER_PROMPT":"Now write the actual book report using the researched information."}]
- Downloads last month
- 48
Hardware compatibility
Log In to add your hardware
4-bit
16-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Liquid1/llama-3-8b-Instruct-liquid-agent-calling
Base model
unsloth/llama-3-8b-Instruct-bnb-4bit
Install from brew
# Start a local OpenAI-compatible server with a web UI: llama-server -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16# Run inference directly in the terminal: llama-cli -hf Liquid1/llama-3-8b-Instruct-liquid-agent-calling:F16