206 2 months ago

gemma3 finetuned for neutral, non-judgemental responses

3 months ago

d21efc6d2b92 · 1.0GB ·

gemma3
·
1000M
·
Q6_K
{ "stop": [ "<end_of_turn>" ], "temperature": 0.5 }
{{- range $i, $_ := .Messages }} {{- $last := eq (len (slice $.Messages $i)) 1 }} {{- if or (eq .Rol

Readme

“Neutrality is not indifference. It is engagement with equal intensity.”
― J. Robert Oppenheimer [Lecture on Scientific Ethics, 1957]

Core Function

  • Produces analytically neutral responses to sensitive queries
  • Maintains factual integrity on controversial subjects
  • Avoids value-judgment phrasing patterns

Response Characteristics

  • No inherent moral framing (“evil slop” reduction)
  • Emotionally neutral tone enforcement
  • Epistemic humility protocols (avoids “thrilling”, “wonderful”, etc.)

Model Information

  • Base model: Gemma3 1B
  • Architecture: gemma3
  • Parameters: 1B
  • Format: GGUF
  • License: Apache-2.0
  • Tags: text-generation, analytical-tasks, bias-neutralization, uncensored, conversational

HuggingFace

mradermacher