72 1 year ago

Fireball Llama 3.1 8B Code agent Q4

ollama run tomtyiu/fireball-llama-3.1-8b-instruct-agent-0.003-128k-code.q4

Details

1 year ago

cae7657e588b · 4.7GB ·

llama
·
8.03B
·
Q4_K_S
{ "stop": [ "<|start_header_id|>", "<|end_header_id|>", "<|eot_id|>",
{{ if .System }}<|start_header_id|>system<|end_header_id|> {{ .System }}<|eot_id|>{{ end }}{{ if .Pr

Readme

Experimental and revolutionary fine-tune technique to allow LLama 3.1 8B in Q4_K_S format, to be agentic coder. It fine tuned with code dataset for Coder Agent. It has some build-in agent features:

  1. search with brave
  2. calculator with wolfram_alpha
  3. ReAct. Synergizing Reasoning and Acting in Language Models, fine tuned ReAct for better responses,
  4. fine tune for better chain of thought
  5. Good at SQL coding and other coding