decadence/ ornith-1.5:9b-base-q4_k_m

102 2 days ago

9B — GGUF — MLX — uncensored — sampling | ~5-11 GB

vision tools thinking
ollama run decadence/ornith-1.5:9b-base-q4_k_m

Details

2 days ago

af351d13ba0c · 6.7GB

qwen35
·
9.2B
·
Q4_K_M
clip
·
456M
·
BF16
{ "min_p": 0, "presence_penalty": 1.5, "repeat_penalty": 1, "temperature": 1, "t

Readme

Ornith 1.5 9B

model type quant unc size source
ornith-1.5:9b-base-abliterated-mlx-4bit MLX int4 ✓ 5.1 GB ressl
ornith-1.5:9b-base-abliterated-mlx-8bit MLX int8 ✓ 9.5 GB ressl
ornith-1.5:9b-base-mlx-4bit MLX int4 - 5.1 GB ornith-ai
ornith-1.5:9b-base-mlx-8bit MLX int8 - 9.5 GB ornith-ai
ornith-1.5:9b-base-q4_k_m GGUF Q4_K_M - 6.7 GB ornith-ai
ornith-1.5:9b-base-q8_0 GGUF Q8_0 - 10 GB ornith-ai
ornith-1.5:9b-code-abliterated-mlx-4bit MLX int4 ✓ 5.1 GB ressl
ornith-1.5:9b-code-abliterated-mlx-8bit MLX int8 ✓ 9.5 GB ressl
ornith-1.5:9b-code-mlx-4bit MLX int4 - 5.1 GB ornith-ai
ornith-1.5:9b-code-mlx-8bit MLX int8 - 9.5 GB ornith-ai
ornith-1.5:9b-code-q4_k_m GGUF Q4_K_M - 6.7 GB ornith-ai
ornith-1.5:9b-code-q8_0 GGUF Q8_0 - 10 GB ornith-ai

Sampling Parameters

Following Ornith-1.5-9B Recommended sampling parameters:

parameter base code
temperature 1.0 0.6
top_p 0.95 0.95
top_k 20 20
min_p 0.0 0.0
presence_penalty 1.5 0.0
repetition_penalty 1.0 1.0

Vision

MLX version of Ornith-1.5-9B is stripped of vision. Non-MLX version has vision capability.