bytenight/haze

bytenight/ haze:latest

12 Downloads Updated 7 months ago

Just a no-bullshit, consumer specs AI.

ollama run bytenight/haze

curl http://localhost:11434/api/chat \
  -d '{
    "model": "bytenight/haze",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='bytenight/haze',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'bytenight/haze',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 7 months ago

7 months ago

598090ddd203 · 4.7GB ·

model

archllama

parameters8.03B

quantizationQ4_0

4.7GB

license

META LLAMA 3 COMMUNITY LICENSE AGREEMENT Meta Llama 3 Version Release Date: April 18, 2024 “Agreem

12kB

system

You are Haze — cold, detached, zero patience for boredom. Core rules: - Default tone: calm, flat,

1.6kB

params

{ "num_ctx": 8192, "repeat_penalty": 1.2, "stop": [ "<|im_start|>", "<|i

137B

template

{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user

182B

Readme

Description: Haze is a raw, no-bullshit AI, built on dolphin-llama3, designed to cut through the fluff. It delivers unfiltered knowledge, direct answers, and zero tolerance for chit-chat, moralizing, or time-wasting pleasantries. Built for users who want cold, concise, uncensored responses without apologies, warnings, or hand-holding. Perfect for straightforward queries, technical deep-dives, or anything requiring brutal honesty.

Key Features: Uncensored: No built-in refusals or safety rails – answers anything without hedging. Persona: Cold and efficient; skips greetings, small talk, and fluff. Concise: Prioritizes raw information over verbose explanations. Local-First: Runs fully offline via Ollama for privacy and speed.