dlasher/ humanizer:IQ4_NL

22 1 month ago

pulled as BF16, requant to IQ4_NL (text_en_large) -- (English only focus)

tools thinking
ollama run dlasher/humanizer:IQ4_NL

Details

1 month ago

a022b8a15770 · 7.0GB

gemma4
·
11.9B
·
IQ4_NL
{ "min_p": 0.05, "num_batch": 2048, "num_ctx": 262144, "presence_penalty": 0, "r
{{- if .System }}<|im_start|>system {{ .System }}<|im_end|> {{- end }} {{- range .Messages }} {{- if

Readme

PERPLEXITY QUALITY SCORECARD

Base Model PPL : 5.4213 (16-bit reference)

Quantized Model PPL : 5.5490 (IQ4_NL target)

Perplexity Drift : +0.1277 (+2.36%)

Quality Assessment : [PASS] EXCELLENT (Near-Lossless / Gold Standard)

image.png

humanizer A 12B model that rewrites AI-written drafts (emails, essays, reports, forum posts; English and Chinese) so they read like a person wrote them. It is trained to keep every number, unit, date, name and quote and to add nothing, and it runs locally.

image.png

SOURCES:

** https://huggingface.co/jialinyyzz/humanizer

** https://huggingface.co/datasets/eaddario/imatrix-calibration

** https://github.com/sgaofen/humanizer-local-model/