Ternary-Bonsai-2-27B PTQ1_0 and own42 LoRA
This folder keeps two upstream files that were used together. groxaxo did not quantize or train either file. Both source model cards report Apache-2.0.
| File | Bytes | Source | Recorded SHA-256 |
|---|---|---|---|
Ternary-Bonsai-2-27B-PTQ1_0.gguf |
5,946,648,928 | prism-ml/Ternary-Bonsai-2-27B-gguf revision b072e1d3b35a0a630cece372c2127528e0994386 |
53107f530aa52eb00912263ab1ee29bd199261c87cd7b4ad4ca1318c1fe33ee3 |
bonsai2-27b-abliterate-own42-lora.gguf |
9,683,168 | AtomicChat/Ternary-Bonsai-2-27B-Abliterate-LoRA-GGUF revision b8486d211e4784b2b2643201f28cba122ec1d9c8 |
8dce2bebffa85792b57139df2d98cd19201a163efdf88f9b9b8353f2d0ced76f |
The base GGUF header records architecture qwen35, size label 27B, basename folded, and general.file_type 143. It has no license field. The adapter header records general.type adapter, general.architecture qwen35, and general.name own42p-all.
The local runtime manifest used the base file with this adapter at LoRA scale 1.5, context 131072, and q8_0 KV cache. That scale is the saved local default, not a new quality claim. The upload checks both recorded hashes before publishing.
llama-server -m Ternary-Bonsai-2-27B-PTQ1_0.gguf \
--lora-scaled bonsai2-27b-abliterate-own42-lora.gguf:1.5 --jinja
This is a different file from groxaxo/Ternary-Bonsai-2-27B-Abliterated-v2-PQ2_0-MTP-GGUF-3060-bench.